Soil hardness prediction method and device, equipment, storage medium and program product

Abstract

The invention discloses a soil hardness prediction method and device, equipment, a storage medium and a program product, and relates to the technical field of image data processing, and the soil hardness prediction method comprises the steps: determining soil attribute data, predicting soil humidity time sequence data through the attribute data, carrying out the soil classification, and finally determining the soil penetration resistance. Comprehensive calculation of the soil hardness is achieved, meanwhile, the change trend in a period of time in the future can be reflected, and the timeliness and the refinement degree of drawing are remarkably improved. Meanwhile, widely available environmental data (such as satellite remote sensing, digital elevation models and meteorological reanalysis products) are used as input and can be quickly applied to areas lacking measured data, the data acquisition cost and time are greatly reduced, and the method has the capability of being popularized and applied to large-scale areas (such as provincial levels and drainage basins).

Description

Technical Field

[0001] This application relates to the field of image data processing technology, and in particular to methods, apparatus, equipment, storage media and program products for predicting soil hardness. Background Technology

[0002] Soil hardness is an important physical property that describes the soil’s ability to resist deformation or damage by external forces. It is also known as soil compaction or soil density and can be measured by soil penetration resistance (SPR).

[0003] Traditional soil hardness measurement mainly relies on in-situ penetration testing (IPT) techniques based on cone penetrometers. This involves inserting a probe equipped with a sensor into a specific depth in the soil and using the penetration resistance encountered when the probe is pressed into the soil at a constant speed to represent the soil hardness. However, this method has insufficient spatial coverage and cannot meet the needs for continuous spatiotemporal characterization of soil hardness in large-scale areas (such as provincial administrative regions or watersheds). Summary of the Invention

[0004] The main purpose of this application is to provide a method, device, equipment, storage medium and program product for predicting soil hardness, which aims to solve the technical problem that traditional soil hardness measurement is difficult to meet the needs of continuous spatiotemporal characterization of soil hardness in large-scale areas.

[0005] To achieve the above objectives, this application proposes a method for predicting soil hardness, the method comprising: Obtain environmental data of the area to be predicted, and determine the soil property data of the area to be predicted based on the environmental data; Based on the environmental data and the soil property data, the soil moisture of the area to be predicted is predicted to obtain the time series data of the soil moisture of the area to be predicted. Based on the soil attribute data, the soil of the area to be predicted is classified to obtain the soil classification data of the area to be predicted. Based on the soil property data, the soil moisture time series data, and the soil classification data, the soil penetration resistance of the area to be predicted is determined, and the soil penetration resistance is used to measure soil hardness.

[0006] In one embodiment, the step of determining the soil property data of the area to be predicted based on the environmental data includes: Based on the environmental data, the environmental covariates of the area to be predicted are determined; The environmental covariates are input into a pre-trained random forest regression model for prediction to obtain soil attribute data for the area to be predicted.

[0007] In one embodiment, before the step of inputting the environmental covariates into a pre-trained random forest regression model for prediction to obtain soil attribute data for the area to be predicted, the method further includes: Acquire training soil property data and training environmental covariate data; wherein the spatial scale of the training soil property data is a first scale, and the spatial scale of the training environmental covariate data is a second scale, and the first scale is larger than the second scale; Based on the training environment covariate data, downsampling processing is performed to obtain the training environment covariate data corresponding to the first scale; Based on the training soil attribute data and the training environment covariate data corresponding to the first scale, a random forest regression model is trained to obtain a pre-trained random forest regression model.

[0008] In one embodiment, the step of inputting the environmental covariates into a pre-trained random forest regression model for prediction to obtain soil attribute data for the area to be predicted includes: The environmental covariates are downsampled to obtain the environmental covariates corresponding to the first scale; The environmental covariates and the environmental covariates corresponding to the first scale are input into the pre-trained random forest regression model for prediction, so as to obtain soil attribute data at the first scale and soil attribute data at the second scale. Based on the soil environmental data at the first scale, residual correction is performed on the soil property data at the second scale to obtain the soil property data of the area to be predicted.

[0009] In one embodiment, before the step of predicting the soil moisture of the area to be predicted based on the environmental data and the soil property data to obtain the time-series data of the soil moisture of the area to be predicted, the method further includes: Acquire training environment data, which includes at least optical remote sensing data, first soil moisture data, and second soil moisture data; Based on the second soil moisture data, the first soil moisture data is corrected for deviation to obtain training soil moisture data; Image fusion is performed based on the optical remote sensing data to obtain the fused spectral index; A short-term prediction model is constructed based on the fused spectral index, the predicted time-series variable data, and the training soil moisture data to obtain a soil moisture prediction model. Accordingly, the step of predicting the soil moisture of the area to be predicted based on the environmental data and the soil property data to obtain the time-series data of the soil moisture of the area to be predicted includes: The environmental data and the soil attribute data are input into the soil moisture prediction model to predict soil moisture, thereby obtaining the time series data of soil moisture in the area to be predicted.

[0010] In one embodiment, before the step of classifying the soil in the area to be predicted based on the soil property data to obtain soil classification data for the area to be predicted, the method further includes: Obtain the first training database and the second training database, and determine the training soil attribute data common to the first training database and the second training database; The training soil attribute data is cleaned based on structured classification rules to obtain cleaned training soil attribute data. A classification model is trained based on the cleaned training soil property data to obtain a soil classification model. Accordingly, the step of classifying the soil in the area to be predicted based on the soil attribute data to obtain soil classification data for the area to be predicted includes: The soil attribute data is input into the soil classification model to classify the soil, thereby obtaining the soil classification data for the area to be predicted.

