Soil splash erosion simulation analysis system

Abstract

The invention relates to the technical field of soil splash erosion, in particular to a soil splash erosion simulation analysis system which comprises a rainfall simulator for simulating rainfall with different intensities and different pH values; the wind simulator adjusts wind speed and wind direction; raindrop data are measured by raindrop characteristic measuring equipment; the splash erosion simulation equipment simulates the process that raindrops drop on target soil to cause splash erosion and generate splash erosion substances; the data acquisition equipment acquires the corresponding initial pH value and initial conductivity of the target soil before rainfall; collecting the target pH value, the target conductivity and the target denudation depth of the target soil after rainfall, and the average weight diameter and the total target splattering amount of the target splattering substance corresponding to the splattering substance; the electronic equipment determines and establishes a splash erosion prediction model based on the raindrop pH value, the raindrop data, the wind speed, the wind direction and the data acquired by the data acquisition equipment, and the soil splash erosion simulation analysis system capable of quantitatively analyzing soil splash erosion in real time is developed.

Description

Technical Field

[0001] The present invention relates to the technical field of soil splash erosion, and in particular to a soil splash erosion simulation and analysis system. Background Art

[0002] Soil erosion, particularly splash erosion caused by raindrop impact, has widespread impacts on agricultural production and the ecological environment. As raindrops strike the soil surface, soil particles are not only compacted or separated but also transported through saltation or by surface runoff, resulting in nutrient loss and deterioration of soil structure. This process poses a potential threat to soil water retention, plant root growth, and farmland productivity.

[0003] Currently, research on soil splash erosion has largely focused on its mechanisms and influencing factors. However, significant gaps remain in practical applications, particularly in the dynamic monitoring and analysis of splash erosion under different soil types and climate conditions. Traditional soil erosion assessment methods often rely on laboratory testing or large-scale monitoring, lacking real-time, accurate soil splash erosion quantification and analysis tools. These tools are unable to accurately reflect the complexity and variability of raindrop impact in different environments.

[0004] Therefore, it is particularly necessary to develop a soil splash erosion simulation analysis system that can analyze soil splash erosion in real time and quantitatively. Summary of the Invention

[0005] In view of this, the present invention provides a soil splash erosion simulation and analysis system to develop a soil splash erosion simulation and analysis system that can perform real-time and quantitative analysis of soil splash erosion.

[0006] In a first aspect, the present invention provides a soil splash erosion simulation and analysis system, comprising: a rainfall simulator, a splash erosion simulation device, a raindrop characteristic measurement device, a data acquisition device, a wind simulator, and an electronic device, wherein the rainfall simulator is mounted above the raindrop characteristic measurement device, the raindrop characteristic measurement device is mounted above the splash erosion simulation device, and the data acquisition device is mounted on the splash erosion simulation device; the wind simulator is mounted on one side of the raindrop characteristic measurement device; the raindrop characteristic measurement device, the wind simulator, and the data acquisition device are all communicatively connected to the electronic device, wherein: the rainfall simulator is used to simulate rainfall of different intensities and different pH values; the wind simulator is used to adjust wind speed and direction under the control of the electronic device; the raindrop characteristic measurement device is used to measure raindrop data, wherein the raindrop data includes raindrop diameter, raindrop terminal velocity, raindrop kinetic energy, and raindrop movement posture; Splash erosion simulation equipment is used to simulate the process of raindrops landing on target soil, causing splash erosion and generating splash erosion materials; data acquisition equipment is used to collect the initial pH value and initial conductivity corresponding to the target soil before rainfall; and collect the corresponding target pH value, target conductivity, target erosion depth of the target soil after rainfall, as well as the target average weight diameter of the splash erosion materials and the target total splash erosion amount; electronic equipment is used to determine and establish a splash erosion prediction model between raindrop data and soil splash erosion data based on raindrop pH value, raindrop data, wind speed, wind direction, initial pH value, initial conductivity, target pH value, target conductivity, target erosion depth, target average weight diameter of the splash erosion materials and the target total splash erosion amount; and to predict the predicted erosion depth, predicted average weight diameter of the splash erosion materials and predicted total splash erosion amount corresponding to other soils to be tested based on the splash erosion prediction model.

[0007] In an optional embodiment, the rainfall simulator includes: a rain barrel, a regulating pump, a rain nozzle, a first liftable support frame and a motor; wherein the regulating pump is connected to the rain barrel and the rain nozzle, the motor is installed on the rain nozzle, the first liftable support frame is installed below the rain nozzle, and the regulating pump, the rain nozzle, the first liftable support frame and the motor are all communicatively connected to the electronic device, wherein: the rain barrel is used to hold rainwater with different pH values; the regulating pump is used to control the flow rate of water drawn to the rain nozzle through the rotation speed under the control of the electronic device to simulate rainfall of different intensities; the motor is used to drive the rain nozzle to rotate under the control of the electronic device to simulate the effect of natural rainfall; the rain nozzle is used to adjust different calibers under the control of the electronic device to adjust the diameter of raindrops, thereby simulating raindrops of different sizes; the first liftable support frame is used to adjust the height of the rain nozzle under the control of the electronic device, thereby controlling the height of rainfall.

[0008] In an optional embodiment, a wind simulator includes a rotatable base, a second liftable support frame, a mixed flow fan unit, and a honeycomb guide device installed at the air outlet of the mixed flow fan unit. The rotatable base, the second liftable support frame and the mixed flow fan unit are all communicatively connected to the electronic device, wherein: the rotatable base is used to rotate in a direction to adjust the wind direction under the control of the electronic device; the second liftable support frame is used to adjust the height under the control of the electronic device to match the height of the rainfall nozzle in the rainfall simulator; the mixed flow fan unit is used to adjust the speed frequency under the control of the electronic device to adjust the wind speed; the honeycomb guide device is used to effectively reduce the turbulence phenomenon generated when the fan discharges air.

[0009] In an optional embodiment, the splash simulation device includes a splash tray, a collection drawer, and a waterproof cover. The splash tray consists of a central source area and a collection area. The central source area is used to place a soil tray with drainage holes at the bottom. The soil tray is used to hold target soil. The collection area is composed of multiple concentric rings, which are separated by steel plates. The ring area of ​​each ring includes two symmetrical small holes for collecting splash materials, and the positions of the small holes in the ring areas of adjacent rings are staggered to maintain sealing and waterproofing between the ring areas. The collection drawer is installed under the splash tray and is used to place a splash material sample collection basin to improve the collection efficiency of the splash material. The waterproof cover is installed on the periphery of the splash tray to ensure that raindrops only act on the soil tray.

[0010] In an optional embodiment, the electronic device is used to calculate the conductivity difference based on the initial conductivity and the target conductivity; calculate the pH value difference based on the initial pH value and the target pH value; input the raindrop pH value, raindrop data, wind speed and wind direction into the first branch of the initial splash prediction network, perform feature extraction on the raindrop pH value and raindrop data to obtain external driving features; input the conductivity difference and the pH value difference into the second branch of the initial splash prediction network, perform feature extraction on the conductivity difference and the pH value difference to obtain soil response features; fuse the external driving features and the soil response features to obtain target features; pass the target features into the output layer, output the virtual erosion depth, the virtual average weight diameter of the splash material and the virtual total splash amount; based on the relationship between the virtual erosion depth, the virtual average weight diameter of the splash material and the virtual total splash amount and the target erosion depth, the target average weight diameter of the splash material and the target total splash amount, train the initial splash prediction network to obtain a splash prediction model.

[0011] In an optional embodiment, the electronic device is used to correct the raindrop diameter based on the wind speed to obtain a diameter correction coefficient; calculate the raindrop falling time based on the raindrop terminal velocity; correct the raindrop falling time based on the soil type corresponding to the target soil to obtain a time correction coefficient; correct the raindrop terminal velocity based on the wind speed to obtain a raindrop velocity correction coefficient; calculate the uniformity coefficient based on the wind speed and the raindrop posture; calculate the basic spatial range characteristics based on the diameter correction coefficient, the time correction coefficient and the raindrop velocity correction coefficient; calculate the impact range based on the basic spatial range characteristics and the uniformity coefficient. Characteristics; the kinetic energy density of raindrops is obtained by dividing the kinetic energy of raindrops by the diameter correction coefficient; the kinetic energy of raindrops is corrected based on wind direction, wind speed and raindrop movement posture to obtain the kinetic energy correction coefficient; the synergy factor is calculated according to the soil type corresponding to the target soil and the pH value of the raindrops; the directional strength coefficient is calculated based on the raindrop movement posture and the angle between wind direction and impact direction; the energy intensity is calculated based on the kinetic energy correction coefficient, the kinetic energy density of raindrops and the synergy factor; the impact intensity characteristics are calculated based on the energy intensity and the directional strength coefficient; the impact range characteristics and the impact intensity characteristics are integrated to generate external driving characteristics.

[0012] In an optional embodiment, the electronic device is used to correct the conductivity difference and the pH difference based on the relationship between the conductivity difference and the pH difference to obtain a corrected conductivity coefficient and a corrected pH coefficient; calculate the soil buffering index based on a first relationship between the corrected conductivity coefficient and the corrected pH coefficient; calculate the structural integrity characteristic based on a second relationship between the corrected conductivity coefficient and the corrected pH coefficient; and fuse the soil buffering index and the structural integrity characteristic to generate a soil response characteristic.

[0013] In an optional embodiment, the external driving characteristics include impact range characteristics and impact intensity characteristics; the soil response characteristics include soil buffering index and structural integrity characteristics; electronic equipment is used to assign initial weights to the external driving characteristics and soil response characteristics, respectively; a correlation coefficient matrix between the external driving characteristics and the soil response characteristics is constructed; based on the correlation coefficient matrix, target association characteristics whose correlation coefficients are greater than a preset correlation threshold are determined; cross-operation is performed on the target association characteristics to generate interactive characteristics; based on the interactive characteristics, the initial weights are corrected to obtain target weights corresponding to the external driving characteristics and the soil response characteristics, respectively; and based on the target weights, the external driving characteristics and the soil response characteristics are feature fused to generate target characteristics.

[0014] In an optional embodiment, an electronic device is used to calculate an erosion relative error and an erosion absolute error based on a virtual erosion depth and a target erosion depth; determine first weight information corresponding to the erosion relative error and the erosion absolute error respectively according to the soil type corresponding to the target soil; calculate the erosion error based on the first weight information, the erosion relative error and the erosion absolute error; calculate the square error of the average weight diameter of the spattering material based on the virtual average weight diameter of the spattering material and the target average weight diameter of the spattering material; calculate the proportion of first particles with a diameter greater than a preset diameter in the average weight diameter of the virtual spattering material and the proportion of second particles with a diameter greater than a preset diameter in the average weight diameter of the target spattering material; calculate the average weight diameter error of the spattering material based on the deviation between the proportion of the first particles and the proportion of the second particles and the square error of the average weight diameter of the spattering material; calculate the total spattering amount error based on the virtual total spattering amount and the target total spattering amount; generate a target loss function based on the erosion error, the square error of the average weight diameter of the spattering material and the total spattering amount error; and train the initial spattering prediction network based on the target loss function to obtain a spattering prediction model.

[0015] In an optional embodiment, the electronic device is also used to calculate the anti-erosion ability index based on the target erosion depth; calculate the aggregate anti-crushing ability index based on the average weight diameter of the target splashing material; calculate the overall anti-loss ability index based on the target total splashing amount; and obtain the target anti-erosion ability corresponding to the target soil based on the anti-erosion ability index, the aggregate anti-crushing ability index, and the overall anti-loss ability index.

[0016] The soil splash erosion simulation and analysis system provided in this application embodiment utilizes a structured layout of various devices to simultaneously collect parameters from the entire "rainfall-wind-soil response" chain. The rainfall simulator precisely controls rainfall intensity and pH. Combined with the wind simulator's wind speed and direction adjustment, it can replicate diverse climatic conditions, resolving the issue of traditional single-use rainfall simulations failing to capture wind influences. Raindrop characterization equipment captures key parameters such as raindrop diameter and kinetic energy in real time. The splash erosion simulation equipment, in conjunction with data acquisition equipment, records the complete evolution of the soil from its initial state to post-splash erosion, forming a closed-loop data chain from "raindrop parameters to soil response," providing a comprehensive basis for subsequent modeling. Electronic equipment then establishes a splash erosion prediction model based on the collected data, overcoming the limitations of traditional empirical assessments. The splash erosion prediction model's inputs include raindrop pH, raindrop data, environmental factors, and initial soil properties. Its output directly correlates to soil splash erosion results, enabling quantitative prediction of splash erosion risk in various scenarios. Compared to traditional laboratory testing, this splash erosion prediction model can rapidly infer splash erosion outcomes for unknown scenarios using historical data, reducing the cost of repeated experiments and providing a scientific prediction tool for agricultural soil and water conservation programs. If, under the same rainfall conditions, the average weight diameter of the target splash material decreases minimally and the total splash erosion is low, the soil is highly resistant to erosion. If weak initial pH buffering capacity leads to increased splash erosion, targeted soil improvement measures can be implemented. The results of erosion resistance assessments directly correlate with the priority of preventive measures. Soils with low erosion resistance can be prioritized for water and soil conservation projects, avoiding blind investments and improving the precision of soil protection.

