Farmland non-point source pollution monitoring risk assessment and prevention and control method based on multiple scales

Abstract

The invention discloses a multi-scale-based farmland non-point source pollution monitoring risk assessment, prevention and control method, and the method comprises the following steps: setting farmland non-point source pollution monitoring points according to different monitoring purposes under the conditions of typical terrains and typical crops, and monitoring the farmland non-point source pollution of a field scale; the method comprises the following steps: installing water quality and water quantity automatic monitoring equipment at key nodes of a small watershed of a field, and monitoring watershed scale farmland non-point source pollution; the method comprises the following steps: collecting farmland non-point source pollution index data of a field size and a watershed size, carrying out online transmission and data storage on the index data, and constructing a farmland non-point source pollution online monitoring and early warning platform; acquiring farmland non-point source pollution related data, evaluating farmland non-point source pollution by using multiple models, and outputting farmland non-point source pollution load distribution characteristics, a key source area and a prevention and control strategy in a target area in combination with cold and hot spot analysis; a farmland non-point source pollution space-ground-air integrated comprehensive monitoring network is constructed by using the Internet of Things technology, and farmland non-point source pollution monitoring, evaluation and early warning are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of farmland non-point source pollution monitoring, and in particular, to a multi-scale farmland non-point source pollution monitoring risk assessment and prevention and control method. Background Art

[0002] Agricultural non-point source pollution is characterized by randomness, dispersion and widespread occurrence, and the temporal and spatial heterogeneity of pollutant output is strong, which brings great difficulties to the accurate monitoring of non-point source pollution. Field monitoring is an evaluation method to characterize the impact of agricultural non-point source pollutants on the water environment. However, since the occurrence of agricultural non-point source pollution involves relatively complex environmental processes and geophysical and chemical transformation processes of nutrients, it will also be affected by point source pollution as the spatial scale expands. Therefore, it is difficult to effectively remove the impact of these non-agricultural activities by simply using monitoring methods.

[0003] However, artificial rainfall simulation methods often require power output and large-scale transportation equipment, which is not easy to carry, has poor mobility, and is very inconvenient to assemble and load and unload. Indicators such as rainfall intensity need to be calibrated on site. The test operation is complicated, and the test results acquisition and post-processing cycle is long and the workload is large, which greatly restricts its use. In addition, traditional monitoring methods such as field monitoring and artificial rainfall simulation methods are limited to measuring the small-scale farmland non-point source pollutant load, and cannot measure the macro-scale farmland non-point source pollution status.

[0004] Existing agricultural non-point source pollution models are mostly used to calculate small-scale pollution loads and lack identification of key source areas. Since most data sources are based on monitoring of field-scale losses, insufficient consideration is given to the losses of pollutants during the generation and aggregation of runoff and their migration to the basin outlet. The influence of factors such as basin topography, hydrology and climate, vegetation cover and land use are not integrated. Therefore, the model calculation results are not the actual pollution load at the basin outlet, resulting in insufficient accuracy of the model in large-scale simulations. Summary of the invention

[0005] The purpose of the present invention is to overcome the problems of difficulty in monitoring farmland non-point source pollution, insufficient accuracy of macro-scale agricultural non-point source pollution assessment, lack of key source area identification and prevention and control strategies under the existing technology, and provide a multi-scale farmland non-point source pollution monitoring risk assessment and prevention and control method, to achieve multi-scale assessment of farmland non-point source pollution monitoring and assessment "field scale-watershed scale-macro scale", accurately output macro-scale farmland non-point source pollution load distribution characteristics, key source areas and prevention and control strategies, and realize real-time monitoring and early warning of farmland non-point source pollution, and the error of farmland non-point source pollution monitoring and assessment results is very small.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] A multi-scale farmland non-point source pollution monitoring risk assessment and prevention and control method includes the following steps:

[0008] Under typical terrain and typical crop conditions, farmland non-point source pollution monitoring points were set up according to different monitoring purposes to monitor farmland non-point source pollution at the field scale, and the LOADEST model was used to calculate the field scale nitrogen and phosphorus pollution fluxes.