[0011] Furthermore, to achieve the above objectives, this application also proposes a soil hardness prediction device, which includes: The soil property determination module is used to acquire environmental data of the area to be predicted and determine the soil property data of the area to be predicted based on the environmental data. The soil moisture prediction module is used to predict the soil moisture of the area to be predicted based on the environmental data and the soil attribute data, and to obtain the time series data of the soil moisture of the area to be predicted. A soil classification model is used to classify the soil in the area to be predicted based on the soil attribute data, so as to obtain soil classification data of the area to be predicted. The soil hardness determination module is used to determine the soil penetration resistance of the area to be predicted based on the soil property data, the soil moisture time series data, and the soil classification data. The soil penetration resistance is used to measure soil hardness.

[0012] In addition, to achieve the above objectives, this application also proposes a soil hardness prediction device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the soil hardness prediction method as described above.

[0013] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the soil hardness prediction method described above.

[0014] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the soil hardness prediction method described above.

[0015] One or more technical solutions proposed in this application have at least the following technical effects: This application acquires environmental data of the area to be predicted and determines soil attribute data based on this data. It then predicts soil moisture in the area based on the environmental and soil attribute data, obtaining time-series soil moisture data. Based on the soil attribute data, it classifies the soil in the area, obtaining soil classification data. Finally, it determines soil penetration resistance, which is used to measure soil hardness, based on the soil attribute data, soil moisture time-series data, and soil classification data. By determining soil attribute data, predicting soil moisture time-series data, and classifying soil, soil penetration resistance is ultimately determined, achieving a comprehensive calculation of soil hardness. This also reflects future trends, significantly improving the timeliness and accuracy of mapping. Furthermore, utilizing widely available environmental data (such as satellite remote sensing, digital elevation models, and meteorological reanalysis products) as input allows for rapid application in areas lacking measured data, significantly reducing data acquisition costs and time, and possessing the capability for widespread application across large areas (such as provincial and watershed levels). Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating an embodiment of the soil hardness prediction method of this application. Figure 2 This is a flowchart illustrating Embodiment 2 of the soil hardness prediction method of this application; Figure 3 This is a roadmap of spatial downscaling technology in one implementation of an embodiment of this application; Figure 4 This is a flowchart illustrating Embodiment 3 of the soil hardness prediction method of this application; Figure 5 This is a technical roadmap for short-term soil moisture prediction in one implementation of the soil hardness prediction method of this application. Figure 6 This is a schematic diagram of the module structure of the soil hardness prediction device according to an embodiment of this application; Figure 7 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the soil hardness prediction method in the embodiments of this application.

[0019] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0021] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0022] The main solution of this application embodiment is as follows: acquire environmental data of the area to be predicted, and determine the soil property data of the area to be predicted based on the environmental data; predict the soil moisture of the area to be predicted based on the environmental data and the soil property data to obtain the soil moisture time series data of the area to be predicted; classify the soil of the area to be predicted based on the soil property data to obtain the soil classification data of the area to be predicted; determine the soil penetration resistance of the area to be predicted based on the soil property data, the predicted soil moisture data, and the soil classification data, and the soil penetration resistance is used to measure soil hardness.

[0023] This application provides a solution, proposing a high-precision prediction method for soil hardness based on regional scale, with the optimization of input data (soil property data, soil classification data, and soil moisture time series data) of transfer function model as the core. By improving the spatial resolution of basic soil properties, enhancing the timeliness and multi-soil layer expression ability of soil moisture data, and introducing engineering soil classification information to constrain the model, the method improves the accuracy, applicability, and engineering application value of soil hardness spatial distribution assessment.

[0024] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a computer or server, or an electronic device or virtual device capable of performing the above functions. The following description uses a soil hardness prediction device (hereinafter referred to as the prediction device) as an example to illustrate this embodiment and the subsequent embodiments.

[0025] Based on this, the embodiments of this application provide a method for predicting soil hardness, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the soil hardness prediction method of this application.

[0026] In this embodiment, the soil hardness prediction method includes steps S10 to S40: Step S10: Obtain environmental data of the area to be predicted, and determine the soil attribute data of the area to be predicted based on the environmental data; Step S20: Based on the environmental data and the soil attribute data, predict the soil moisture of the area to be predicted to obtain the time series data of the soil moisture of the area to be predicted.

[0027] It is understood that the aforementioned environmental data may be surface environmental factors related to soil formation or spatial distribution that can be obtained through remote sensing technology or geographic information technology, such as high-resolution remote sensing data, topographic data, meteorological data, spatial location data, coarse-resolution soil attribute data, etc., and the embodiments of this application do not limit them.

[0028] It should be understood that the aforementioned soil property data is data that can characterize the physical or chemical properties of the soil itself. The corresponding soil property data may differ for soils at different depths. The soil properties in this application embodiment may include, but are not limited to, organic carbon content, cation exchange capacity, soil bulk density, and gravel content. These soil property data can be represented in raster form or in other forms. In this application embodiment, higher resolution soil property data can be obtained by processing environmental data.

[0029] For example, in some embodiments of this application, the soil properties may be 24, specifically covering four soil depths: ALR1, ALR2, soil organic carbon, soil bulk density, cation exchange capacity, and gravel content, for soil layers of 0-5, 5-15, 15-30, and 30-60 cm.