[0017] Furthermore, soil erosion resistance is quantified separately across three key erosion stages: erosion, aggregate fragmentation, and loss, avoiding a one-sided assessment of soil erosion resistance based on a single indicator. Integrating these three indicators comprehensively covers the entire process of soil erosion resistance, from splashing and stripping to particle fragmentation and final loss, reflecting the overall level of soil erosion resistance. This can directly pinpoint weak links in soil erosion resistance (e.g., weak erosion resistance and easily broken aggregates), providing a clear basis for targeted improvements (e.g., increasing the application of organic fertilizer to enhance aggregate stability). BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 is a schematic structural diagram of a soil splash erosion simulation and analysis system according to an embodiment of the present invention; Figure 2 is a schematic structural diagram of a raindrop characteristic measuring device according to an embodiment of the present invention; Figure 3 1 is a flow chart of another soil splash erosion simulation and analysis system method according to an embodiment of the present invention; Figure 4 is a structural block diagram of a sputtering simulation device according to an embodiment of the present invention; Figure 5 1 is a flow chart of establishing a splash erosion prediction model between raindrop data and soil splash erosion data according to an embodiment of the present invention; Figure 6 1 is a schematic diagram of a process for obtaining a target corrosion resistance corresponding to a target soil according to an embodiment of the present invention; Among them, there are a rainfall simulator 1; a splash erosion simulation device 2; a raindrop characteristic measurement device 3; a data acquisition device 4; a wind simulator 5; an electronic device 6; a rain barrel 11; a regulating pump 12; a rainfall nozzle 13; a first liftable support frame 14; a motor 15; a rotatable base 51; a second liftable support frame 52; a mixed flow fan unit 53; a honeycomb guide device 54; a splash erosion plate 21; a collection drawer 22; a waterproof cover 23; a central source area 211; a collection area 212; a soil plate 213; and a steel plate 214. DETAILED DESCRIPTION

[0020] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0021] The present application embodiment provides a soil splash erosion simulation analysis system, such as Figure 1 As shown, the system includes: a rainfall simulator 1, a splash erosion simulation device 2, a raindrop characteristic measurement device 3, a data acquisition device 4, a wind simulator 5 and an electronic device 6, wherein the rainfall simulator 1 is installed above the raindrop characteristic measurement device 3, the raindrop characteristic measurement device 3 is installed above the splash erosion simulation device 2, and the wind simulator 5 is installed on one side of the raindrop characteristic measurement device; the raindrop characteristic measurement device 3, the wind simulator 5 and the data acquisition device 4 are all communicatively connected to the electronic device 6, wherein: Rainfall simulator 1, used to simulate rainfall of different intensities and pH values; A wind simulator 5 is used to adjust wind speed and direction under the control of an electronic device 6; Raindrop characteristic measuring device 3, used to measure raindrop data, including raindrop diameter, raindrop terminal velocity, raindrop kinetic energy and raindrop movement posture; Splash erosion simulation equipment 2, used to simulate the process of raindrops landing on the target soil causing splash erosion and generating splash erosion materials; Data acquisition equipment 4 is used to collect the initial pH value and initial conductivity of the target soil before rainfall; and to collect the target pH value, target conductivity, target erosion depth of the target soil after rainfall, and the target average weight diameter of the splash material and the target total splash amount corresponding to the splash material; Electronic device 6 is used to establish a splash erosion prediction model between raindrop data and soil splash erosion data based on raindrop pH value, raindrop data, wind speed, wind direction, initial pH value, initial conductivity, target pH value, target conductivity, target erosion depth, target average weight diameter of splashing material, and target total splashing amount; and to predict the predicted erosion depth, predicted average weight diameter of splashing material, and predicted total splashing amount corresponding to other soils to be tested based on the splash erosion prediction model.

[0022] Specifically, press Figure 1 The layout completes the coordinated deployment of equipment: Rainfall simulator 1 is installed directly above raindrop characterization device 3 (at a distance of 2-3 meters to ensure a complete raindrop path), which is then located directly above splash erosion simulator 2 (at a distance of 1.5 meters to accurately capture raindrop characteristics before they strike the soil). Wind simulator 5 is placed to the side of raindrop characterization device 3 (at a horizontal distance of 1.5 meters to simulate the effect of natural wind on raindrop trajectory). Electronic device 6 connects these devices via a communication interface to complete parameter calibration (such as the correspondence between the diameter of the rainfall nozzle 13 and the raindrop diameter, and the linear calibration between the fan speed and wind speed).

[0023] The target soil (5 cm thick, matching the specifications of the soil pan 213 of the splash erosion simulation device 2) was placed in the soil pan 213 of the splash erosion simulation device 2. After standing for 24 hours, the initial state parameters were measured by the data acquisition device 4: initial pH value (using a soil pH meter, measuring 3 times and taking the average value) and initial conductivity (using a portable conductivity meter, reflecting the basic state of soil salinity).

[0024] Electronic device 6 sends control instructions to rainfall simulator 1 and wind simulator 5. Rainfall simulator 1 adjusts the speed of pump 12 to set the rainfall intensity (e.g., 20 mm / h) and selects rainwater from rain barrel 11 with a corresponding pH value (e.g., pH 4.5 simulates acidic rain), simulating rainfall of varying intensities and pH values. Wind simulator 5 can adjust wind speed and direction. As raindrops fall from the rainfall simulator 1, the raindrop characteristic measurement device 3 synchronously captures data: raindrop diameter (measured by a laser particle size analyzer with an accuracy of 0.1 mm), raindrop terminal velocity (the trajectory is captured by a high-speed camera and the instantaneous velocity is calculated), raindrop kinetic energy (calculated based on mass and velocity: kinetic energy = 0.5 × raindrop mass × velocity²), and raindrop movement posture (tilt angle α, reflecting the impact of wind on the raindrop trajectory), and transmits the raindrop data in real time to the electronic device 6. Specifically, if Figure 2 As shown, the raindrop characteristic measuring device 3 is a remote sensing sensor comprising a transmitting end and a receiving end. At the transmitting end, a laser transmitter is used to generate a parallel laser beam; at the receiving end, the laser signal is received by a combination of a photodiode and a lens. By analyzing the attenuation and amplitude change of the laser beam when passing through precipitation particles, the parameters of raindrops produced by drippers of different diameters are measured, including the mean diameter of raindrops, the terminal velocity of raindrops, and the kinetic energy of raindrops. The raindrop characteristic measuring device records the number of raindrops passing through the measurement area per minute and measures the diameter and velocity of each raindrop. The raindrop cumulative volume percentage curve of the rainfall is used to obtain the corresponding raindrop diameter when the cumulative volume reaches 50%. This diameter is called the mean diameter of raindrops and is represented by D50. Assuming that the raindrop is a sphere, its volume is then calculated.

[0025] The kinetic energy of raindrops KE (Jm-2h-1) is calculated as follows: (1) Where m i represents the mass of raindrop i in kg; v i Indicates the speed of raindrop i in ms -2 .

[0026] The calculation formula for rainfall erosivity is expressed by the product of rainfall kinetic energy and rainfall intensity. This product is considered to be a key indicator for measuring rainfall erosivity.

[0027] (2) Where R represents rainfall erosivity, KE is rainfall kinetic energy, and I 30 Indicates the maximum rainfall intensity within 30 minutes.

[0028] Under the influence of wind, raindrops will be affected by the wind during the falling process, causing them to fall at an angle. This increases the kinetic energy of the raindrops, so its kinetic energy needs to be corrected accordingly according to the tilt angle of the rainfall.

[0029] (3) Among them, KE fis the corrected kinetic energy of rainfall under the influence of wind; α represents the angle between the direction of raindrop motion and the direction perpendicular to gravity. The following is the calculation process of angle α.

[0030] According to the force of uniform motion of raindrops during the falling process, it can be known that the gravity of raindrops is equal to the air resistance, and the formula is: (4) After transformation, the formula can be rewritten as: (5) After the above transformation, we can get the vertical velocity V of the raindrops: s .

[0031] (6) The force on a raindrop when it moves at a uniform speed in the horizontal direction can be expressed by Newton's drag formula as follows: (7) The horizontal velocity V of the raindrops can be obtained from the above formula w : (8) According to the calculation method of trigonometric function, the angle α between the direction of raindrop movement and the direction perpendicular to gravity is the vertical velocity V of the raindrop. s The horizontal velocity V of the raindrop w After substituting the ratio into equations (7) and (8), the calculation results are as follows: (9) In the above formula: m is the mass of a single raindrop, g is the acceleration of gravity, F a is the air resistance encountered by raindrops in the process of falling due to gravity. r is the radius of the raindrop, ρ w is the water density, C is the air drag coefficient, is the air density, S r is the cross-sectional area of ​​the raindrop, and F is the horizontal force on the raindrop when it falls.

[0032] After raindrops and wind force act synergistically on the target soil, the data acquisition device 4 starts multi-dimensional recording: immediately after rainfall, the target pH value and target conductivity are measured (compared with the initial value to reflect the physical and chemical changes of the soil).

[0033] Data collection equipment 4 may include a soil pH meter, a soil conductivity meter, a laser rangefinder, and a concentric ring collector. The soil pH meter can measure the initial pH value of the target soil before rainfall and the target pH value of the target soil after rainfall. The soil conductivity meter can measure the initial conductivity of the target soil before rainfall and the target conductivity of the target soil after rainfall.

[0034] A laser rangefinder measures the initial height from the target soil before rainfall. Then, after rainfall, the target height from the target soil is measured. The target erosion depth is calculated by subtracting the initial height from the target height. Simultaneously, data acquisition equipment 4 uses concentric ring collectors to weigh the total mass of the eroded material in each ring area to obtain the target total erosion volume. Furthermore, the average weight diameter of the eroded material is determined by sieving. All data are linked to timestamps, forming a complete data chain from "initial state - simulation parameters - erosion results."

[0035] The electronic device 6 data is based on the raindrop pH value, raindrop data, wind speed, wind direction, initial pH value, initial conductivity, target pH value, target conductivity, target erosion depth, target average weight diameter of splashing material and target total splashing amount, to establish a splashing prediction model between the raindrop data and the soil splashing data; based on the splashing prediction model, the predicted erosion depth, predicted average weight diameter of splashing material and predicted total splashing amount corresponding to other soils to be tested are predicted.

[0036] The soil splash erosion simulation and analysis system provided in this embodiment utilizes a structured layout of various devices (a rainfall simulator 1, a raindrop characteristic measurement device 3, and a splash erosion simulation device 2, linked vertically, with a wind simulator 5 providing lateral coordination) to achieve simultaneous collection of parameters across the entire "rainfall-wind-soil response" chain. The rainfall simulator 1 precisely controls rainfall intensity and pH. Combined with the wind simulator 5's wind speed and direction adjustment, it can replicate diverse climate conditions (e.g., acidic rainfall with crosswinds, neutral rainfall with no wind), resolving the issue of traditional single-use rainfall simulations failing to capture wind influences. The raindrop characteristic measurement device 3 captures key parameters such as raindrop diameter and kinetic energy in real time. The splash erosion simulation device 2, in conjunction with the data acquisition device 4, records the complete soil changes from the initial state (initial pH and electrical conductivity) to the post-splash erosion state (target erosion depth, average weight diameter of the splashed material, etc.), forming a closed-loop data chain linking "raindrop parameters-soil response," providing a comprehensive basis for subsequent modeling. The electronic device 6 establishes a splash erosion prediction model based on the collected data, overcoming the limitations of traditional empirical evaluation. The splash erosion prediction model's inputs include raindrop pH, raindrop data, environmental factors (wind speed, wind direction), and initial soil properties. Its output directly correlates to soil splash erosion results (total splash erosion volume, erosion depth, etc.), enabling quantitative prediction of splash erosion risks in different scenarios (e.g., the splash erosion volume of sandy soils from a combination of large-diameter raindrops and high wind speeds). Compared to traditional laboratory testing, this splash erosion prediction model can rapidly infer splash erosion results for unknown scenarios using historical data, reducing the cost of repeated experiments and providing a scientific predictive tool for agricultural soil and water conservation programs (e.g., selecting arable land cover methods). If, under the same rainfall conditions, the average weight diameter of the target splash material decreases slightly and the total splash erosion volume is low, it indicates strong erosion resistance (e.g., clay soils are generally better than sandy soils). If weak initial pH buffering capacity (large variations in target pH) leads to increased splash erosion, targeted soil improvement measures (e.g., applying organic fertilizer to enhance aggregate stability) can be implemented. The results of erosion resistance assessment are directly related to the priority of prevention and control measures. For soils with weak erosion resistance, priority can be given to implementing water and soil conservation projects (such as terrace construction) to avoid blind investment and improve the accuracy of soil protection.