[0009] Install automatic water quality and quantity monitoring equipment at key nodes of small watersheds in fields to monitor non-point source pollution of farmland at the watershed scale, use the InVEST model to assess non-point source pollution at the watershed scale, and calibrate the InVEST model with observation data;

[0010] Collect farmland non-point source pollution index data at the field scale and watershed scale, transmit and store the index data online, build an online monitoring and early warning platform for farmland non-point source pollution, and conduct online monitoring of farmland non-point source nitrogen and phosphorus loss;

[0011] Obtain relevant data on farmland non-point source pollution, use the InVEST model and machine learning model to evaluate farmland non-point source pollution, and combine hot and cold spot analysis to output the distribution characteristics of farmland non-point source pollution load, key source areas and prevention and control strategies in the target area;

[0012] By integrating multi-source heterogeneous data from field-scale monitoring, watershed-scale monitoring, and farmland non-point source pollution assessment, the Internet of Things technology is used to build an integrated monitoring network for farmland non-point source pollution that integrates land, sky, and air, to achieve monitoring, assessment, and early warning of farmland non-point source pollution.

[0013] Furthermore, field-scale monitoring of typical crops was carried out under typical terrain conditions, with typical crops representing four land use types: paddy fields, dry land, irrigated land, and gardens;

[0014] Establish standardized monitoring plots for nitrogen and phosphorus loss in farmland, with monitoring indicators including runoff and leaching water quantity and quality, microclimate, soil moisture, fertilizer application and irrigation water quantity;

[0015] The LOADEST model was used to calculate the nitrogen and phosphorus losses at the field scale, and the impact of soil physical and chemical properties, crop types, fertilization methods, fertilizer ratios, fertilizer amounts and rainfall agricultural environmental information on non-point source pollution was comprehensively analyzed.

[0016] Furthermore, the LOADEST model principle is as follows:

[0017] The flux of non-point source pollutants from farmland is the total amount of pollutants passing through a certain section within a certain period of time:

[0018]

[0019] In the formula, is the pollutant flux in the time period τ; Δt is the time interval; NP is the number of discrete time intervals; Q is the average flow rate during the time period; C is the instantaneous water quality; is the instantaneous pollutant flux;

[0020] The LOADEST model is based on equation (1) and uses the multivariate linear regression method to calculate the river pollutant flux. Make an estimate:

[0021]

[0022] In the formula, a0, a j is the coefficient of the equation; NV is the number of independent variables; X j is the independent variable;

[0023] The LOADEST model provides 11 pollutant flux regression equations, and uses the AIC criterion and SPPC criterion to calibrate and optimize the parameters. When the AIC and SPPC values ​​are minimized, it is the optimal pollutant flux regression equation.

[0024]

[0025] In the formula, For the data set D(X i , i=1,n) the maximum likelihood value; is the maximum likelihood estimate of the equation parameters; d is the number of equation parameters; n is the number of data sets used for estimating the equation parameters;

[0026] The LOADEST model establishes parameter estimation methods based on three statistical estimation methods: asymptotic maximum likelihood estimation AMLE, maximum likelihood estimation MLE, and least absolute deviation method LAD; when the residuals obey the normal distribution, AMLE is used to estimate the parameters of the censored data, as shown in formula (4); MLE is used to estimate the parameters of the non-censored data, as shown in formula (5); when the residuals do not obey the normal distribution, LAD is used to estimate the parameters, as shown in formula (6);

[0027]

[0028] In the formula, is the pollutant flux estimated by AMLE, MVUE, and LAD methods; a0, a j are the coefficients of the equation calculated by different methods; H(a, b, s 2 ,α,k) is the likelihood approximation function of infinite series; g m (m,s 2 ,V) is the Bessel function; a, b, V are the independent variable functions; α, h are the parameters of the gamma distribution; m is the degree of freedom; s 2 is the residual variance; ek is the residual error; f is the number of missing data in the data.

[0029] Furthermore, automatic water quality and quantity monitoring equipment is installed at key nodes of small watersheds in the fields to carry out in-situ monitoring of water quality and quantity at the watershed scale. The water quantity monitoring equipment includes Parshall flume, pressure water level sensor and data acquisition transmitter, and the water quality monitoring equipment includes automatic analyzer for total nitrogen, total phosphorus, ammonia nitrogen and chemical oxygen demand;

[0030] Online transmission and data storage of index data are realized through the pollutant online monitoring system, which includes the perception layer, network layer and application layer;

[0031] The water quality and quantity automatic monitoring equipment is the perception layer, which is installed at the pollutant monitoring site and the process nodes that affect pollutant emissions. It is used to monitor pollutant emissions and process parameters and complete communication with the host computer.