[0030] It should be noted that the embodiments of this application do not limit the method of determining the above-mentioned soil attribute data. It can be obtained by downscaling existing coarse-resolution soil attribute data, or it can be determined by machine learning, neural networks or other methods.

[0031] Additionally, it should be noted that the aforementioned soil moisture time-series data can be a type of continuous spatiotemporal data used to describe soil moisture over a future period. For example, soil moisture data for the next three consecutive days (t+1, t+2, t+3); or soil moisture data for the next five consecutive days. The determination of the aforementioned soil moisture time-series data can be based on machine learning models, optical index methods, remote sensing inversion, etc., and this application does not impose any limitations on this method.

[0032] In its specific implementation, this application embodiment acquires environmental data of the area to be predicted and determines soil attribute data of the area based on the environmental data. Soil moisture in the area to be predicted is then predicted based on the environmental data and the soil attribute data to obtain time-series soil moisture data. By downscaling the environmental data, the coarse-resolution soil attributes are upgraded to higher-resolution soil attributes, filling the gap in high-resolution soil attribute data. By determining the time-series soil moisture data of the area to be predicted, soil moisture prediction for the area to be predicted is achieved, laying a crucial data foundation for subsequent refined prediction of soil hardness.

[0033] Step S30: Based on the soil attribute data, classify the soil in the area to be predicted to obtain soil classification data for the area to be predicted. Step S40: Based on the soil property data, the soil moisture time series data, and the soil classification data, determine the soil penetration resistance of the area to be predicted. The soil penetration resistance is used to measure soil hardness.

[0034] It is understandable that soils can be classified into different categories based on certain standards (such as small particle size, organic matter content, etc.), and these different categories can be identified by different symbols, names, or identification numbers. The soil grouping data mentioned above is used to represent the soil categories corresponding to different locations in the area to be predicted. This application does not limit the specific method of soil classification; it can be selected according to the needs of actual applications, such as using the Unified Soil Classification System (USCS) for soil classification.

[0035] For example, soil can be classified into coarse-grained soil (such as gravel, sand, etc.) and fine-grained soil (such as silt, clay, etc.) based on its particle size.

[0036] For example, based on the organic matter content of the soil, it can be divided into high organic matter soil, low organic matter soil, etc.

[0037] In some embodiments of this application, in order to determine the soil penetration resistance of the area to be predicted, the data can be processed by a soil penetration resistance transfer function model.

[0038] It should be noted that the soil penetration resistance transfer function model in this application embodiment can be an expression for the relationship between soil moisture and soil hardness established in the form of an exponential function. In this application embodiment, since the area to be predicted may cover various surface covers and soil uses, the soil penetration resistance transfer function model used needs to have high versatility and generalization ability.

[0039] For example, the soil penetration resistance transfer function model used in the embodiments of this application can be the second version of the exponential-logarithmic composite function prediction model of soil moisture-hardness (SMSP II), and its formula can be as follows: ; in, The cone index (CPI) can be used to measure soil penetration resistance. ) indicates (psi); The soil moisture content (%) can be determined based on soil property data and soil moisture time series data; The coefficients corresponding to different USCS soil types are undetermined and can be determined based on soil type data.

[0040] This application embodiment acquires environmental data of the area to be predicted and determines soil attribute data based on the environmental data. Soil moisture in the area is predicted based on the environmental data and soil attribute data to obtain time-series soil moisture data. Soil classification is performed based on the soil attribute data to obtain soil classification data. Soil penetration resistance is determined based on the soil attribute data, soil moisture time-series data, and soil classification data; soil penetration resistance is used to measure soil hardness. By determining soil attribute data, predicting soil moisture time-series data using the attribute data, and performing soil classification, soil penetration resistance is finally determined, achieving a comprehensive calculation of soil hardness. It also reflects the changing trend over a future period, significantly improving the timeliness and accuracy of mapping. Furthermore, using widely available environmental data (such as satellite remote sensing, digital elevation models, and meteorological reanalysis products) as input allows for rapid application in areas lacking measured data, significantly reducing data acquisition costs and time, and possessing the capability for widespread application across large areas (such as provincial and watershed levels).

[0041] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2, Figure 2 This is a flowchart illustrating Embodiment 2 of the soil hardness prediction method of this application.

[0042] like Figure 2 As shown in this embodiment, the step of determining the soil attribute data of the area to be predicted based on the environmental data includes: Step S11: Determine the environmental covariates of the area to be predicted based on the environmental data; Step S12: Input the environmental covariates into a pre-trained random forest regression model for prediction to obtain soil attribute data for the area to be predicted.

[0043] It should be noted that the aforementioned environmental covariates can be external environmental factors related to soil formation and spatial distribution, acquired using remote sensing or geographic information technology, and used to predict soil properties. These include spatial location variables, biometeorological variables, topographic variables based on digital elevation models, and remote sensing variables based on spaceborne optical sensors. By analyzing these environmental covariates, the soil properties of the entire area to be predicted can be determined.