[0037] In an optional embodiment of the present application, as Figure 3 As shown, the rainfall simulator 1 includes: a rain barrel 11, a regulating pump 12, a rain nozzle 13, a first liftable support frame 14 and a motor 15; wherein the regulating pump 12 is connected to the rain barrel 11 and the rain nozzle 13, the motor 15 is installed above the rain nozzle 13, and the first liftable support frame 14 is installed below the rain nozzle 13. The regulating pump 12, the rain nozzle 13, the first liftable support frame 14 and the motor 15 are all connected to the electronic device 6 for communication, wherein: A rainwater barrel 11 is used to hold rainwater of different pH values; The regulating pump 12 is used to control the water flow rate pumped to the rainfall nozzle 13 by the rotation speed under the control of the electronic device 6 to simulate rainfall of different intensities; The motor 15 is used to drive the rainfall nozzle 13 to rotate under the control of the electronic device 6 to simulate the effect of natural rainfall; The rainfall nozzle 13 is used to adjust different calibers under the control of the electronic device 6 to adjust the diameter of the raindrops, thereby simulating raindrops of different sizes; The first liftable support frame 14 is used to adjust the height of the rainfall nozzle 13 under the control of the electronic device 6, thereby controlling the height of the rainfall.

[0038] Specifically, rainwater barrels 11 are pre-filled with rainwater of varying pH values ​​(e.g., acidic rainwater with a pH of 5.0, neutral rainwater with a pH of 7.0) and labeled with pH values. The rainwater barrels 11 are connected to the inlet of a regulating pump 12 via a pipe, ensuring that water can be pumped into the rainfall sprinklers 13.

[0039] In the initial state, the first elevating support frame 14 adjusts the rainfall nozzle 13 to a reference height (e.g., 2.0 m, serving as a starting point for subsequent height adjustments); the rainfall nozzle 13 is reset to a default caliber (e.g., 1.5 mm, corresponding to medium-sized raindrops); the motor 15 stops rotating, and the rainfall nozzle 13 is in a stationary state; Then, the electronic device 6 controls the pipeline valve (implicit in the connection logic between the rainwater barrel 11 and the regulating pump 12) to specify the extraction of rainwater with a target pH value (such as selecting a rainwater barrel 11 with pH=5.0) to ensure that the water flow entering the sprinkler meets the pH requirements.

[0040] Electronic device 6 sends a speed command to regulating pump 12 based on the rainfall intensity-speed correspondence (e.g., a rainfall intensity of 20 mm / h corresponds to 3,000 rpm for regulating pump 12). Regulating pump 12 controls the amount of water pumped per unit time by varying its speed: a high speed (e.g., 4,000 rpm) results in a high flow rate, simulating heavy rainfall (e.g., 30 mm / h); a low speed (e.g., 1,000 rpm) results in a low flow rate, simulating light rainfall (e.g., 10 mm / h).

[0041] Electronic device 6 sends an aperture adjustment command to rainfall nozzle 13, which changes the water outlet diameter via the nozzle's internal valve or aperture switching mechanism: a large aperture (e.g., 2.5mm) produces large raindrops (2-3mm in diameter); a small aperture (e.g., 1.0mm) produces small raindrops (0.5-1mm in diameter). The correspondence between aperture and raindrop diameter requires pre-calibration (e.g., experimentally, "a 2.0mm aperture corresponds to a raindrop diameter of 2.2mm ± 0.2mm").

[0042] Electronic device 6 controls the extension and retraction of first elevating support frame 14 based on the relationship between raindrop falling height and terminal velocity (the higher the height, the greater the terminal velocity and the greater the impact kinetic energy). To simulate high-altitude rainfall (high kinetic energy), the support frame is raised to 2.5 meters; to simulate near-ground rainfall (low kinetic energy), the support frame is lowered to 1.5 meters. The height adjustment accuracy is typically ±0.1 meters, ensuring that the kinetic energy of the raindrop is controllable.

[0043] To simulate the randomness of raindrop distribution in natural rainfall, the electronic device 6 sends a rotation instruction to the motor 15: the motor 15 drives the rainfall nozzle 13 to rotate at a set speed (such as 30r / min) at a constant speed, so that the rainwater sprayed from the nozzle evenly covers the splash simulation device 2 below (avoiding local rainfall concentration caused by spraying water in a fixed direction); if "local heavy rainfall" needs to be simulated, the motor 15 can be instructed to rotate intermittently (such as rotating for 10 seconds and stopping for 5 seconds).

[0044] Next, the electronic device 6 receives feedback on the operating status of each component (such as the actual speed of the regulating pump 12 and the real-time height of the nozzle). If a deviation is found (such as a speed fluctuation of ±5%), a correction instruction is immediately sent: if the rainfall intensity is low (the actual flow rate is less than the set value), increase the speed of the regulating pump 12 by 5%; if the height of the nozzle is offset due to vibration, instruct the first liftable support frame 14 to fine-tune to the set height.

[0045] After the experiment ends, electronic device 6 sends a stop command: the speed of regulating pump 12 gradually decreases to 0 (to avoid sudden pump stop and water pressure fluctuations in the pipeline), and water supply to the sprinkler stops. Motor 15 stops rotating, and rainfall sprinkler 13 returns to its initial position. The diameter of rainfall sprinkler 13 is reset to the default value (1.5mm). The first elevating support frame 14 lowers the sprinkler to the reference height (2.0m). The electronic device 6 records the complete parameters of this rainfall (pH value, intensity, raindrop diameter, etc.) for subsequent correlation analysis with the splash erosion data.

[0046] In the soil splash erosion simulation and analysis system provided by the present embodiment, the rainwater barrel 11 can directly hold rainwater of varying pH values, providing a basis for simulating various types of rainfall, including acidic, neutral, and alkaline. The regulating pump 12 controls the water flow rate by rotating its speed, accurately simulating rainfall of varying intensities. The regulation process is controlled by the electronic device 6, ensuring the stability and repeatability of rainfall intensity parameters. The rain nozzle 13 can be adjusted to different diameters. Combined with the control of the electronic device 6, the raindrop diameter can be precisely adjusted to meet the needs of simulating raindrops of varying sizes. These designs enable the simulator to comprehensively cover the core parameters of rainfall, providing accurate "rainfall input" for subsequent soil splash erosion simulation. The motor 15 drives the rain nozzle 13 to rotate, ensuring a more uniform distribution of falling raindrops, more similar to the random distribution of raindrops in natural rainfall. The first elevating support frame 14 adjusts the height of the rain nozzle 13, thereby controlling the rainfall height and simulating the impact of raindrops falling on the soil at different heights. Through these designs, the simulator can reduce the deviation between artificially simulated rainfall and natural rainfall, making the simulated scene more realistic and natural. Each component is connected to the electronic device 6 for communication, which enables the automation and intelligence of parameter adjustment. Researchers can use the electronic device 6 to easily adjust parameters such as water flow, nozzle speed, nozzle diameter, and support frame height, and quickly switch between different rainfall scenes without frequent manual operations, which greatly improves the efficiency of the experiment and can better meet the needs of splash erosion simulation for different soil types and different climatic conditions. Finally, the rainfall simulator 1 can stably output controllable rainfall parameters. Combined with other equipment in the system, it can form a complete data chain of "rainfall parameter input-soil splash erosion response". The simulated rainfall scene can accurately reflect the effects of different rainfall conditions on the soil, and provide reliable experimental data support for studying the correlation mechanism between raindrop splashing and soil splash erosion, and establishing a splash erosion prediction model.

[0047] In an optional embodiment of the present application, as Figure 3 As shown, the wind simulator 5 includes a rotatable base 51, a second liftable support frame 52, a mixed flow fan unit 53, and a honeycomb guide device 54 installed at the air outlet of the mixed flow fan unit 53. The rotatable base 51, the second liftable support frame 52 and the mixed flow fan unit 53 are all communicatively connected to the electronic device 6, wherein: A rotatable base 51 is used to rotate under the control of the electronic device 6 to adjust the wind direction; The second liftable support frame 52 is used to adjust the height under the control of the electronic device 6 to match the height of the rainfall nozzle 13 in the rainfall simulator 1; The mixed flow fan unit 53 is used to adjust the rotation speed and frequency under the control of the electronic equipment 6, thereby adjusting the wind speed; The honeycomb-shaped flow guide device 54 is used to effectively reduce the turbulence generated when the fan discharges air.

[0048] Specifically, the rotatable base 51 enables precise control of wind direction. It houses a built-in stepper motor 15 (rotation control) and an angle encoder (accuracy ±0.5°). The base supports the entire wind simulator 5 and can rotate 360° without any blind spots. Under the control of the electronic device 6, the orientation of the fan outlet is adjusted to simulate different wind directions (such as tailwind, headwind, and 45° crosswind). For example, to simulate raindrops impacting the soil at an angle, the electronic device 6 sends a "30°" command, causing the base to rotate to 30°, causing the wind to act on the raindrops along a 30° angle.

[0049] The second, elevating support frame 52 enables height adaptation to wind speeds. It includes an electrically operated telescopic rod (travel range 0.5-3m) and a displacement sensor (accuracy ±0.01m). It supports and adjusts the height of the mixed flow fan assembly 53. The support frame extends and retracts according to the instructions of the electronic device 6, ensuring that the fan outlet height is aligned with the nozzle height of the rainfall simulator 1 (for example, if the rainfall nozzle 13 is 2.2m high, the support frame adjusts to 2.2m simultaneously).

[0050] The mixed flow fan unit 53 can achieve continuous adjustment of wind speed. The mixed flow fan unit 53 includes: a variable frequency motor 15 (to control the speed), an impeller (to generate airflow) and a wind speed sensor (range 0-12m / s, accuracy ±0.2m / s), which supports continuous adjustment of wind speed of 0.5-10m / s. The speed frequency is controlled by the electronic device 6 (such as 20Hz corresponds to 2m / s, 50Hz corresponds to 8m / s), and the target wind speed is accurately output. For example: 15Hz (1.5m / s) is set when simulating a breeze environment, and 60Hz (10m / s) is set when simulating a strong wind environment. The mixed flow fan has the large flow rate of the axial flow fan and the high wind pressure characteristics of the centrifugal fan. It can not only output a stable wind speed, but also ensure that the wind covers the soil surface of the splash erosion simulation device 2 (covering diameter ≥1m).

[0051] The honeycomb-shaped flow guide device 54 enables stable airflow output. The structure of the honeycomb-shaped flow guide device 54 consists of a honeycomb grille (made of ABS plastic, with a single honeycomb aperture of 8mm and a length of 5cm) installed at the fan outlet, arranged in a regular hexagonal array. When the fan directly discharges air, there is turbulence (disturbed airflow and large fluctuations in wind speed). The honeycomb grille can "comb" the turbulent airflow into laminar flow (parallel airflow and uniform wind speed), reducing the turbulence at the air outlet from 30% to below 5%. Turbulence can cause uneven force on raindrops (some raindrops deflect excessively, while others do not deflect at all), while laminar flow ensures that raindrops at the same height are subjected to uniform wind force, making subsequent measurements of "raindrop kinetic energy and trajectory" more reliable (avoiding errors caused by turbulence).