[0032] The data acquisition and transmission equipment is the network layer, which collects data from the automatic monitoring equipment for water quality and quantity, and completes the single-chip microcomputer, industrial computer, embedded computer, programmable automation controller or programmable controller for data storage and data transmission with the monitoring center;

[0033] The monitoring center is the application layer, which is installed in the competent departments at all levels. It is linked to the automatic monitoring equipment through the network and issues query and control instructions to it.

[0034] Furthermore, the InVEST model results were corrected using field-scale and watershed-scale observation data. The model output error was evaluated by adjusting the Borselli K parameter and calculating the goodness of fit under different Borselli K parameters. The statistic of goodness of fit was the determination coefficient R 2 , R 2 The maximum value is 1, R 2 The closer the value is to 1, the smaller the error of the model simulation value is than the observed value;

[0035]

[0036] In the formula, y 观测 is the observed value; is the average value of the observations; y 模拟 is the simulated value; is the average of the simulated values.

[0037] Furthermore, the InVEST model principle is as follows:

[0038] First, the nitrogen and phosphorus loss loads are distributed to the surface and subsurface according to the proportion, and the calculation formula is:

[0039] loadsurf,p =(1-proportion subsurface )×load_ βp

[0040] load sub,p =proportion subsurface ×load_ βp

[0041] Where β is the nutrient; p is a single pixel; load subsurf,p is the subsurface nutrient load;

[0042] The surface nutrient transfer rate is calculated as:

[0043]

[0044] Where, NDR 0,p is the nutrient transfer rate retained by downstream pixels; IC j is the terrain index; IC0 and k are calibration parameters;

[0045] The formula for calculating the subsurface nutrient transfer rate is:

[0046]

[0047] Where, NDR sub,p is the underground nutrient transfer rate; eff subs To achieve maximum nutrient retention efficiency in underground flow; subs is the interception length of groundwater flow; that is, the distance at which the soil is assumed to retain nutrients at its maximum capacity, and l is the distance from the pixel to the flow;

[0048] The formula for calculating nutrient output is:

[0049] U exp,p =load surf,p ×NDR surf,p +load subs,[ ×NDR subs,p

[0050]

[0051] Where U exp,p Load is the nutrient output load of each grid cell; surf,p is the surface nutrient load; NDR surf,p is the surface nutrient transfer rate; load subs,p is the subsurface nutrient load; NDR subs,p is the subsurface nutrient transfer rate; Export the total nutrient load for the sub-basin.

[0052] Furthermore, the monitoring and early warning platform includes a remote intelligent control platform, an early warning platform and a nitrogen and phosphorus non-point source loss monitoring platform, which can realize online monitoring and early warning of nitrogen and phosphorus loss from farmland non-point sources;

[0053] Based on the IoT hardware monitoring system of nitrogen and phosphorus non-point source pollution in farmland at the watershed scale, supplemented by field experiments to obtain basic data on nitrogen and phosphorus loss in different crops, InVEST parameter correction and optimization verification are carried out to predict and evaluate the water, nitrogen and phosphorus material cycles in regional farmland ecosystems, build a monitoring and early warning platform, realize remote online intelligent monitoring of non-point source pollution in farmland, and issue real-time early warnings for indicator data that exceed the standard. The constructed network data are mutually verified and correlated, and gradually radiate to other areas.

[0054] Furthermore, through nitrogen and phosphorus loss assessment, remote sensing, UAVs and meteorological stations were used to obtain farmland distribution, terrain slope and meteorological data in the basin. The NDR module of the InVEST model was used to calibrate and optimize the InVEST parameters in combination with the observation data at the field scale and the basin scale, and finally the output of nitrogen and phosphorus nutrients at the basin scale was simulated.

[0055] Furthermore, through regional farmland nonpoint source assessment, machine learning models were used and professional external survey software "External Survey Assistant" was introduced to obtain basic information on the target area's planting area, soil type, farming method, soil nutrients, fertilization method, land use method, terrain slope and historical meteorological data;

[0056] After collecting various types of agricultural non-point source pollution data, we used Geodatabase as the database based on the ArcGIS platform, and established agricultural data database, meteorological data database and geospatial database for agricultural non-point source pollution. We revised the InVEST model through the measured data at the field scale and watershed scale, and converted the watershed scale simulation into a macro-scale simulation through the machine learning model to establish the farmland non-point source model for the target area. Combined with ArcGIS hot and cold spot analysis, we output the distribution characteristics of farmland non-point source pollution load, key source areas and prevention and control strategies in the target area.