[0044] In practical applications, due to insufficient investment in ground surveys and limitations imposed by conflicts in the spatiotemporal resolution of satellite remote sensing, soil information related to soil hardness is often used as input to prediction models with low temporal or spatial resolution. This leads to a decline in the availability of regionalized products and fails to meet the soil data requirements of applications such as precision agriculture and military decision-making. For example, soil data products supporting small-scale agricultural production and data used for simplified evaluation of equipment mobility performance in pre-planning scenarios should have a spatial resolution of no less than 30 m. However, in practical applications, high-quality basic soil attribute data (such as soil texture and bulk density) with a spatial resolution (i.e., spatial scale) of no less than 30 m are often lacking. Furthermore, the difficulty of soil sampling within a region, insufficient representative soil profile data, and uneven spatial distribution of samples make it difficult for traditional digital soil mapping methods to directly construct high-resolution regional soil attribute datasets. This application, by constructing an environmental covariate system, enables the generation of high-resolution low-scale environmental data (such as soil attribute data with a spatial scale of 30 m) from low-resolution high-scale environmental data (such as soil attribute data with a spatial scale of 250 m). Specifically, before the step of inputting the environmental covariates into a pre-trained random forest regression model for prediction to obtain soil attribute data of the area to be predicted, the method further includes: acquiring training soil attribute data and training environmental covariate data; wherein the spatial scale of the training soil attribute data is a first scale, and the spatial scale of the training environmental covariate data is a second scale, and the first scale is larger than the second scale; performing downsampling processing on the training environmental covariate data to obtain training environmental covariate data corresponding to the first scale; and training the random forest regression model based on the training soil attribute data and the training environmental covariate data corresponding to the first scale to obtain a pre-trained random forest regression model.

[0045] In this embodiment of the application, the process of determining soil attribute data in this application is illustrated by taking the effective downscaling of soil attribute data at a spatial scale of 250m (first scale) to soil attribute data at a spatial scale of 30m (second scale).

[0046] It should be noted that, in order to effectively downscale the current 250 m spatial scale soil property data to a 30 m spatial scale, a comprehensive and high-quality covariate system can be constructed. In this embodiment, using the Scorpan theoretical model as a guiding framework, training environmental covariate data with a 30 m spatial scale was constructed. This training environmental covariate data can cover the causal factors of soil spatial distribution, including surface information acquired by remote sensing, topographic factors derived from digital elevation models, regional climate characteristic indicators, and spatial location information, maximizing the explanatory power of environmental variables on soil heterogeneity. The final constructed covariates may include a total of 50 items.

[0047] For example, remote sensing variables based on spaceborne optical sensors: targeting the spectral absorption characteristics of chlorophyll, Landsat near-infrared and red bands are used to provide surface vegetation information, and information enhancement is achieved by calculating different vegetation indices, including normalized vegetation index, enhanced vegetation index, and soil-adjusted vegetation index. This improves the sensitivity and adaptability of images to vegetation status under complex surface backgrounds and climatic conditions, enhancing expression and differential extraction. Embodiments of this application can introduce short-wave infrared reflectance and its combination to form bare soil index and clay index, utilizing the difference in reflectance between visible light and short-wave infrared bands to enhance the physical characteristics of surface soil, indirectly characterizing parent material characteristics and geological background. First, derived spectral index variables are calculated using Landsat 5-8 reflectance data. Then, the median value of each pixel over a certain time dimension (e.g., 30 years) is calculated. Finally, the median image of each variable is obtained to characterize the long-term steady-state level of the pixel within this time interval.

[0048] For example, topographic variables based on digital elevation models: In complex geomorphological study areas, topography is a key factor influencing soil formation and spatial distribution. This application's embodiments utilize a 30-meter spatial scale digital elevation model and derive various topographic variables (e.g., implemented using the built-in algorithms of SAGA GIS 8.5.1 software), including basic slope, aspect, and hill shadow, as well as complex topographic structure indicators such as topographic humidity index, Melton ruggedness index, and topographic location index. These variables effectively reflect the impact of solar radiation, erosion and deposition processes, and micro-topographic changes on soil formation.

[0049] For example, biometeorological variables: Based on the biometeorological variable dataset, several initial biometeorological variables (e.g., 19) are obtained, covering key climate dimensions such as annual average temperature, seasonal variation, extreme precipitation, and temperature and humidity distribution, to effectively assist the model in capturing the climate response characteristics of soil environmental changes. Since biometeorological variables often exhibit significant multicollinearity, highly collinear terms need to be screened out from the 19 initial variables using the Pearson correlation coefficient.

[0050] For example, spatial location variables: This application introduces latitude and longitude coordinates and an Euclidean distance field constructed based on the image corners and center positions to characterize the spatial position of each pixel in the fine image relative to the overall structure, thereby effectively reducing the edge artifact problem commonly encountered in spatial downscaling. Furthermore, considering that water bodies and artificial surfaces were not masked in the original data, spatial location variables can also help the model make more robust judgments when identifying invalid pixel regions.