[0052] After receiving information about parameters such as the height of the rainfall nozzle 13 and rainfall intensity, the electronic device 6 automatically adjusts the height of the second elevating support frame 52 and the wind speed of the fan (e.g., to match a higher wind speed during heavy rainfall to simulate severe weather). When wind acts on raindrops, the raindrop characteristic measurement device 3 captures the changes in the raindrop's posture (e.g., tilt angle) caused by the wind. This data is fed back to the electronic device 6 to verify that the "wind force-raindrop trajectory" correlation meets expectations. (If the deviation is too large, the electronic device 6 will fine-tune the fan parameters.)

[0053] In the soil splash erosion simulation and analysis system provided by the embodiment of the present application, the mixed flow fan unit 53 can accurately control the wind speed output by adjusting the speed frequency to meet the simulation requirements of different wind speed scenarios. The rotatable base 51 can flexibly adjust the wind direction under the control of the electronic device 6, and combined with the wind speed adjustment, it can reproduce different wind conditions. This design breaks through the limitation of single parameters in traditional wind simulation, can accurately simulate the complex wind conditions in the natural environment, and provide reliable wind parameter input for studying the impact of wind on raindrop movement and soil splash erosion. The second liftable support frame 52 can adjust the height to match the height of the rainfall nozzle 13, ensuring that the wind action range and the rainfall range effectively overlap, avoiding the problem of insufficient wind action on raindrops due to height differences; the honeycomb guide device 54 can reduce turbulence, make the output wind more stable, and ensure the uniformity of wind action on raindrops, thereby improving the authenticity of the "rainfall-wind" collaborative simulation and more closely fitting the actual scene of wind and rain acting together on the soil in the natural environment. The rotatable base 51, the second liftable support frame 52 and the mixed flow fan unit 53 are all connected to the electronic device 6 for communication, realizing the automation of wind direction, height and wind speed adjustment. Researchers can quickly set wind parameters through the electronic device 6 without manual adjustment, which greatly shortens the switching time of different wind scenes and improves experimental efficiency. It is especially suitable for multiple groups of comparative experiments that require frequent changes in wind conditions. The stable and controllable wind simulation combined with rainfall simulation can accurately capture the impact of wind on raindrop kinetic energy, raindrop motion trajectory and soil particle splash migration, providing high-quality experimental data to reveal the mechanism of wind in the soil splash erosion process, and facilitate deeper soil erosion research.

[0054] In an optional embodiment of the present application, as Figure 4 As shown, the splash simulation device 2 includes a splash tray 21, a collection drawer 22, and a waterproof cover 23. The splash tray 21 consists of a central source area 211 and a collection area 212. The central source area 211 is used to place a soil tray 213 with drainage holes at the bottom. The soil tray 213 is used to hold the target soil. The collection area 212 consists of multiple concentric rings, each separated by a steel plate 214. Each ring includes two symmetrical small holes for collecting splash materials, and the small holes in the ring areas of adjacent rings are staggered to maintain a sealed and waterproof environment. The collection drawer 22 is installed below the spatter tray 21 and is used to place the spatter material sample collection basin to improve the collection efficiency of the spatter material; The waterproof cover 23 is installed on the periphery of the splash plate 21 to ensure that raindrops only act on the soil plate 213.

[0055] Specifically, the splash plate 21 forms the main body of the device, consisting of a central source area 211 and a collection area 212. Made of stainless steel (corrosion-resistant and easy to clean), the central source area 211 (20 cm diameter) houses a soil plate 213 (15 cm diameter, 5 cm height), the site of raindrops' direct impact on the soil. The bottom of the soil plate 213 is equipped with three to five 2 mm diameter drainage holes to allow excess water to escape after rainfall (preventing soil accumulation and impacting the splash process). These holes are smaller than soil particles to prevent soil from leaking out. The target soil is loaded into the soil plate 213 (compacted to a bulk density of 1.3 g / cm³, simulating natural soil conditions). Raindrops falling from above directly impact the soil surface, creating splash particles (clods or individual particles) that are scattered in all directions.

[0056] Collection zone 212 (surrounding central source zone 211, 80 cm wide) consists of five concentric rings (5 cm, 10 cm, 20 cm, 30 cm, and 40 cm from the center, respectively; the distance from the edge of central source zone 211 is the particle migration distance). The rings are separated by 2 mm thick steel plates 214 (3 cm high to prevent mixing of particles from adjacent rings). Each ring has two symmetrical holes (1 cm in diameter) at the bottom to guide splashed material into the collection drawer 22 below. The holes in adjacent rings are staggered (e.g., holes at 0° and 180° in the 5 cm ring, and at 90° and 270° in the 10 cm ring) to prevent particles from outer rings from being "accidentally collected" by the inner holes. The gaps between the rings are sealed with silicone to prevent rainwater or splashed material from seeping in (ensuring that the collection volume of each ring exactly matches its migration distance).

[0057] The collection drawer 22 is an auxiliary structure that improves the efficiency of collecting splashed materials. It is horizontally embedded directly below the splash plate 21 (vertically corresponding to the small holes in the rings) and can be pulled out (similar to a drawer design). "Sample collection basins" matching the number of rings are placed in the drawer (each basin corresponds to a small hole in the ring area). The splashed materials falling from the small holes in the rings fall directly into the corresponding basins. This avoids the material residue caused by the traditional "direct collection at the bottom of the ring area" (the drawer can be completely pulled out, and the collection basin is easy to pour); the collection basin is marked with the ring area number (such as "5cm" and "10cm"), which can be directly corresponded to the migration distance during subsequent weighing, reducing data recording errors.

[0058] The waterproof cover 23 is a protective structure designed to eliminate secondary splash interference. Made of transparent acrylic (without obstructing observation), it is cylindrical and covers the periphery of the splash plate 21 (1.2m high and 1.2m in diameter). The bottom is sealed to the edge of the splash plate 21. This ensures that raindrops falling from the rainfall simulator 1 fall vertically only onto the soil plate 213 in the central source area 211 (preventing raindrops from splashing onto the collection area 212 or outside the device). If splash particles splash onto the inner wall of the waterproof cover 23, they slide along the wall to the corresponding annular area (rather than being dispersed by raindrops again), ensuring that the collected material is a product of "primary splash" (truly reflecting the initial splash state of the soil).

[0059] Raindrops impact the target soil surface, shattering sand particles and scattering them in all directions—small particles have a long range (falling into a 40cm ring), while large particles have a short range (falling into a 5cm ring). The splashed material from each ring is collected along the bottom slope to the small holes and falls into the corresponding collection basin in the collection drawer 22 below. A waterproof cover 23 prevents raindrops from escaping and protects the splashed particles from being disturbed by external airflow.

[0060] After the experiment, the collection drawer 22 was pulled out, and the materials in each basin were dried (105°C, 24 hours). After weighing, the “splashing amount at different migration distances” was obtained (e.g., 15 g for a 5 cm ring area and 3 g for a 40 cm ring area).

[0061] In the soil splash erosion simulation and analysis system provided in the embodiment of the present application, a central source area 211 is specifically used to place a soil pan 213 containing target soil, and a drainage hole is provided at the bottom of the soil pan 213. This ensures that raindrops directly splash the soil while also allowing excess water to be drained away in a timely manner. This design makes the initial conditions for splash erosion closer to nature, reducing the deviation between the experimental scenario and the actual environment. The collection area 212 is composed of multiple concentric rings with fixed spacing between the rings. The particle migration distance can be directly determined by the "ring area where the splash material is located" - small particles (easy to splash) generally fall into the far ring area, while large particles (resistant to breakage) fall into the near ring area. This hierarchical collection method can accurately capture the relationship between "soil particle size and migration distance", providing direct data support for analyzing soil erosion resistance (for example, soil with strong aggregate stability has a high proportion of large particles and a high amount of splash erosion in the near ring area). The circular rings are separated by steel plates 214 and sealed to prevent splashing materials from one ring zone from mixing into other ring zones due to water flow or vibration. The positions of the small holes in adjacent ring zones are staggered (for example, the small holes in the 5cm ring zone are at 0° and 180°, and those in the 10cm ring zone are at 90° and 270°) to prevent splashing materials from "crossing ring zones" and falling into the wrong collection holes. These two designs physically ensure that "the materials collected in each ring zone only come from splashing within the range of that ring zone," making the data reliable and reducing misjudgment of results due to collection errors. The collection drawer 22 is installed below the splash tray 21, and can directly hold the sample collection basin of the corresponding ring zone. The splashing materials naturally fall into the basin through the small holes in the ring zone, eliminating the need for manual collection. After the experiment, the drawer can be pulled out as a whole, and the samples can be quickly transferred to the drying and weighing stages, greatly improving operational efficiency. It is particularly suitable for multiple comparative experiments. The core function of the waterproof cover 23 is to block "secondary interference." Firstly, it confines raindrops to the soil pan 213 in the central source area 211, preventing them from splashing into the collection area 212 or outside the device. Secondly, it prevents splashed material already in the collection area 212 from being re-splashed by subsequent raindrops (i.e., "secondary splash"). Without the waterproof cover 23, material in the collection area 212 might be dispersed by raindrops and re-splashed, distorting the collected volume. The waterproof cover 23 ensures that the collected material is a direct product of "primary splash," accurately reflecting the soil's splash state under the initial impact. Data collected by the splash simulation device 2, such as "splash volume per annular zone," "total splash volume," and "splash material particle size (subsequently screened to obtain the average weight diameter of the splash material)," can be directly correlated with parameters of the rainfall simulator 1 (raindrop diameter and kinetic energy) and the wind simulator 5 (wind speed and direction), providing the foundational data for the electronic device 6 to establish an "external driver-splash erosion result" model.

[0062] In an optional embodiment of the present application, "a splash erosion prediction model is established between raindrop data and soil splash erosion data based on raindrop pH value, raindrop data, wind speed, wind direction, initial pH value, initial conductivity, target pH value, target conductivity, target erosion depth, target average weight diameter of splash erosion material, and target total splash erosion amount", such as Figure 5 As shown, the following steps may be included: Step S101 : calculating a conductivity difference based on an initial conductivity and a target conductivity; and calculating a pH difference based on an initial pH value and a target pH value.

[0063] Specifically, the electronic device 6 may calculate the conductivity difference by subtracting the initial conductivity from the target conductivity, and then calculate the pH difference by subtracting the initial pH from the target pH.

[0064] In step S102 , the pH value of raindrops, raindrop data, wind speed, and wind direction are input into the first branch of the initial splash erosion prediction network, and feature extraction is performed on the pH value of raindrops and raindrop data to obtain external driving features.

[0065] Specifically, the above step S102 may include the following steps: Step a1: Correct the raindrop diameter based on the wind speed to obtain a diameter correction coefficient.

[0066] Specifically, the electronic device 6 may compare the wind speed with a preset wind speed threshold. If the wind speed is greater than the preset wind speed threshold, the diameter correction coefficient is obtained by multiplying the raindrop diameter by a preset diameter coefficient.

[0067] For example, when wind speeds exceed 3m / s, large-diameter raindrops (>3mm) experience increased flattening due to wind shear forces, leading to a preset diameter coefficient of 1.1 for large raindrops. Small-diameter raindrops (<1mm) are easily deflected by wind, reducing their actual impact area by 5% compared to calm conditions. Therefore, the preset diameter coefficient is set to 0.95. After this correction, the error between the impact range characteristics at different wind speeds and the measured splash distribution radius is reduced to within 5%.

[0068] Step a2: Calculate the raindrop falling time based on the raindrop terminal velocity.

[0069] Specifically, the electronic device 6 may calculate the raindrop falling time based on the raindrop terminal velocity.

[0070] Step a3: Based on the soil type corresponding to the target soil, the raindrop falling time is corrected to obtain a time correction coefficient.

[0071] Specifically, the electronic device 6 may determine a preset time coefficient according to the soil type corresponding to the target soil and the raindrop falling time, and then multiply the raindrop falling time by the preset time coefficient to obtain a time correction coefficient.

[0072] For example, sandy soil: if the falling time of a raindrop is greater than 2s, the preset time coefficient is 0.95; if the falling time of a raindrop is less than 0.5s, the preset time coefficient is 1.15 (sandy soil particles are loose and easy to spread even if the falling time is short); clay soil: if the falling time of a raindrop is greater than 2s, the preset time coefficient is 0.85; if the falling time of a raindrop is less than 0.5s, the preset time coefficient is 1.05 (clay soil particles are cohesive, and the range shrinkage is more obvious when the falling time is long).

[0073] Step a4: Correcting the raindrop terminal velocity based on the wind speed to obtain a raindrop velocity correction coefficient.

[0074] Specifically, the electronic device 6 corrects the raindrop terminal velocity according to the wind speed based on the following formula to obtain a raindrop velocity correction coefficient.