[0057] Furthermore, the principles of ArcGIS hot and cold spot analysis are as follows:

[0058] ArcGIS hot spot analysis is based on Getis-OrdG i * The index divides hot and cold spots into different types; Getis-OrdG i * The index is a statistical method for identifying clusters of high and low values. It is used in geographical research to identify and explain spatial distribution patterns. It considers highly significant hot spots at a 99% confidence level to be key pollution source areas.

[0059] The calculation formula is:

[0060]

[0061] In the formula, G i * is the Getis-Ord index; V j is the nitrogen and phosphorus emission intensity of the jth study unit; is the mean value of nitrogen emission intensity; W oj is the spatial weight coefficient of regions o and j; N total is the total number of research units.

[0062] Compared with the prior art, the present invention utilizes the "1234" mode of farmland non-point source pollution monitoring, that is, through 4 crops, 3 platforms and 2 sets of models, and uses the Internet of Things technology to construct a macro-regional farmland non-point source pollution sky-ground integrated monitoring network, breaking the limitations of traditional monitoring methods, and targeting the problems of difficulty in monitoring and evaluating farmland non-point source pollution, inability to accurately measure the macro-scale agricultural non-point source pollution status, lack of key source area identification and prevention and control strategies, to achieve a multi-scale assessment of farmland non-point source pollution monitoring and assessment of "field scale-watershed scale-macro scale", accurately output macro-scale farmland non-point source pollution load distribution characteristics, key source areas and prevention and control strategies, to achieve real-time monitoring and early warning of farmland non-point source pollution, and the error of farmland non-point source pollution monitoring and assessment results is very small. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a flowchart of the risk assessment and prevention and control methods for farmland non-point source pollution monitoring based on multiple scales.

[0064] Figure 2 This is a schematic diagram of the structure of the automatic monitoring equipment for water quality and quantity.

[0065] Figure 3 This is a schematic diagram of the structure of the pollutant online monitoring system.

[0066] Figure 4 Schematic diagram of the overall planning system of the Internet of Things.

[0067] Figure 5 Schematic diagram of the machine learning algorithm flow.

[0068] Figure 6 Schematic diagram of the machine learning model process. DETAILED DESCRIPTION

[0069] The multi-scale farmland non-point source pollution monitoring risk assessment and prevention and control method of the present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0070] See also Figure 1The present invention discloses a multi-scale farmland non-point source pollution monitoring risk assessment and prevention and control method, comprising the following steps:

[0071] Under typical terrain and typical crop conditions, farmland non-point source pollution monitoring points were set up according to different monitoring purposes to monitor farmland non-point source pollution at the field scale, and the LOADEST model was used to calculate the field scale nitrogen and phosphorus pollution fluxes.

[0072] Install automatic water quality and quantity monitoring equipment at key nodes of small watersheds in fields to monitor non-point source pollution of farmland at the watershed scale, use the InVEST model to assess non-point source pollution at the watershed scale, and calibrate the InVEST model with observation data;

[0073] Collect farmland non-point source pollution index data at the field scale and watershed scale, transmit and store the index data online, build an online monitoring and early warning platform for farmland non-point source pollution, and conduct online monitoring of farmland non-point source nitrogen and phosphorus loss;

[0074] Obtain relevant data on farmland non-point source pollution, use the InVEST model and machine learning model to evaluate farmland non-point source pollution, and combine hot and cold spot analysis to output the distribution characteristics of farmland non-point source pollution load, key source areas and prevention and control strategies in the target area;

[0075] By integrating multi-source heterogeneous data from field-scale monitoring, watershed-scale monitoring, and farmland non-point source pollution assessment, the Internet of Things technology is used to build an integrated monitoring network for farmland non-point source pollution that integrates land, sky, and air, to achieve monitoring, assessment, and early warning of farmland non-point source pollution.