[0051] In some embodiments of this application, the data can be downsampled to a spatial scale of 250m using bilinear or nearest neighbor methods to obtain training environment covariate data corresponding to the first scale. Spatial location variables at 250m can be generated from these downsampled images. This training environment covariate data corresponding to the first scale can be used as predictor variables for model construction based on the training soil property data corresponding to the first scale. Figure 3 As shown, Figure 3 This is a roadmap of spatial downscaling technology in one implementation of an embodiment of this application. For example... Figure 3 As shown, by inputting low-resolution soil data (soil attribute data) and corresponding covariate data as training data for model construction, a pre-trained random forest regression model can be obtained. In practical applications, by inputting high-resolution covariates into the pre-trained random forest regression model, high-resolution soil attribute data and low-resolution soil attribute data can be predicted. The low-resolution soil attribute data is then used to perform residual correction on the high-resolution soil attribute data using a residual correction module to obtain the corrected high-resolution soil attribute data. Specifically, the step of inputting the environmental covariates into the pre-trained random forest regression model for prediction to obtain the soil attribute data of the area to be predicted includes: downsampling the environmental covariates to obtain the environmental covariates corresponding to the first scale; inputting the environmental covariates and the environmental covariates corresponding to the first scale into the pre-trained random forest regression model for prediction to obtain soil attribute data at the first scale and soil attribute data at the second scale; and performing residual correction on the second-scale soil attribute data based on the soil environmental data at the first scale to obtain the soil attribute data of the area to be predicted.

[0052] For example, in this embodiment of the application, a low-resolution residual can be obtained by subtracting the original low-resolution soil attribute data (i.e., the original soil attribute data corresponding to the first scale) from the predicted low-resolution soil attribute data, and a high-resolution residual can be obtained by performing spline interpolation based on the low-resolution residual; by adding the high-resolution residual to the predicted high-resolution soil data, the soil attribute data of the area to be predicted after residual correction can be obtained. That is, the step of performing residual correction on the soil attribute data of the second scale based on the soil environmental data of the first scale to obtain the soil attribute data of the area to be predicted includes: obtaining the original soil attribute data corresponding to the first scale; performing a subtraction operation on the original soil attribute data and the soil attribute data of the first scale to obtain a low-resolution residual; performing spline interpolation on the low-resolution residual to obtain a high-resolution residual; and adding the high-resolution residual to the soil attribute data of the second scale to obtain the soil attribute data of the area to be predicted.

[0053] In some embodiments of this application, examples of constructing 24 random forest regression models can be shown below: ; in, It is raw soil property data with a spatial resolution of 250 m. , , , The data represents the 250 m spatial resolution covariate data obtained after downsampling.

[0054] The fitted random forest regression model, based on covariate data at a spatial resolution of 30 m, can predict soil properties at this resolution. ; in, It is the predicted soil property result at a resolution of 30 m (i.e., soil property data of the area to be predicted). , , , It has a spatial resolution of 30 m. Scorpan Raw data for covariates (i.e., environmental covariates).

[0055] This application embodiment determines environmental covariates of the area to be predicted based on environmental data; the environmental covariates are then input into a pre-trained random forest regression model for prediction to obtain soil attribute data for the area to be predicted. Because a random forest regression model is constructed, second-scale environmental data is determined based on second-scale environmental covariates and first-scale environmental data, thus avoiding the insufficiency of second-scale environmental data.

[0056] Based on the first and / or second embodiments of this application, in the third embodiment of this application, the content that is the same as or similar to the first and / or second embodiments described above can be referred to the above description and will not be repeated hereafter. Based on this, please refer to... Figure 4 , Figure 4 This is a flowchart illustrating Embodiment 3 of the soil hardness prediction method of this application.

[0057] like Figure 4 As shown in the embodiment of this application, before the step of predicting the soil moisture of the area to be predicted based on the environmental data and the soil attribute data to obtain the time series data of the soil moisture of the area to be predicted, the method further includes: Step A100: Obtain training environment data, which includes at least optical remote sensing data, first soil moisture data, and second soil moisture data; Step A200: Based on the second soil moisture data, the first soil moisture data is corrected for deviation to obtain training soil moisture data; Step A300: Perform image fusion based on the optical remote sensing data to obtain the fused spectral index; Step A400: Construct a short-term prediction model based on the fused spectral index, the predicted time-series variable data, and the training soil moisture data to obtain a soil moisture prediction model; Accordingly, the step of predicting the soil moisture of the area to be predicted based on the environmental data and the soil property data to obtain the time-series data of the soil moisture of the area to be predicted includes: The environmental data and the soil attribute data are input into the soil moisture prediction model to predict soil moisture, thereby obtaining the time series data of soil moisture in the area to be predicted.

[0058] It should be noted that, in order to achieve the prediction of soil moisture time-series data, this application proposes an integrated "downscaling-prediction" framework for short-term soil moisture prediction, which can be specifically described as follows: Figure 5 As shown, Figure 5 This is a technical roadmap for short-term soil moisture prediction in one implementation of the soil hardness prediction method of this application. First, remote sensing data from different sources (such as Landsat and MODIS remote sensing data, which can be fused using GEE image fusion) can be fused to generate a spatiotemporally continuous fused spectral index. This index is then combined with meteorological variables, topographic variables (static), soil variables (static), and annual cumulative day data at a spatial resolution of 30m to determine the predicted time-series variables at a spatial resolution of 30m. Second, soil moisture data of different dimensions can be acquired, namely, first soil moisture data (such as soil moisture data at a spatial resolution of 1km in the SMCI library, which can be SMCI1.0, including data at different soil depths such as 10cm, 20cm, and 50cm) and second soil moisture data (such as soil moisture data at 20° in the GLDAS library, which can be GLDASv2.1, including data at soil depths of 5-20cm, 20-40cm, and 40-100cm).