[0075] Raindrop velocity correction factor = Raindrop terminal velocity × (1 + 0.02 × min(wind speed, 5)). (Note: Wind speed is in m / s. When the wind speed is greater than 5 m / s, it is calculated as 5 m / s to avoid excessive amplification of extreme wind speeds.)

[0076] It should be noted that for every 1 m / s increase in wind speed, the effective impact velocity of raindrops on the soil increases by approximately 2% (for example, for a moderate-speed raindrop in a wind speed of 3 m / s, the actual velocity coefficient = 1.2 × (1 + 0.02 × 3) = 1.272). This correction makes the velocity coefficient no longer fixed, but instead dynamically adapts to wind conditions.

[0077] Step a5: Calculate the uniformity coefficient based on the wind speed and the raindrop posture.

[0078] The raindrop posture is the angle α between the raindrop movement direction and the direction perpendicular to gravity. The electronic device 6 can calculate the uniformity coefficient based on the following formula.

[0079] Uniformity coefficient = 1.0 - 0.01 × wind speed - 0.02 × α. Wind speed is measured in m / s, α is measured in degrees, and the uniformity coefficient should be between 0.6 and 1.0.

[0080] For example, wind speed of 3 m / s, α = 10° → uniformity coefficient = 1.0-0.01×3-0.02×10 = 0.77 (between dispersion and concentration); wind speed of 5 m / s, α = 25° → coefficient = 1.0-0.01×5-0.02×25 = 0.45 (forced to increase to 0.6 to avoid being too low).

[0081] Step a6: Calculate the basic spatial range characteristics based on the diameter correction coefficient, the time correction coefficient, and the raindrop velocity correction coefficient.

[0082] Specifically, the electronic device 6 multiplies the diameter correction coefficient, the time correction coefficient, and the raindrop velocity correction coefficient to obtain the basic spatial range feature.

[0083] Step a7: Calculate the impact range characteristics based on the basic spatial range characteristics and the uniformity coefficient.

[0084] Specifically, the electronic device 6 may calculate the impact range feature by multiplying the basic spatial range feature by the uniformity coefficient.

[0085] Step a8: derive the raindrop kinetic energy density based on the raindrop kinetic energy divided by the diameter correction coefficient.

[0086] Specifically, the electronic device 6 obtains the raindrop kinetic energy density based on the raindrop kinetic energy divided by the diameter correction coefficient.

[0087] Step a9: Based on the wind direction, wind speed and raindrop movement posture, the kinetic energy of the raindrop is corrected to obtain a kinetic energy correction coefficient.

[0088] Specifically, electronic device 6 can calculate the angle between the wind direction and the raindrop's travel direction based on the wind direction and the raindrop's travel posture. This angle is then compared with a first preset angle threshold and a second preset angle threshold. If the angle between the wind direction and the raindrop's travel direction is less than the first preset angle threshold, the terminal velocity of the raindrop is multiplied by the first preset velocity coefficient to correct the terminal velocity of the raindrop, thereby obtaining a raindrop velocity correction coefficient. The kinetic energy of the raindrop is then corrected based on the raindrop velocity correction coefficient to obtain an initial kinetic energy coefficient.

[0089] If the angle between the wind direction and the raindrop's travel direction is greater than a second preset angle threshold, the terminal velocity of the raindrop is multiplied by the second preset velocity coefficient to correct the terminal velocity of the raindrop, thereby obtaining a raindrop velocity correction coefficient. The kinetic energy of the raindrop is then corrected based on the raindrop velocity correction coefficient to obtain an initial kinetic energy coefficient.

[0090] For example, when the angle between the wind direction and the raindrop's travel direction is less than 30°, the raindrop is boosted by the wind, and the terminal velocity of the raindrop increases by 5% (raindrop velocity correction coefficient = raindrop terminal velocity × 1.05). Then, based on the raindrop velocity correction coefficient, the raindrop kinetic energy is corrected to obtain the initial kinetic energy correction coefficient. If the angle between the wind direction and the raindrop's travel direction is greater than 60°, the terminal velocity of the raindrop decreases by 8% (raindrop velocity correction coefficient = raindrop terminal velocity × 0.92). Then, based on the raindrop velocity correction coefficient, the raindrop kinetic energy is corrected to obtain the initial kinetic energy coefficient.

[0091] Then, the electronic device 6 calculates the kinetic energy correction coefficient based on the initial kinetic energy coefficient. The specific formula is as follows: Kinetic energy correction coefficient = initial kinetic energy coefficient × (1 + 0.03 × min(wind speed, 6)). Wind speed is in m / s. If the wind speed is greater than 6 m / s, it is calculated as 6 m / s to avoid interference from extreme values.

[0092] Step a10, calculating the synergy factor based on the soil type corresponding to the target soil and the pH value of the raindrops; Specifically, electronic device 6 can determine a soil buffer coefficient based on the soil type corresponding to the target soil. The pH enhancement factor can be determined based on the corresponding relationship between the raindrop pH value and the pH enhancement factor. The synergy factor can then be calculated by multiplying the pH enhancement factor by the soil buffer coefficient.

[0093] Among them, when the target soil is sandy soil, the soil buffer coefficient is 1.2; when the target soil is loamy soil, the soil buffer coefficient is 1.0; when the target soil is clay soil, the soil buffer coefficient is 0.8.

[0094] Example: Acidic raindrops with pH = 4.0 (enhancement factor 1.3) act on sandy soil (buffering coefficient 1.2), the synergy factor = 1.3 × 1.2 = 1.5.

[0095] Step a11: Calculate the directional strength coefficient based on the raindrop movement posture and the angle between the wind direction and the impact direction.

[0096] Specifically, the raindrop's motion posture is used to characterize the angle α between the raindrop's motion direction and the direction perpendicular to gravity. Electronic device 6 can calculate the vertical intensity coefficient based on the raindrop's motion posture. The calculation formula is as follows: Vertical Intensity Coefficient = cos(α) × (1 - 0.005 × α); where α is the raindrop's motion posture, measured in degrees, with a value of 0° ≤ α ≤ 60°. When the angle is greater than 60°, the value is forcibly taken as 0.5 to avoid excessive discounting.

[0097] Electronic device 6 can calculate the angle between the wind direction and the impact direction based on the raindrop's trajectory and wind direction. The impact direction, in a splash erosion scenario, generally refers to the direction in which the raindrop (or other impactor) strikes the soil (in natural rainfall, this is typically approximately vertically downward; in simulated experiments, this may be tilted due to device angle adjustment).

[0098] Electronic device 6 then calculates the horizontal strength coefficient based on the angle between the wind direction and the impact direction. The calculation formula is as follows: horizontal strength coefficient = 1.0 + 0.003 × θ (θ is the angle between the wind direction and the impact direction, 0°≤θ≤90°). When θ = 0° (downwind): horizontal strength coefficient = 1.1 (wind assists particle migration, increasing horizontal strength by 10%). When θ=90° (crosswind): horizontal intensity coefficient = 1.27 (crosswind expands the particle diffusion range, indirectly enhancing horizontal damage).

[0099] Then, the electronic device 6 multiplies the vertical direction strength coefficient and the horizontal direction strength coefficient by the corresponding weight coefficients respectively, and adds them together to obtain the direction strength coefficient.

[0100] Step a12: Calculate the energy intensity based on the kinetic energy correction coefficient, the raindrop kinetic energy density, and the synergy factor.

[0101] Specifically, the electronic device 6 multiplies the kinetic energy correction coefficient, the raindrop kinetic energy density, and the synergy factor to obtain the energy intensity.

[0102] Step a13: Calculate the impact strength characteristics based on the energy intensity and directional strength coefficient.

[0103] Specifically, the electronic device 6 multiplies the energy intensity by the directional intensity coefficient to obtain the impact intensity characteristic.

[0104] Step a14: Fusion the impact range feature and the impact intensity feature to generate an external driving feature.

[0105] Specifically, electronic device 6 may compare the shock intensity feature with a preset shock intensity threshold. If the shock intensity feature exceeds the preset shock intensity threshold, electronic device 6 determines that the weight corresponding to the shock intensity feature is greater than the weight corresponding to the shock range feature. Then, based on the weights corresponding to the shock range feature and the shock intensity feature, the shock range feature and the shock intensity feature are fused to generate the external driving feature.

[0106] Step S103: input the conductivity difference and the pH value difference into the second branch of the initial splash prediction network, perform feature extraction on the conductivity difference and the pH value difference, and obtain soil response features.

[0107] Specifically, the above step S103 may include the following steps: Step b1: Based on the relationship between the conductivity difference and the pH value difference, correct the conductivity difference and the pH value difference to obtain a corrected conductivity coefficient and a corrected pH value coefficient.

[0108] Specifically, the electronic device 6 may correct the conductivity difference and the pH difference based on the positive and negative relationship between the conductivity difference and the pH difference to obtain a corrected conductivity coefficient and a corrected pH coefficient.

[0109] For example, if the pH difference is negative (soil acidification) and the conductivity difference is positive (increased salt loss), it means that the acidic conditions have exacerbated soil structural damage. In this case, the conductivity difference is multiplied by the conductivity coefficient to obtain a corrected conductivity coefficient (amplifying the effect of salt loss), and the pH difference is multiplied by the pH coefficient to obtain a corrected pH coefficient (strengthening the associated effect of acidification). Among them, the conductivity coefficient can be 1.1, 1.2, or other values. The pH coefficient can be 1.05, 1.06, or other values. The embodiments of this application do not specifically limit the conductivity coefficient and pH coefficient.

[0110] If the directions of change of the two are unrelated (e.g., the pH value remains unchanged but the conductivity increases), it is determined to be a non-splashing factor (e.g., natural salt migration in the soil), and the correction factor is multiplied by 0.8 (to weaken the interference).

[0111] Effect: Eliminate irrelevant interference, so that the corrected parameters can more accurately reflect the physical and chemical changes of soil caused by splash erosion, and avoid misjudgment of a single parameter.

[0112] Step b2: calculating the soil buffer index based on the first relationship between the corrected conductivity coefficient and the corrected pH coefficient.

[0113] Specifically, the electronic device 6 may calculate the soil buffer index based on the synergistic relationship (first relationship) between the corrected conductivity coefficient and the corrected pH coefficient: Formula example: Soil buffer index = 1 / (corrected pH coefficient + corrected conductivity coefficient / 2). A larger value indicates that the soil has a stronger ability to resist changes in acidity, alkali, and salinity (for example, when the buffer index is greater than 2, it means that the soil can effectively maintain physical and chemical balance).

[0114] This allows the two scattered physical and chemical parameters to be integrated into a single index, which intuitively reflects the soil's ability to withstand the physical and chemical disturbances caused by splash erosion.

[0115] Step b3: calculating the structural integrity characteristic based on the second relationship between the corrected conductivity coefficient and the corrected pH coefficient.

[0116] Specifically, the electronic device 6 may calculate the structural integrity feature based on the inverse correlation (second relationship) between the corrected conductivity coefficient and the corrected pH coefficient: Formula example: Structural integrity characteristic = 1-(corrected pH coefficient × 0.6 + corrected conductivity coefficient × 0.4) × soil texture coefficient.

[0117] Among them, for sandy soil, the soil texture coefficient is 1.2; for clay soil, the soil texture coefficient is 0.7. The larger the value, the more complete the soil structure (for example, when the structural integrity is greater than 0.7, the degree of aggregate fragmentation is low).

[0118] Step b4: Fusion the soil buffer index and the structural integrity characteristics to generate soil response characteristics.

[0119] Specifically, the electronic device 6 can integrate the soil buffer index (reflecting the physical and chemical stability) and the structural integrity characteristics (reflecting the anti-destruction ability of the structure).

[0120] Fusion logic: Dynamically allocate weights based on the buffering index - when the buffering index is greater than 2 (strong buffering), the weight of physical and chemical stability is 60%; when the buffering index is less than 1 (weak buffering), the weight of structural integrity is 70% (structural damage dominates corrosion resistance).

[0121] As a result, the output soil response characteristics cover both the anti-interference ability at the physical and chemical level and the anti-crushing ability at the structural level, completely quantifying the soil's intrinsic resistance potential to splash erosion and providing reliable "soil intrinsic property" input for subsequent fusion with external driving characteristics.

[0122] Step S104: fusing the external driving features and the soil response features to obtain target features.