[0076] Under the typical terrain conditions of dry slopes and flat land, typical crops represent four types of land use: paddy fields, dry land, irrigated land, and gardens. According to different monitoring purposes, standardized monitoring plots for nitrogen and phosphorus loss in farmland were established. Several treatments were set for each crop, and 3 replicates were set for each treatment. The plot monitoring test plan is shown in Table 1. The monitoring indicators include runoff and leaching water volume and quality, microclimate, soil moisture, fertilizer application amount, and irrigation water volume; the LOADEST model was used to calculate the field-scale nitrogen and phosphorus loss, and the impact of soil physical and chemical properties, crop types, fertilization methods, fertilizer ratios, fertilizer application amounts, and rainfall on agricultural environmental information on non-point source pollution was comprehensively analyzed.

[0077] Table 1 Cell monitoring test plan

[0078]

[0079] Under typical terrain, field monitoring experiments of typical crops were carried out to monitor farmland non-point source pollution at the field scale. Water quality and quantity index data of farmland non-point source pollution at the field scale were collected, and the farmland non-point source pollution data at the field scale were processed and analyzed using the LOADEST model to estimate the farmland non-point source pollutant flux at the field scale, providing observation data for correcting the simulation results of the INVEST model.

[0080] The LOADEST model works as follows:

[0081] The flux of non-point source pollutants from farmland is the total amount of pollutants passing through a certain section within a certain period of time:

[0082]

[0083] In the formula, is the pollutant flux in the time period τ; Δt is the time interval; NP is the number of discrete time intervals; Q is the average flow rate during the time period; C is the instantaneous water quality; is the instantaneous pollutant flux.

[0084] The LOADEST model is based on equation (1) and uses the multivariate linear regression method to calculate the river pollutant flux. Make an estimate:

[0085]

[0086] In the formula, a0, a j is the coefficient of the equation; NV is the number of independent variables; X j is the independent variable.

[0087] The LOADEST model provides 11 pollutant flux regression equations, as shown in Table 2, and uses the AIC criterion (Akaikeinformation criterion) and SPPC criterion (Schwarz posterior probability criteria) to calibrate and optimize parameter selection. When the AIC and SPPC values ​​are minimized, it is the optimal pollutant flux regression equation.

[0088]

[0089] In the formula, For the data set D(X i , i=1,n) the maximum likelihood value; is the maximum likelihood estimate of the equation parameters; d is the number of equation parameters; n is the number of data sets used for estimating the equation parameters.

[0090] Table 2 Commonly used river load estimation models in the LOADEST model

[0091]

[0092] The LOADEST model establishes parameter estimation methods based on three statistical estimation methods: asymptotic maximum likelihood estimation AMLE, maximum likelihood estimation MLE, and least absolute deviation method LAD. When the residual follows a normal distribution, AMLE is used to estimate the parameters of the censored data, as shown in formula (4); MLE is used to estimate the parameters of the non-censored data, as shown in formula (5); when the residual does not follow a normal distribution, LAD is used to estimate the parameters, as shown in formula (6).

[0093]

[0094] In the formula, is the pollutant flux estimated by AMLE, MVUE, and LAD methods; a0, a j are the coefficients of the equation calculated by different methods; H(a, b, s 2 ,α,k) is the likelihood approximation function of infinite series; g m (m,s 2 ,V) is the Bessel function; a, b, V are the independent variable functions; α, h are the parameters of the gamma distribution; m is the degree of freedom; s 2 is the residual variance; e k is the residual error; f is the number of missing data in the data.

[0095] Set up automatic water quality and quantity monitoring equipment at key points in small watersheds, such as Figure 2 As shown, in-situ monitoring of water quality and quantity at the basin scale is carried out. The water quantity monitoring equipment includes Parshall flume, pressure water level sensor and data acquisition transmitter that meet the industry standards of the Ministry of Water Resources, and the water quality monitoring equipment includes automatic analyzers for total nitrogen, total phosphorus, ammonia nitrogen and chemical oxygen demand.

[0096] Real-time monitoring, online transmission and data storage of indicator data are realized through the pollutant online monitoring system, such as Figure 3 As shown in the figure, the pollutant online monitoring system includes the perception layer, network layer and application layer. The automatic monitoring equipment for water quality and quantity is the perception layer, which is installed at the pollutant monitoring point and the process node that affects the pollutant emission, and is used to monitor the pollutant emission status and process parameters and complete the communication with the host computer. The data acquisition and transmission equipment is the network layer, which collects the data of the automatic monitoring equipment for water quality and quantity, completes the single-chip microcomputer, industrial computer, embedded computer, programmable automation controller or programmable controller for data storage and data transmission with the monitoring center. The monitoring center is the application layer, which is installed in the competent departments at all levels, and is linked to the automatic monitoring equipment through the network and issues query and control instructions to it.