[0059] In some embodiments of this application, the first soil moisture data can be globally biased and corrected using the second soil moisture data as a reference. Spatial downscaling modeling can then be performed using the upsampled data of the predicted time-series variable data (e.g., spatial downscaling modeling via the XGBoost algorithm with extreme gradient boosting) to obtain soil moisture data with a spatial resolution of 30m (e.g., SMCI1.0 downscaling data). Finally, in a pixel region excluding soil moisture monitoring stations and surrounding pixels, extracting 30m spatial resolution soil moisture data from some pixel locations and combining it with the predicted time-series variable data allows for iterating the soil moisture prediction model obtained from the XGBoost short-term prediction model. This generates the short-term predicted spatial distribution of soil moisture for the next three days at a specific time phase, i.e., soil moisture time-series data. The accuracy of the generated soil moisture prediction model can be evaluated using pixel soil moisture values ​​and ground-observed soil moisture values ​​(which can be determined based on SDR-SMN ground-observed data).

[0060] In some embodiments of this application, the GEE image fusion algorithm is used to fuse remote sensing data from different databases and generate a high spatiotemporal resolution spectral reflectance image to determine the fused spectral index. This algorithm is an improved version of the ESTARFM algorithm (with the same principle) and can run on the GEE cloud computing platform. The fused spatiotemporally continuous high-resolution image can be directly downloaded to the local disk for model building.

[0061] In some embodiments of this application, SMCI1.0 generally overestimates soil moisture. This application can use GLDASv2.1 second soil moisture data, which is unbiased overall but has poor correlation, for global correction. The Noah model of GLDASv2.1 divides soil depth into segments such as 0-5 cm, 5-20 cm, 20-40 cm, and 40-100 cm. To ensure that the soil moisture data of the two datasets correspond in vertical distribution, this application proposes to use GLDASv2.1 data from 5–20 cm to correct 10 cm SMCI1.0 data, GLDASv2.1 data from 20–40 cm to correct 20 cm SMCI1.0 data, and GLDASv2.1 data from 40–100 cm to correct 50 cm SMCI1.0 data. The bias correction process proposed in this application is as follows: First, the SMCI1.0 soil moisture data with a spatial resolution of 1 km is spatially aggregated by calculating the mean of all 1 km pixels in each 0.25° pixel to generate data with the same 0.25° resolution as GLDASv2.1. Since the bias is a comprehensive result of long-term series in a specific region, this invention defines the correction value of each 0.25° pixel as the difference between the means of the two datasets in the same time series.

[0062]

[0063] in, This represents the deviation value of a single pixel at a resolution of 0.25°. and The values ​​are the time series averages of the SMCI1.0 spatially aggregated soil moisture data and the GLDASv2.1 data for this pixel, respectively. Then, the deviation at 0.25° resolution is interpolated to 1 km resolution using cubic spline spatial interpolation, and the SMCI1.0 data is corrected.

[0064]

[0065]

[0066] in, The deviation at 1 km resolution obtained by interpolation, This is the corrected SMCI1.0 soil moisture. This method preserves the high correlation and unbiased root mean square error based on ERA-5 Land data while reducing bias. Compared to the daily correction method, this method reduces the possibility that the underestimation bias inherent in GLDASv2.1 on a small number of dates could further increase the total bias for those dates.

[0067] It should be noted that in the XGBoost spatial downscaling modeling stage, topographic data (elevation, slope, hill shadow), meteorological data (rainfall, temperature, wind speed), and soil data at different depths (clay content, silt content, sand content, gravel content, soil bulk density, residual water content, saturated water content, saturated hydraulic conductivity) at a spatial resolution of 30 m were first downsampled to a resolution of 1 km using bilinear methods. Because the NDVI and NSDSI databases, fused from remote sensing data from different sources, are affected by cloud cover and low-similarity pixels, the images contain many missing values ​​for some dates, making mathematical interpolation impossible. Therefore, the nearest neighbor method was used for downsampling. Environmental auxiliary variables after sampling were used as predictors of soil moisture at 1 km. In the downscaling stage, meteorological variables only used the target time phase and the meteorological variables of the past 5 days for that time phase as model input, introducing the lag effect of soil moisture into the model and improving the accuracy of the downscaling model. The target variable of the downscaling model is the soil moisture data at the next 1 km resolution at depths of 10 cm, 20 cm, and 50 cm in the corrected SMCI1.0 dataset, which are used to characterize soil moisture at depths of 5–15 cm, 15–30 cm, and 30–60 cm, respectively.

[0068]

[0069] in, It is the soil moisture content of SMCI1.0 at a certain soil depth. , , These are the topographic, meteorological, and soil auxiliary variables used in the study. During the specific model construction process, soil properties below or covering the target depth were selected as input variables to provide complete soil hydrological driving information. For example, when downscaling soil moisture at a depth of 10 cm, soil property data from the 0–5 cm and 5–15 cm soil layers were used. After spatial matching of the predictors and the target variables, all 1 km sample cells with missing values ​​in NDVI or NSDSI were excluded from the modeling process.

[0070] In some embodiments of this application, the Extreme Gradient Boosting (XGBoost) algorithm can be used to construct soil moisture downscaling models at three soil depths of 10 cm, 20 cm, and 50 cm. Finally, the trained models are applied to auxiliary variables with a spatial resolution of 30 m to predict soil moisture time-series data with a spatial resolution of 30 m.