[0123] Specifically, the external driving characteristics include impact range characteristics and impact intensity characteristics; the soil response characteristics include soil buffer index and structural integrity characteristics; the above step S104 may include the following steps: In step c1, initial weights are assigned to external driving features and soil response features respectively.

[0124] Specifically, the electronic device 6 may assign initial weights to the external driving characteristics and the soil response characteristics respectively.

[0125] For example, initial weights are assigned to the external driving features and soil response features (e.g., 50% each) to ensure that both are taken into account in the initial fusion: This allocation method avoids a single feature dominating the fusion result, laying the foundation for subsequent weight adjustment based on data correlation, so that the fusion process starts from an equilibrium starting point, taking into account the dual attributes of "external shock" and "soil resistance".

[0126] Step c2: construct a correlation coefficient matrix between external driving characteristics and soil response characteristics.

[0127] Specifically, the electronic device 6 can calculate the correlation coefficient between the external driving characteristics (such as impact range and impact intensity) and the soil response characteristics (such as soil buffer index and structural integrity) to clarify the correlation strength between the two, and construct a correlation coefficient matrix between the external driving characteristics and the soil response characteristics based on the calculated correlation coefficient.

[0128] For example, the impact intensity characteristics may be significantly negatively correlated with the structural integrity characteristics (correlation coefficient -0.8), indicating that the stronger the impact, the easier it is to destroy the soil structure; the impact range characteristics may be negatively correlated with the soil buffering index (correlation coefficient -0.7), indicating that the wider the impact range, the easier it is to break the physical and chemical balance of the soil.

[0129] Effect: The matrix intuitively presents the intrinsic connections between features, provides data basis for subsequent focus on key correlations, and avoids the ineffective fusion of unrelated features.

[0130] Step c3: determining target association features having correlation coefficients greater than a preset correlation threshold based on the correlation coefficient matrix.

[0131] Specifically, electronic device 6 can select target correlation feature pairs (e.g., "impact strength - structural integrity" and "impact range - soil buffer index") from the correlation coefficient matrix with correlation coefficients greater than a preset threshold (e.g., 0.7). Only highly correlated feature pairs are retained for subsequent calculations, while weakly correlated or uncorrelated features (e.g., impact range and structural integrity with a correlation coefficient less than 0.3) are eliminated to reduce redundant information interference. This allows the fusion process to focus on the relationships that have a core impact on the spatter erosion results, improving the targetedness and efficiency of feature fusion.

[0132] Step c4, performing cross operation on target related features to generate interactive features; Specifically, cross operations are performed on target-related features to generate interactive features.

[0133] Exemplarily, the interaction feature is calculated as follows: Structural excess impact = impact strength characteristic × (1-structural integrity characteristic); physical meaning: when the structural integrity characteristic = 0.5, if the impact strength characteristic = 0.8, then the excess impact = 0.8 × (1-0.5) = 0.4 - indicating that 40% of the impact strength exceeds the bearing capacity of the soil structure.

[0134] Physical and chemical excess shock = shock range characteristics × (1-soil buffer index); physical meaning: when the soil buffer index = 0.6, if the shock range characteristics = 0.7, then the excess shock = 0.7 × (1-0.6) = 0.28 - indicating that 28% of the shock range exceeds the physical and chemical buffering capacity of the soil, corresponding to "local soil acid-base imbalance."

[0135] The interactive characteristics directly quantify the "potential driving force of splash erosion". The higher the value (such as >0.5), the more it indicates that the damage to the soil caused by the impact has entered the "irreversible" stage.

[0136] In step c5, the initial weights are modified based on the interaction characteristics to obtain the target weights corresponding to the external driving characteristics and the soil response characteristics respectively.

[0137] Specifically, the electronic device 6 may modify the initial weights based on the interaction characteristics to obtain target weights corresponding to the external driving characteristics and the soil response characteristics, respectively.

[0138] For example, if the interaction eigenvalue is greater than 0.5 (the impact significantly exceeds the soil's bearing capacity), the weight of the external driving feature is increased (for example, from 50% to 60%); if the interaction eigenvalue is less than 0.3 (the impact is effectively resisted), the weight of the soil response feature is increased (for example, from 50% to 60%). This allows the weight distribution to no longer be fixed, but to dynamically adapt to the actual strength contrast of the "drive-response" relationship, ensuring that the fusion results are consistent with the actual mechanism of splash erosion.

[0139] Step c6: Fusing the external driving features and the soil response features based on the target weights to generate target features.

[0140] Specifically, electronic device 6 fuses the external driving characteristics with the soil response characteristics based on the target weights. The fusion result incorporates both the destructive force of the external impact and the soil's inherent erosion resistance. For example, when the external driving force is high and the soil response is low, the target characteristic value is high, indicating a high risk of splash erosion. This allows the target characteristics to fully capture the comprehensive conditions for splash erosion, providing comprehensive and accurate input for subsequent splash erosion prediction, avoiding the biased nature of single-dimensional assessments.

[0141] Step S105 , transferring the target features to the output layer, and outputting the virtual ablation depth, the virtual average weight diameter of the sputtered material, and the virtual total sputtered amount.

[0142] Specifically, the electronic device 6 may transfer the target features to the output layer, and output the virtual ablation depth, the virtual average weight diameter of the sputtered material, and the virtual total sputtered amount.

[0143] Step S106 , based on the relationship between the virtual ablation depth, the virtual average weight diameter of the sputtered material, and the virtual total sputtered amount and the target ablation depth, the target average weight diameter of the sputtered material, and the target total sputtered amount, the initial sputtering prediction network is trained to obtain a sputtering prediction model.

[0144] Specifically, the above step S106 may include the following steps: Step d1, calculating the erosion relative error and the erosion absolute error based on the virtual erosion depth and the target erosion depth; determining first weight information corresponding to the erosion relative error and the erosion absolute error, respectively, according to the soil type corresponding to the target soil; and calculating the erosion error based on the first weight information, the erosion relative error, and the erosion absolute error; Specifically, the electronic device 6 can calculate the relative erosion error and the absolute erosion error based on the virtual erosion depth and the target erosion depth. The relative erosion error = |virtual erosion depth-target erosion depth|; the absolute erosion error = |(virtual erosion depth-target erosion depth) / target erosion depth|.

[0145] Since different soil types have different erosion sensitivities (eg, sandy soil is easily eroded, while clay soil is difficult to erode), the electronic device 6 may determine first weight information corresponding to the relative erosion error and the absolute erosion error, respectively, based on the soil type corresponding to the target soil.

[0146] For example, the soil type is sandy soil: when the target erosion depth is greater than 1 cm, the first weight corresponding to the absolute erosion error is increased to 0.7 (because deep erosion has a greater impact on the productivity of sandy soil), and the first weight corresponding to the relative erosion error is 0.3; the soil type is clay soil: when the target erosion depth is less than 0.5 cm, the weight corresponding to the relative erosion error is increased to 0.5 (it is more sensitive to shallow erosion and needs to be captured accurately), and the first weight corresponding to the absolute erosion error is 0.5.

[0147] Then, the error is calculated based on the first weight information, the relative erosion error, and the absolute erosion error, so that the erosion error calculation is adapted to the soil type, avoiding a "one-size-fits-all" assessment of different soil types.

[0148] Example of the corrected formula (sandy soil): Erosion error = 0.7 × |virtual erosion depth - target erosion depth| + 0.3 × |(virtual erosion depth - target erosion depth) / target erosion depth| (when target erosion depth is greater than 1 cm).

[0149] Step d2: calculating the square error of the average weight diameter of the sputtered material based on the virtual average weight diameter of the sputtered material and the target average weight diameter of the sputtered material.

[0150] Specifically, the electronic device 6 may calculate the square error of the average weight diameter of the sputtered material based on the virtual average weight diameter of the sputtered material and the target average weight diameter of the sputtered material.

[0151] The square error of the average weight diameter of the spattered material = (the average weight diameter of the virtual spattered material - the average weight diameter of the target spattered material)².

[0152] Step d3, calculating the proportion of first particles having a diameter greater than a preset diameter in the average weight diameter of the virtual sputtered material and the proportion of second particles having a diameter greater than a preset diameter in the average weight diameter of the target sputtered material.

[0153] Logically speaking, the average weight diameter of the spattered material is a key indicator for measuring the water stability of the agglomerates (the larger the average weight diameter of the spattered material, the stronger the water stability of the agglomerates), and the proportion of particles with a diameter greater than the preset diameter threshold is an important auxiliary feature reflecting the integrity of the agglomerates - a high proportion of large particles usually corresponds to a larger average weight diameter of the spattered material and more stable agglomerates.

[0154] Therefore, the electronic device 6 can calculate the proportion of first particles having a diameter greater than the preset diameter in the virtual average weight diameter of the sputtered material and the proportion of second particles having a diameter greater than the preset diameter in the target average weight diameter of the sputtered material.

[0155] Among them, the preset diameter can be 0.25mm or 0.26mm. The implementation of this application does not make any specific limitation on the preset diameter.

[0156] Step d4, calculating the average weight diameter error of the sputtered material according to the deviation between the first particle ratio and the second particle ratio and the square error of the average weight diameter of the sputtered material.

[0157] Specifically, the electronic device 6 calculates the deviation between the first particle ratio and the second particle ratio, and then calculates the average weight diameter error of the sputtered material based on the formula.

[0158] The average weight diameter error of the spattered material is [(the average weight diameter of the virtual spattered material - the average weight diameter of the target spattered material)² × 1.2] × [1 + a × I (the deviation of the particle ratio > 15%)]. Where I is the indicator function, which takes the value of 1 if the condition is met and 0 otherwise, and a is the first penalty coefficient.

[0159] For example, if the proportion of particles >0.25 mm calculated from the virtual average weight diameter of the spattered material deviates by more than 15% from the proportion of particles corresponding to the actual target average weight diameter of the spattered material, it means that although the virtual value may be close to the target in terms of diameter value, it does not truly reflect the actual state of the particle composition (for example, the virtual average weight diameter of the spattered material meets the standard but the proportion of large particles is insufficient, which may underestimate the degree of agglomerate breakup).

[0160] Therefore, adding an additional penalty (e.g., 0.2) to the original squared error ((virtual average weight diameter of spattered material - target average weight diameter of spattered material)² × 1.2) essentially forces the virtual value to match the particle distribution characteristics of the target value through quantitative means, avoiding the pursuit of diameter matching while ignoring the inherent consistency of soil structure. This makes the error calculation more consistent with the actual mechanism that "the average weight diameter of spattered material reflects aggregate stability." This links the error in the average weight diameter of spattered material to the actual particle distribution, avoiding the one-sidedness of focusing only on diameter while ignoring particle composition.

[0161] Step d5: calculating the total sputtering amount error according to the virtual total sputtering amount and the target total sputtering amount.

[0162] Specifically, the electronic device 6 calculates the absolute value of the difference between the virtual total sputtering amount and the target total sputtering amount as the basic error.

[0163] Then, based on the virtual total sputtering amount, the virtual proportion of the virtual ring area sputtering amount corresponding to each ring area in the collection area 212 to the virtual total sputtering amount is calculated; then based on the target total sputtering amount, the target proportion of the target ring area sputtering amount corresponding to each ring area in the collection area 212 to the target total sputtering amount is calculated.

[0164] For each ring zone, the ring zone deviation between the virtual proportion and the target proportion corresponding to each ring zone is calculated. If the ring zone deviation corresponding to at least one ring zone is greater than the preset ring zone deviation threshold, a second penalty coefficient is determined.

[0165] Then, the total sputtering amount error is calculated based on the second penalty coefficient and the basic error.

[0166] Exemplarily, the calculation formula is: total sputtering error = basic error × [1 + b × I (annular area ratio deviation > 10%)].

[0167] Where I is the indicator function, which takes 1 when the condition is met and 0 otherwise, and b is the first penalty coefficient.

[0168] Step d6: generating a target loss function based on the ablation error, the square error of the average weight diameter of the sputtered material, and the total sputtering amount error.

[0169] Specifically, the electronic device 6 can generate a target loss function based on the ablation error, the square error of the average weight diameter of the sputtered material, and the total sputtering amount error. The formula is as follows: Target loss function = (erosion error × c + square error of average weight diameter of splashed material × d + total splash amount error × f) × risk coefficient × dynamic balance coefficient.