[0097] Automatic water quality and quantity monitoring equipment was set up at key nodes of the small watershed where the field monitoring test was conducted, and water quality and quantity indicator data of farmland non-point source pollution at the watershed scale were collected. Topography, land use data and meteorological data were obtained through remote sensing satellites, drones and agricultural meteorological stations. Combined with the hydrological and water quality data collected at the field scale and the watershed scale, the InVEST model was used to simulate the farmland non-point source pollution load at the watershed scale.

[0098] The InVEST model results were corrected using field-scale and watershed-scale observation data. The model output error was evaluated by adjusting the Borselli K parameter and calculating the goodness of fit under different Borselli K parameters. The goodness of fit statistic was the determination coefficient R 2 , R 2 The maximum value is 1, R 2 The closer the value is to 1, the smaller the error of the model simulation value is than the observed value.

[0099]

[0100] In the formula, y 观测 is the observed value; is the average value of the observations; y 模拟 is the simulated value; is the average of the simulated values.

[0101] The InVEST model works as follows:

[0102] First, the nitrogen and phosphorus loss loads are distributed to the surface and subsurface according to the proportion, and the calculation formula is:

[0103] load surf,p =(1-proportion subsurface )×load_ βp

[0104] load sub,p =proportion subsurface ×load_ βp

[0105] Where β is the nutrient; p is a single pixel; load subsurf,p is the subsurface nutrient load;

[0106] The surface nutrient transfer rate is calculated as:

[0107]

[0108] Where, NDR 0,p is the nutrient transfer rate retained by downstream pixels; IC j is the terrain index; IC0 and k are calibration parameters;

[0109] The formula for calculating the subsurface nutrient transfer rate is:

[0110]

[0111] Where, NDR sub,p is the underground nutrient transfer rate; eff subs To achieve maximum nutrient retention efficiency in underground flow; subs is the interception length of groundwater flow; that is, the distance at which the soil is assumed to retain nutrients at its maximum capacity, and l is the distance from the pixel to the flow;

[0112] The formula for calculating nutrient output is:

[0113] U exp,p =load surf,p ×NDR surf,p +load subs,[ ×NDR subs,p

[0114]

[0115] Where U exp,p Load is the nutrient output load of each grid cell; surf,p is the surface nutrient load; NDR surf,p is the surface nutrient transfer rate; load subs,p is the subsurface nutrient load; NDR subs,p is the subsurface nutrient transfer rate; Export the total nutrient load for the sub-basin.

[0116] Collect the index data at the field scale and watershed scale, transmit and store the index data online, build an online monitoring and early warning platform for farmland non-point source pollution, construct an online intelligent monitoring and early warning network for farmland non-point source pollution, form an integrated Internet of Things overall planning system, and use the integrated Internet of Things for all-round monitoring to achieve online monitoring and early warning of regional farmland non-point source pollution, such as Figure 4 shown.

[0117] The monitoring and early warning platform includes a remote intelligent control platform, an early warning platform, and a nitrogen and phosphorus non-point source loss monitoring platform, which can realize online monitoring and early warning of farmland non-point source nitrogen and phosphorus loss. The IoT hardware monitoring system for farmland nitrogen and phosphorus non-point source pollution at the watershed scale is the main body, supplemented by field experiments to obtain basic data on nitrogen and phosphorus loss of different crops, perform InVEST parameter correction and optimization verification, predict and evaluate the water, nitrogen, and phosphorus material cycles of regional farmland ecosystems, build a monitoring and early warning platform, realize remote online intelligent monitoring of farmland non-point source pollution, and issue real-time early warnings for indicator data that exceeds the standard, verify and correlate the constructed network point data, and gradually radiate to other areas.

[0118] The present invention uses the NDR module of the InVEST model to evaluate nitrogen and phosphorus loss, and uses a machine learning model to convert the calibrated InVEST model simulation results into macro-scale simulation results to evaluate farmland non-point source pollution in the target area. Remote sensing, drones and meteorological stations are used to obtain farmland distribution, terrain slope and meteorological data in the basin. By using the NDR module of the InVEST model, combined with field-scale and basin-scale observation data, InVEST parameter calibration and optimization verification are performed.