[0071] In some embodiments of this application, in order to implement soil data components, before the step of classifying the soil in the area to be predicted based on the soil attribute data to obtain the soil classification data of the area to be predicted, the method further includes: acquiring a first training database and a second training database, and determining the training soil attribute data common to the first training database and the second training data; cleaning the training soil attribute data based on structured classification rules to obtain cleaned training soil attribute data; training a classification model based on the cleaned training soil attribute data to obtain a soil classification model; correspondingly, the step of classifying the soil in the area to be predicted based on the soil attribute data to obtain the soil classification data of the area to be predicted includes: inputting the soil attribute data into the soil classification model to perform soil classification to obtain the soil classification data of the area to be predicted.

[0072] It should be noted that this application first addresses the discrepancy between soil particle size classification standards in engineering and agricultural soil classification by proposing a reasonable method for converting engineering soil types. Secondly, it extracts soil physicochemical variables closely related to engineering soil classification from both a soil unit gridded database (first training database, such as the gSSURGO database) and a soil information database (second training database, such as the ChinaSoilInfoGrids database), i.e., training soil attribute data. The training soil attribute data can be cleaned using structured classification rules to obtain cleaned training soil attribute data. Then, based on the cleaned training soil attribute data and the corresponding USCS engineering soil classification, an XGBoost model suitable for structured data is constructed. By transferring learning from this XGBoost model of structured data, a soil classification model can be obtained; the transfer learning process is not limited in this embodiment. Using the soil classification model to downscale the spatially obtained soil attribute data, an engineering soil classification map covering multiple vertical soil depths can be generated, i.e., soil classification data.

[0073] This application embodiment acquires training environment data, which includes at least optical remote sensing data, first soil moisture data, and second soil moisture data. Based on the second soil moisture data, the first soil moisture data is corrected for bias to obtain training soil moisture data. Image fusion is performed based on the optical remote sensing data to obtain a fused spectral index. A short-term prediction model is constructed based on the fused spectral index, prediction time-series variable data, and the training soil moisture data to obtain a soil moisture prediction model. By correcting the bias of the first soil moisture data, the negative impact of the generally overestimated first soil moisture data on model construction is avoided. By fusing different remote sensing data, spectral index image data with spatiotemporal continuity is generated. Through multi-source data collaboration, a soil moisture model that can output high resolution and predict the future is achieved.

[0074] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the soil hardness prediction method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0075] This application also provides a soil hardness prediction device, please refer to... Figure 6 , Figure 6 This is a schematic diagram of the module structure of the soil hardness prediction device according to an embodiment of this application. The soil hardness prediction device includes: Soil property determination module 10 is used to acquire environmental data of the area to be predicted and determine the soil property data of the area to be predicted based on the environmental data. Soil moisture prediction module 20 is used to predict the soil moisture of the area to be predicted based on the environmental data and the soil attribute data, and obtain the time series data of the soil moisture of the area to be predicted. Soil classification model 30 is used to classify the soil in the area to be predicted based on the soil attribute data, and to obtain soil classification data of the area to be predicted. The soil hardness determination module 40 is used to determine the soil penetration resistance of the area to be predicted based on the soil property data, the soil moisture time series data and the soil classification data. The soil penetration resistance is used to measure soil hardness.

[0076] The soil hardness prediction device provided in this application, employing the soil hardness prediction method described in the above embodiments, can solve the technical problem that traditional soil hardness measurements are insufficient to meet the requirements for continuous spatiotemporal characterization of soil hardness in large-scale areas. Compared with the prior art, the beneficial effects of the soil hardness prediction device provided in this application are the same as those of the soil hardness prediction method provided in the above embodiments, and other technical features of the soil hardness prediction device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0077] This application provides a soil hardness prediction device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the soil hardness prediction method in the first embodiment described above.

[0078] The following is for reference. Figure 7 The diagram illustrates a structural schematic suitable for implementing the soil hardness prediction device in the embodiments of this application. The soil hardness prediction device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), vehicle terminals (e.g., vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 7 The soil hardness prediction device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of this application.

[0079] like Figure 7As shown, the soil hardness prediction device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the soil hardness prediction device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. The communication device 1009 allows the soil hardness prediction device to communicate wirelessly or wiredly with other devices to exchange data. Although the figures show soil hardness prediction devices with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0080] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0081] The soil hardness prediction device provided in this application, employing the soil hardness prediction method described in the above embodiments, can solve the technical problem that traditional soil hardness measurements are insufficient to meet the requirements for continuous spatiotemporal characterization of soil hardness in large-scale areas. Compared with the prior art, the beneficial effects of the soil hardness prediction device provided in this application are the same as those of the soil hardness prediction method provided in the above embodiments, and other technical features of this soil hardness prediction device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0082] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0083] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0084] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the soil hardness prediction method in the above embodiments.

[0085] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), or flash memory, optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0086] The aforementioned computer-readable storage medium may be included in the soil hardness prediction device; or it may exist independently and not assembled into the soil hardness prediction device.

[0087] The aforementioned computer-readable storage medium carries one or more programs that, when executed by the soil hardness prediction device, cause the soil hardness prediction device to: Obtain environmental data of the area to be predicted, and determine the soil property data of the area to be predicted based on the environmental data; Based on the environmental data and the soil property data, the soil moisture of the area to be predicted is predicted to obtain the time series data of the soil moisture of the area to be predicted. Based on the soil attribute data, the soil of the area to be predicted is classified to obtain the soil classification data of the area to be predicted. Based on the soil property data, the soil moisture time series data, and the soil classification data, the soil penetration resistance of the area to be predicted is determined, and the soil penetration resistance is used to measure soil hardness.