[0170] Among them, c, d, and f are all weighted data. The risk coefficient is: 1.2 when the target total spattering amount is greater than 200g / m²; 1.0 when it is 100-200g / m²; and 0.8 when it is less than 100g / m². The dynamic balance coefficient is: after every 20 rounds of training, if the contribution of an error term is greater than 50%, the corresponding weight is reduced by 10% (for example, the error weight of the average weight diameter of the spattering material is reduced from 0.4 to 0.36), and the balance coefficient is adjusted synchronously to maintain the total weight logic).

[0171] Step d7: training the initial sputtering prediction network based on the target loss function to obtain a sputtering prediction model.

[0172] Specifically, the electronic device 6 trains the initial sputtering prediction network based on the target loss function until the function value corresponding to the target loss function stabilizes at a preset function value, stops training, and obtains the sputtering prediction model.

[0173] The soil splash erosion simulation and analysis system provided in the embodiment of the present application can directly quantify the changes in soil physical and chemical properties before and after rainfall by calculating the difference in electrical conductivity and pH value. The difference in electrical conductivity reflects the degree of soil salt loss or migration, and the difference in pH value reflects the change in soil acid-base balance. The two together serve as the "physical and chemical signal" of the impact of soil splash erosion, providing a quantitative basis for the subsequent evaluation of soil erosion resistance. The raindrop diameter is corrected based on the wind speed to obtain a diameter correction coefficient, so that the corrected diameter correction coefficient is more consistent with the actual impact area. The raindrop falling time is calculated based on the raindrop terminal velocity. Based on the soil type corresponding to the target soil, the raindrop falling time is corrected to obtain a time correction coefficient, so that the influence of the time factor on the impact range matches the actual characteristics of the soil, thereby improving the rationality of the basic spatial range. The raindrop terminal velocity is corrected based on the wind speed to obtain a raindrop velocity correction coefficient. The uniformity coefficient is calculated based on the wind speed and the raindrop posture. Based on the diameter correction coefficient, the time correction coefficient and the raindrop velocity correction coefficient, the basic spatial range characteristics are calculated to avoid the one-sidedness dominated by a single parameter. Based on the basic spatial range characteristics and the uniformity coefficient, the impact range characteristic is calculated. This not only reflects the spatial breadth of raindrop coverage but also reflects the distribution concentration through uniformity, aligning with the reality of uneven raindrop distribution caused by wind in natural rainfall. The raindrop kinetic energy density is calculated by dividing the raindrop kinetic energy by the diameter correction coefficient. This standardizes the energy per unit volume of raindrops and eliminates the apparent kinetic energy advantage of large-diameter raindrops. The raindrop kinetic energy is corrected based on wind direction, wind speed, and raindrop motion to obtain a kinetic energy correction coefficient, which more closely approximates the actual impact energy. A synergy factor is calculated based on the target soil type and the pH value of the raindrop. A directional intensity coefficient is calculated based on the raindrop motion and the angle between wind direction and the impact direction. This accurately captures the effective vertical and horizontal energy contribution and avoids neglecting the influence of direction on destructive force. Energy intensity is calculated based on the kinetic energy correction coefficient, raindrop kinetic energy density, and the synergy factor. The impact intensity characteristic is calculated based on the energy intensity and directional intensity coefficient. The impact range characteristic and impact intensity characteristic are combined to generate an external driving characteristic. The output external driving characteristics cover both spatial distribution and energy intensity, fully describing the external effects of raindrops on the soil and providing comprehensive input for subsequent splash erosion prediction.

[0174] The second branch corrects the conductivity and pH differences based on the relationship between them, generating corrected conductivity and pH coefficients. This eliminates irrelevant interference, making the corrected parameters more accurately reflect the physical and chemical changes in the soil caused by splash erosion and avoiding misjudgment based on a single parameter. Based on the first relationship between the corrected conductivity and pH coefficients, the soil buffering index is calculated. These two discrete corrected conductivity and pH coefficients are integrated into a single index that intuitively reflects the soil's ability to withstand the physical and chemical disturbances caused by splash erosion. Based on the second relationship between the corrected conductivity and pH coefficients, the structural integrity characteristic is calculated. This infers the soil's structural state from physical and chemical parameters, addressing the complexity of direct structural measurement. The soil buffering index and structural integrity characteristic are then integrated to generate a soil response signature. The output soil response signature encompasses both physical and chemical interference resistance and structural resistance to fragmentation, fully quantifying the soil's inherent resistance to splash erosion and providing reliable "intrinsic soil property" input for subsequent integration with externally driven characteristics.

[0175] Initial weights are assigned to external driving features and soil response features separately to prevent a single feature from dominating the fusion results. This lays the foundation for subsequent weight adjustments based on data correlation, ensuring that the fusion process starts from a balanced starting point and accounts for both the "external impact" and "soil resistance" attributes. A correlation coefficient matrix is ​​constructed for the external driving features and soil response features. This matrix visually demonstrates the intrinsic connections between features, providing a data basis for subsequent focus on key relationships and avoiding ineffective fusion of unrelated features. Based on this correlation matrix, target related features with correlation coefficients greater than a preset correlation threshold are identified, allowing the fusion process to focus on relationships that have a significant impact on the splash erosion results, improving the targetedness and efficiency of feature fusion. A cross-operation is performed on these target related features to generate interactive features, transforming abstract relationships into quantifiable features that accurately reflect the dynamic game between "external impact and soil resistance" and providing a key basis for weight adjustment. Based on the interactive features, the initial weights are modified to obtain target weights for the external driving features and soil response features, respectively. This eliminates the need for fixed weight assignments and instead dynamically adapts to the actual strength of the "driver-response" relationship, ensuring that the fusion results accurately reflect the true mechanism of splash erosion. Based on target weights, the external driving features and soil response features are fused to generate target features. These target features fully capture the comprehensive conditions under which splash erosion occurs, providing comprehensive and accurate input for subsequent splash erosion prediction and avoiding the biased nature of single-dimensional assessments. The output layer generates virtual erosion depth, virtual average weight diameter of splashed material, and virtual total splash erosion volume. These three prediction targets correspond to the depth of soil physical damage, changes in aggregate stability, and total material loss, respectively, encompassing the core dimensions of splash erosion assessment.

[0176] Next, for "erosion depth," the model combines absolute and relative errors, assigning different weights based on soil type to avoid the limitations of a single error metric and making the erosion-related error calculation more tailored to soil characteristics. For "splashed material particles," the model not only quantifies the overall particle size deviation using the "mean weight-diameter squared error," but also focuses on the splash patterns of coarse particles using the "deviation of the proportion of particles with diameters greater than a preset value." These two metrics combine to more comprehensively reflect the particle distribution characteristics of the splashed material, avoiding the loss of detail that results from focusing solely on the mean diameter. For "total splash amount," the model directly quantifies the total amount deviation to ensure the model's accuracy in predicting the overall splash amount. This multi-dimensional error coverage allows the model to simultaneously learn core metrics such as erosion depth, particle characteristics, and total amount, resulting in predictions that are more accurate than the actual splash process. The objective loss function integrates the erosion error, particle diameter error, and total splash amount error, rather than relying on a single metric. This "multi-objective fusion" design avoids optimizing one metric at the expense of others, ensuring the model's generalization across diverse scenarios. Even with varying input conditions, the model maintains stable predictions across multiple key dimensions. The trained splash erosion prediction model can accurately output key parameters such as erosion depth, particle distribution, and total splash erosion amount, and can provide data support for soil erosion prevention and control - for example, by predicting the splash erosion risk of specific soils under rainfall, vegetation cover, engineering protection and other measures can be formulated in advance.

[0177] In an optional embodiment of the present application, as Figure 6 As shown, the electronic device 6 is further configured to perform the following steps: Step S201: Calculate an anti-erosion capability index based on a target erosion depth.

[0178] Specifically, the electronic device 6 may obtain the soil type corresponding to the target soil, and then correct the target erosion depth according to the soil type to obtain the anti-erosion capability index.

[0179] For example, different soil types have different basic erosion resistance capabilities, so a texture correction factor is introduced (0.8 for sandy soil, 1.0 for loamy soil, and 1.2 for clay soil). The corrected formula is: Erosion Resistance Index = [1 / (1 + target erosion depth)] × texture correction factor. The specific logic is: Clay soil has strong cohesion and higher actual erosion resistance at the same erosion depth; sandy soil is prone to stripping and requires a lower score. For example, for clay soil with a target erosion depth of 0.3 cm, the Erosion Resistance Index = [1 / (1 + 0.3)] × 1.2 ≈ 0.92, which better reflects the actual erosion resistance characteristics of clay soil.

[0180] Optionally, if the target erosion depth is greater than a preset depth threshold, the anti-erosion capability index is additionally multiplied by a corresponding penalty coefficient.

[0181] For example, if the target erosion depth is greater than 2 cm (exceeding 40% of the height of the soil pan 213 , assuming the soil pan 213 is 5 cm high in this application), an additional penalty factor of 0.8 is multiplied because deep erosion causes more irreversible damage to the soil structure.

[0182] Step S202: calculating the anti-breakage capability index of the agglomerates based on the average weight diameter of the target sputtered material.

[0183] Specifically, the electronic device 6 can determine the average weight diameter threshold of the spattered material corresponding to the target soil based on the soil type corresponding to the target soil, and then divide the target average weight diameter of the spattered material by the average weight diameter threshold of the spattered material to obtain the initial aggregate anti-crushing ability. The electronic device 6 then calculates the proportion of second particles with a diameter greater than a preset diameter in the target average weight diameter of the spattered material. If the proportion of second particles is greater than the first preset proportion threshold, the initial aggregate anti-crushing ability is multiplied by a first coefficient to obtain an aggregate anti-crushing ability index. If the proportion of second particles is less than the second preset proportion threshold, the initial aggregate anti-crushing ability is multiplied by a second coefficient to obtain an aggregate anti-crushing ability index. Wherein, the first preset proportion threshold is greater than the second preset proportion threshold, the first coefficient is greater than 1, and the second coefficient is less than 1. If the proportion of second particles is greater than or equal to the second preset proportion threshold and less than or equal to the first preset proportion threshold, the initial aggregate anti-crushing ability is determined as the aggregate anti-crushing ability index.

[0184] For example, a threshold for the average weight diameter of splash material is set based on soil type (2mm for sandy soil, 3mm for loamy soil, and 4mm for clay soil). The formula is adjusted to: Initial aggregate resistance to breakage = Target average weight diameter of splash material divided by the average weight diameter threshold of splash material. If the proportion of particles >0.25mm in the splash material is >60% (the first preset proportion threshold) (predominantly coarse particles), the aggregate resistance to breakage is multiplied by 1.1 (coarse particle aggregates are more stable); if it is <20% (the second preset proportion threshold) (predominantly fine particles), the resistance is multiplied by 0.9 (fine particles are more easily dispersed).

[0185] Example optimization: For sandy soil with an average weight diameter of 2.1 mm for target splash material (threshold 2 mm), this indicator is 2.1 / 2 = 1.0 (capped at 1.0). Since coarse particles account for 70%, the final indicator is 1.0 × 1.1 = 1.1 (normalized to 1.0), which more accurately reflects the actual stability of sandy soil aggregates.

[0186] Step S203: Calculate the overall anti-loss capability index based on the target total sputtering amount.

[0187] Specifically, the electronic device 6 may calculate the initial anti-loss capability based on the formula: initial anti-loss capability=1 / (1+target total sputtering amount / 100).

[0188] For example, the target total spattering amount is 80 g / m², and the initial anti-loss ability = 1 / (1+80 / 100)≈0.56.

[0189] The electronic device 6 then calculates the proportion of the spattering amount in the farthest annular region to the target total spattering amount and the proportion of the spattering amount in the closest annular region to the target total spattering amount. If the proportion of the spattering amount in the farthest annular region to the target total spattering amount is greater than a third preset proportion threshold, the electronic device 6 multiplies the proportion by a third coefficient; if the proportion of the spattering amount in the closest annular region to the target total spattering amount is greater than a fourth preset proportion threshold, the electronic device 6 multiplies the proportion by a fourth coefficient.

[0190] The third preset ratio threshold value may be 15% or 18%, and the fourth preset ratio threshold value may be 60% or 62%. The third preset ratio threshold value and the fourth preset ratio threshold value are not specifically limited in this embodiment of the application. The third coefficient is less than 1, and the fourth coefficient is greater than 1.

[0191] For example, if the 40cm ring area accounts for >15% of the splash volume (particles migrate far), the anti-loss capability is multiplied by 0.9 (wider splash range, higher risk). If the 0cm ring area accounts for >60% (particles are not dispersed), the anti-loss capability is multiplied by 1.1 (concentrated splash, easier to control). For a target total splash volume of 80g / m², and a 40cm ring area accounts for 20%, the overall anti-loss capability index = 1 / (1 + 80 / 100) × 0.9 ≈ 0.50, which better reflects the impact of splash range on risk.