[0119] The machine learning model is used to convert the simulation results of the revised InVEST model into macro-scale simulation results. At the same time, combined with ArcGIS hot and cold spot analysis, the distribution characteristics of farmland non-point pollution load, key source areas and prevention and control strategies in the target area are output. The principle of the machine learning model is as follows Figure 5 and Figure 6 As shown, Figure 5 For the machine learning algorithm process, Figure 6 For machine learning model process.

[0120] By adopting machine learning models and introducing professional external survey software "External Survey Assistant", we can obtain basic information on the target area's planting area, soil type, farming method, soil nutrients, fertilization method, land use method, terrain slope and meteorological data over the years. After collecting various types of agricultural non-point source pollution data, we use Geodatabase as a database based on the ArcGIS platform, and establish an agricultural data database, meteorological data database and geospatial database for agricultural non-point source pollution in agricultural areas. We calibrate and revise the InVEST model through field-scale and watershed-scale measured data, and convert the watershed-scale simulation into a macro-scale simulation through a machine learning model to establish a farmland non-point source model for the target area. At the same time, combined with ArcGIS hot and cold spot analysis, we output the distribution characteristics of farmland non-point source pollution loads, key source areas and prevention and control strategies in the target area.

[0121] The principles of ArcGIS hot and cold spot analysis are as follows:

[0122] ArcGIS hot spot analysis is based on Getis-OrdG i * The index is used to classify hot and cold spots. i * The index is a statistical method used to identify high-value and low-value clusters. It is often used to identify and explain spatial distribution patterns in geographical research. It is usually considered that extremely significant hot spots at a 99% confidence level are key pollution source areas. The calculation formula is as follows:

[0123]

[0124] In the formula, G i * is the Getis-Ord index; V j is the nitrogen and phosphorus emission intensity of the jth study unit; is the mean value of nitrogen emission intensity; W oj is the spatial weight coefficient of regions o and j; N total is the total number of research units.

[0125] The present invention utilizes the "1234" mode of farmland non-point source pollution monitoring, that is, through 4 crops, 3 platforms and 2 sets of models, and uses the Internet of Things technology to construct a macro-regional farmland non-point source pollution sky-ground integrated monitoring network, breaking the limitations of traditional monitoring methods, and targeting the problems of difficulty in monitoring and evaluating farmland non-point source pollution, inability to accurately measure the macro-scale agricultural non-point source pollution status, lack of key source area identification and prevention and control strategies, to achieve a multi-scale assessment of farmland non-point source pollution monitoring and assessment of "field scale-watershed scale-macro scale", accurately output macro-scale farmland non-point source pollution load distribution characteristics, key source areas and prevention and control strategies, to achieve real-time monitoring and early warning of farmland non-point source pollution, and the error of farmland non-point source pollution monitoring and assessment results is very small.

[0126] The above description is a detailed description of the preferred feasible embodiments of the present invention, but the embodiments are not intended to limit the scope of the patent application of the present invention. All equivalent changes or modified changes completed under the technical spirit disclosed by the present invention should fall within the patent scope covered by the present invention.

Claims

1. A multi-scale farmland non-point source pollution monitoring risk assessment and prevention and control method, characterized in that: The following steps are involved: Under typical terrain and typical crop conditions, farmland non-point source pollution monitoring points were set up according to different monitoring purposes to monitor farmland non-point source pollution at the field scale, and the LOADEST model was used to calculate the field scale nitrogen and phosphorus pollution fluxes. Install automatic water quality and quantity monitoring equipment at key nodes of small watersheds in fields to monitor non-point source pollution of farmland at the watershed scale, use the InVEST model to assess non-point source pollution at the watershed scale, and calibrate the InVEST model with observation data; Collect farmland non-point source pollution index data at the field scale and watershed scale, transmit and store the index data online, build an online monitoring and early warning platform for farmland non-point source pollution, and conduct online monitoring of farmland non-point source nitrogen and phosphorus loss; Obtain relevant data on farmland non-point source pollution, use the InVEST model and machine learning model to evaluate farmland non-point source pollution, and combine hot and cold spot analysis to output the distribution characteristics of farmland non-point source pollution load, key source areas and prevention and control strategies in the target area; By integrating multi-source heterogeneous data from field-scale monitoring, watershed-scale monitoring, and farmland non-point source pollution assessment, the Internet of Things technology is used to build an integrated monitoring network for farmland non-point source pollution that integrates land, sky, and air, to achieve monitoring, assessment, and early warning of farmland non-point source pollution.