[0088] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0089] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0090] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0091] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described soil hardness prediction method. This solves the technical problem that traditional soil hardness measurements are insufficient to meet the requirements for continuous spatiotemporal characterization of soil hardness in large-scale areas. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the soil hardness prediction method provided in the above embodiments, and will not be elaborated upon here.

[0092] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the soil hardness prediction method described above.

[0093] The computer program product provided in this application can solve the technical problem that traditional soil hardness measurement methods are insufficient to meet the requirements for continuous spatiotemporal characterization of soil hardness in large-scale areas. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the soil hardness prediction method provided in the above embodiments, and will not be repeated here.

[0094] The above description is only a part of the embodiments of this application and does not limit the scope of protection of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included in the scope of protection of this application.

Claims

1. A method for predicting soil hardness, characterized in that, The method includes: Obtain environmental data of the area to be predicted, and determine the soil property data of the area to be predicted based on the environmental data; Based on the environmental data and the soil property data, the soil moisture of the area to be predicted is predicted to obtain the time series data of the soil moisture of the area to be predicted. Based on the soil attribute data, the soil of the area to be predicted is classified to obtain the soil classification data of the area to be predicted. Based on the soil property data, the soil moisture time series data, and the soil classification data, the soil penetration resistance of the area to be predicted is determined, and the soil penetration resistance is used to measure soil hardness.

2. The soil hardness prediction method as described in claim 1, characterized in that, The step of determining the soil property data of the area to be predicted based on the environmental data includes: Based on the environmental data, the environmental covariates of the area to be predicted are determined; The environmental covariates are input into a pre-trained random forest regression model for prediction to obtain soil attribute data for the area to be predicted.

3. The soil hardness prediction method as described in claim 2, characterized in that, Before the step of inputting the environmental covariates into a pre-trained random forest regression model for prediction to obtain the soil attribute data of the area to be predicted, the method further includes: Acquire training soil property data and training environmental covariate data; wherein the spatial scale of the training soil property data is a first scale, and the spatial scale of the training environmental covariate data is a second scale, and the first scale is larger than the second scale; Based on the training environment covariate data, downsampling processing is performed to obtain the training environment covariate data corresponding to the first scale; Based on the training soil attribute data and the training environment covariate data corresponding to the first scale, a random forest regression model is trained to obtain a pre-trained random forest regression model.

4. The soil hardness prediction method as described in claim 3, characterized in that, The steps of inputting the environmental covariates into a pre-trained random forest regression model for prediction to obtain soil attribute data for the area to be predicted include: The environmental covariates are downsampled to obtain the environmental covariates corresponding to the first scale; The environmental covariates and the environmental covariates corresponding to the first scale are input into the pre-trained random forest regression model for prediction, so as to obtain soil attribute data at the first scale and soil attribute data at the second scale. Based on the soil environmental data at the first scale, residual correction is performed on the soil property data at the second scale to obtain the soil property data of the area to be predicted.

5. The soil hardness prediction method as described in claim 1, characterized in that, Before the step of predicting the soil moisture of the area to be predicted based on the environmental data and the soil property data to obtain the time series data of the soil moisture of the area to be predicted, the method further includes: Acquire training environment data, which includes at least optical remote sensing data, first soil moisture data, and second soil moisture data; Based on the second soil moisture data, the first soil moisture data is corrected for deviation to obtain training soil moisture data; Image fusion is performed based on the optical remote sensing data to obtain the fused spectral index; A short-term prediction model is constructed based on the fused spectral index, the predicted time-series variable data, and the training soil moisture data to obtain a soil moisture prediction model. Accordingly, the step of predicting the soil moisture of the area to be predicted based on the environmental data and the soil property data to obtain the time-series data of the soil moisture of the area to be predicted includes: The environmental data and the soil attribute data are input into the soil moisture prediction model to predict soil moisture, thereby obtaining the time series data of soil moisture in the area to be predicted.

6. The soil hardness prediction method as described in claim 1, characterized in that, Before the step of classifying the soil in the area to be predicted based on the soil property data to obtain the soil classification data of the area to be predicted, the method further includes: Obtain the first training database and the second training database, and determine the training soil attribute data common to the first training database and the second training database; The training soil attribute data is cleaned based on structured classification rules to obtain cleaned training soil attribute data. A classification model is trained based on the cleaned training soil property data to obtain a soil classification model. Accordingly, the step of classifying the soil in the area to be predicted based on the soil attribute data to obtain soil classification data for the area to be predicted includes: The soil attribute data is input into the soil classification model to classify the soil, thereby obtaining the soil classification data for the area to be predicted.

7. A soil hardness prediction device, characterized in that, The soil hardness prediction device includes: The soil property determination module is used to acquire environmental data of the area to be predicted and determine the soil property data of the area to be predicted based on the environmental data. The soil moisture prediction module is used to predict the soil moisture of the area to be predicted based on the environmental data and the soil attribute data, and to obtain the time series data of the soil moisture of the area to be predicted. A soil classification model is used to classify the soil in the area to be predicted based on the soil attribute data, so as to obtain soil classification data of the area to be predicted. The soil hardness determination module is used to determine the soil penetration resistance of the area to be predicted based on the soil property data, the soil moisture time series data, and the soil classification data. The soil penetration resistance is used to measure soil hardness.

8. A soil hardness prediction device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the soil hardness prediction method as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the soil hardness prediction method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the soil hardness prediction method as described in any one of claims 1 to 6.

Images