[0192] Optionally, if the target total spattering amount is greater than a preset total spattering amount threshold (e.g., 200 g / m²) (high risk), regardless of the calculated result, the maximum anti-loss capability is the preset anti-loss capability threshold (e.g., 0.6)—strengthening risk warnings for high-loss scenarios.

[0193] Step S204: obtaining the target anti-erosion capacity corresponding to the target soil based on the anti-erosion capacity index, the aggregate anti-breakage capacity index, and the overall anti-loss capacity index.

[0194] Specifically, the electronic device 6 may assign weights according to the physical meanings of the erosion resistance index, the aggregate anti-breakage ability index, and the overall anti-loss ability index, and fuse them to obtain the target corrosion resistance.

[0195] For example, the weight distribution is: aggregate anti-breakage ability index (40%) > anti-erosion ability index (30%) > overall anti-loss ability index (30%) (because aggregate stability is the core of corrosion resistance).

[0196] Calculation formula: Target corrosion resistance = (aggregate anti-crushing ability index × 0.4) + (anti-erosion ability index × 0.3) + (overall anti-loss ability index × 0.3).

[0197] Then, the electronic device 6 maps the target corrosion resistance to the interval [0,1] and divides it into 5 levels to clarify the strength of the corrosion resistance: extremely strong (0.8-1.0): the agglomerates are intact, the erosion is slight, and there is almost no particle loss; strong (0.6-0.8): the agglomerates are relatively stable, the erosion is shallow, and the loss is small (for example, 0.67 in the example corresponds to "strong"); medium (0.4-0.6): the agglomerates are partially broken, the erosion is moderate, and there is a certain loss; weak (0.2-0.4): the agglomerates are largely broken, the erosion is deep, and the loss is obvious; extremely weak (0-0.2): the agglomerates are completely broken, the erosion is serious, and the loss is large.

[0198] The soil splash erosion simulation and analysis system provided in the present application quantifies soil erosion resistance across three key erosion stages: erosion, aggregate fragmentation, and loss, avoiding the biased assessment of soil erosion resistance based on a single indicator. Integrating indicators across these three stages comprehensively covers the soil's erosion resistance from splashing and stripping to particle fragmentation and final loss, reflecting the overall erosion resistance level. This system can directly identify weak links in soil erosion resistance (e.g., weak erosion resistance and easily broken aggregates), providing a clear basis for targeted improvements (e.g., increasing the application of organic fertilizer to enhance aggregate stability).

[0199] Although the embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention. Such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A soil splash erosion simulation and analysis system, characterized in that: The system includes: a rainfall simulator, a splash erosion simulation device, a raindrop characteristic measurement device, a data acquisition device, a wind simulator, and an electronic device, wherein the rainfall simulator is installed above the raindrop characteristic measurement device, the raindrop characteristic measurement device is installed above the splash erosion simulation device, and the wind simulator is installed on one side of the raindrop characteristic measurement device; the raindrop characteristic measurement device, the wind simulator, and the data acquisition device are all communicatively connected to the electronic device, wherein: The rainfall simulator is used to simulate rainfall of different intensities and different pH values; The wind simulator is used to adjust the wind speed and direction under the control of the electronic device; The raindrop characteristic measuring device is used to measure raindrop data, wherein the raindrop data includes raindrop diameter, raindrop terminal velocity, raindrop kinetic energy and raindrop movement posture; The splash erosion simulation device is used to simulate the process of raindrops landing on the target soil, causing splash erosion and generating splash erosion materials; The data acquisition device is used to collect the initial pH value and initial conductivity of the target soil before rainfall; and to collect the target pH value, target conductivity, target erosion depth of the target soil after rainfall, and the target average weight diameter of the splash material and the target total splash amount corresponding to the splash material; The electronic device is used to establish a splash erosion prediction model between raindrop data and soil splash erosion data based on the raindrop pH value, the raindrop data, the wind speed, the wind direction, the initial pH value, the initial conductivity, the target pH value, the target conductivity, the target erosion depth, the target average weight diameter of the splashing material, and the target total splashing amount; and to predict the predicted erosion depth, the predicted average weight diameter of the splashing material, and the predicted total splashing amount corresponding to other soils to be tested based on the splash erosion prediction model.

2. The soil splash erosion simulation and analysis system according to claim 1, characterized in that: The rainfall simulator includes: a rain barrel, a regulating pump, a rainfall nozzle, a first liftable support frame, and a motor; wherein the regulating pump is connected to the rain barrel and the rainfall nozzle, the motor is installed on the rainfall nozzle, the first liftable support frame is installed below the rainfall nozzle, and the regulating pump, the rainfall nozzle, the first liftable support frame, and the motor are all communicatively connected to the electronic device, wherein: The rainwater barrel is used to hold rainwater with different pH values; The regulating pump is used to control the flow of water pumped to the rainfall nozzle by rotating speed under the control of the electronic device to simulate rainfall of different intensities; The motor is used to drive the rainfall sprinkler head to rotate under the control of the electronic device to simulate the effect of natural rainfall; The rainfall nozzle is used to adjust different calibers under the control of the electronic device to adjust the diameter of raindrops, thereby simulating raindrops of different sizes; The first liftable support frame is used to adjust the height of the rainfall nozzle under the control of the electronic device, thereby controlling the height of the rainfall.

3. The soil splash erosion simulation and analysis system according to claim 1, characterized in that: The wind simulator includes a rotatable base, a second liftable support frame, a mixed flow fan unit, and a honeycomb guide device installed at the air outlet of the mixed flow fan unit. The rotatable base, the second liftable support frame, and the mixed flow fan unit are all communicatively connected to the electronic device, wherein: The rotatable base is used to rotate under the control of the electronic device to adjust the wind direction; The second elevating support frame is used to adjust its height under the control of the electronic device to match the height of the rainfall nozzle in the rainfall simulator; The mixed flow fan unit is used to adjust the rotation speed and frequency under the control of the electronic equipment, thereby adjusting the wind speed; The honeycomb-shaped flow guide device is used to effectively reduce the turbulence phenomenon generated when the fan discharges air.

4. The soil splash erosion simulation and analysis system according to claim 1, characterized in that: The splash erosion simulation device includes a splash erosion tray, a collection drawer, and a waterproof cover. The splash erosion tray consists of a central source area and a collection area. The central source area is used to place a soil tray with drainage holes at the bottom; the soil tray is used to hold the target soil. The collection area consists of multiple concentric rings, each separated by a steel plate. The ring area of ​​each ring includes two symmetrical small holes for collecting splash erosion materials, and the small holes of adjacent rings are staggered to maintain a sealed and waterproof environment. The collection drawer is installed below the spatter plate and is used to place a spatter material sample collection basin to improve the collection efficiency of the spatter material; The waterproof cover is installed on the periphery of the splash plate to ensure that raindrops only act on the soil plate.

5. The soil splash erosion simulation and analysis system according to claim 1, characterized in that: The electronic device is used to calculate a conductivity difference based on the initial conductivity and the target conductivity; and calculate a pH difference based on the initial pH value and the target pH value; Inputting the raindrop pH value, the raindrop data, the wind speed, and the wind direction into a first branch of the initial splash erosion prediction network, performing feature extraction on the raindrop pH value and the raindrop data to obtain external driving features; Inputting the conductivity difference and the pH value difference into the second branch of the initial splash prediction network, performing feature extraction on the conductivity difference and the pH value difference to obtain soil response features; fusing the external driving feature and the soil response feature to obtain a target feature; The target features are transferred to the output layer, and the virtual ablation depth, the virtual average weight diameter of the spattered material and the virtual total spattered amount are output; Based on the relationship between the virtual erosion depth, the virtual average weight diameter of the sputtering material, and the virtual total sputtering amount and the target erosion depth, the target average weight diameter of the sputtering material, and the target total sputtering amount, the initial sputtering prediction network is trained to obtain the sputtering prediction model.

6. The soil splash erosion simulation and analysis system according to claim 5, characterized in that: The electronic device is configured to correct the raindrop diameter based on the wind speed to obtain a diameter correction coefficient; Calculating raindrop falling time based on the raindrop terminal velocity; Correcting the raindrop falling time based on the soil type corresponding to the target soil to obtain a time correction coefficient; Correcting the raindrop terminal velocity based on the wind speed to obtain a raindrop velocity correction coefficient; Calculating a uniformity coefficient based on the wind speed and the raindrop posture; Calculating a basic spatial range feature based on the diameter correction coefficient, the time correction coefficient, and the raindrop velocity correction coefficient; Calculating an impact range characteristic based on the basic spatial range characteristic and the uniformity coefficient; Obtaining a raindrop kinetic energy density based on the raindrop kinetic energy divided by the diameter correction coefficient; Correcting the kinetic energy of the raindrops based on the wind direction, the wind speed, and the raindrop movement posture to obtain a kinetic energy correction coefficient; Calculating a synergistic factor according to the soil type corresponding to the target soil and the pH value of the raindrop; Calculating a directional intensity coefficient based on the raindrop movement posture and the angle between the wind direction and the impact direction; Calculating energy intensity based on the kinetic energy correction coefficient, the raindrop kinetic energy density, and the synergy factor; Calculating an impact strength characteristic based on the energy intensity and the directional strength coefficient; The impact range feature and the impact intensity feature are fused to generate the external driving feature.

7. The soil splash erosion simulation and analysis system according to claim 5, characterized in that: The electronic device is configured to correct the conductivity difference and the pH difference based on a relationship between the conductivity difference and the pH difference to obtain a corrected conductivity coefficient and a corrected pH coefficient; Calculating a soil buffer index based on a first relationship between the corrected conductivity coefficient and the corrected pH coefficient; calculating a structural integrity characteristic based on a second relationship between the corrected conductivity coefficient and the corrected pH coefficient; The soil buffer index and the structural integrity characteristic are fused to generate the soil response characteristic.

8. The soil splash erosion simulation and analysis system according to claim 5, characterized in that: The external driving characteristics include an impact range characteristic and an impact intensity characteristic; the soil response characteristics include a soil buffer index and a structural integrity characteristic; the electronic device is used to assign initial weights to the external driving characteristics and the soil response characteristics respectively; Constructing a correlation coefficient matrix between the external driving characteristics and the soil response characteristics; Determining, based on the correlation coefficient matrix, target correlation features having a correlation coefficient greater than a preset correlation threshold; Performing cross operations on the target-related features to generate interactive features; Based on the interaction characteristics, the initial weights are modified to obtain target weights corresponding to the external driving characteristics and the soil response characteristics respectively; The external driving feature and the soil response feature are fused based on the target weight to generate the target feature.

9. The soil splash erosion simulation and analysis system according to claim 5, characterized in that: The electronic device is configured to calculate an erosion relative error and an erosion absolute error based on the virtual erosion depth and the target erosion depth; determine first weight information corresponding to the erosion relative error and the erosion absolute error, respectively, based on the soil type corresponding to the target soil; and calculate an erosion error based on the first weight information, the erosion relative error, and the erosion absolute error; Calculating a square error of the average weight diameter of the sputtered material based on the virtual average weight diameter of the sputtered material and the target average weight diameter of the sputtered material; Calculating the proportion of first particles with a diameter greater than a preset diameter in the virtual average weight diameter of the sputtered material and the proportion of second particles with a diameter greater than the preset diameter in the target average weight diameter of the sputtered material; Calculating an average weight diameter error of the sputtered material according to a deviation between the first particle ratio and the second particle ratio and a square error of an average weight diameter of the sputtered material; Calculating a total sputtering amount error according to the virtual total sputtering amount and the target total sputtering amount; generating a target loss function based on the ablation error, the square error of the average weight diameter of the sputtered material, and the total sputtering amount error; The initial sputtering prediction network is trained based on the target loss function to obtain the sputtering prediction model.

10. The soil splash erosion simulation and analysis system according to claim 1, characterized in that: The electronic device is further configured to calculate an anti-erosion capability index based on the target erosion depth; Calculating the anti-crushing ability index of the agglomerate based on the average weight diameter of the target spattered material; Calculating an overall anti-loss capability index based on the target total spattering amount; Based on the anti-erosion capacity index, the aggregate anti-breakage capacity index, and the overall anti-loss capacity index, the target anti-erosion capacity corresponding to the target soil is obtained.

Images