2. The method for risk assessment and prevention of farmland non-point source pollution monitoring based on multiple scales according to claim 1 is characterized in that: Carry out field-scale monitoring of typical crops under typical terrain conditions. Typical crops represent four land use types: paddy fields, dry land, irrigated land, and gardens. Establish standardized monitoring plots for nitrogen and phosphorus loss in farmland, with monitoring indicators including runoff and leaching water quantity and quality, microclimate, soil moisture, fertilizer application and irrigation water quantity; The LOADEST model was used to calculate the nitrogen and phosphorus losses at the field scale, and the impact of soil physical and chemical properties, crop types, fertilization methods, fertilizer ratios, fertilizer amounts and rainfall agricultural environmental information on non-point source pollution was comprehensively analyzed.

3. The method for risk assessment and prevention of farmland non-point source pollution monitoring based on multiple scales according to claim 1 is characterized in that: The LOADEST model works as follows: The flux of non-point source pollutants from farmland is the total amount of pollutants passing through a certain section within a certain period of time: In the formula, is the pollutant flux in the time period τ; Δt is the time interval; NP is the number of discrete time intervals; Q is the average flow rate during the time period; C is the instantaneous water quality; is the instantaneous pollutant flux; The LOADEST model is based on equation (1) and uses the multivariate linear regression method to calculate the river pollutant flux. Make an estimate: In the formula, a0, a j is the coefficient of the equation; NV is the number of independent variables; X j is the independent variable; The LOADEST model provides 11 pollutant flux regression equations, and uses the AIC criterion and SPPC criterion to calibrate and optimize the parameters. When the AIC and SPPC values ​​are minimized, it is the optimal pollutant flux regression equation. In the formula, For the data set D(X i , i=1,n) the maximum likelihood value; is the maximum likelihood estimate of the equation parameters; d is the number of equation parameters; n is the number of data sets used for estimating the equation parameters; The LOADEST model establishes parameter estimation methods based on three statistical estimation methods: asymptotic maximum likelihood estimation AMLE, maximum likelihood estimation MLE, and least absolute deviation method LAD; when the residuals obey the normal distribution, AMLE is used to estimate the parameters of the censored data, as shown in formula (4); MLE is used to estimate the parameters of the non-censored data, as shown in formula (5); when the residuals do not obey the normal distribution, LAD is used to estimate the parameters, as shown in formula (6); In the formula, is the pollutant flux estimated by AMLE, MVUE, and LAD methods; a0, a j are the coefficients of the equation calculated by different methods; H(a, b, s 2 ,α,k) is the likelihood approximation function of infinite series; g m (m,s 2 ,V) is the Bessel function; a, b, V are the independent variable functions; α, h are the parameters of the gamma distribution; m is the degree of freedom; s 2 is the residual variance; e k is the residual error; f is the number of missing data in the data.

4. The method for risk assessment and prevention of farmland non-point source pollution monitoring based on multi-scale according to claim 1, characterized in that: Set up automatic water quality and quantity monitoring equipment at key nodes of small watersheds in fields to carry out in-situ water quality and quantity monitoring at the watershed scale. The water quantity monitoring equipment includes Parshall flume, pressure water level sensor and data acquisition transmitter. The water quality monitoring equipment includes automatic analyzer for total nitrogen, total phosphorus, ammonia nitrogen and chemical oxygen demand; Online transmission and data storage of index data are realized through the pollutant online monitoring system, which includes the perception layer, network layer and application layer; The water quality and quantity automatic monitoring equipment is the perception layer, which is installed at the pollutant monitoring site and the process nodes that affect pollutant emissions. It is used to monitor pollutant emissions and process parameters and complete communication with the host computer. The data acquisition and transmission equipment is the network layer, which collects data from the automatic monitoring equipment for water quality and quantity, and completes the single-chip microcomputer, industrial computer, embedded computer, programmable automation controller or programmable controller for data storage and data transmission with the monitoring center; The monitoring center is the application layer, which is installed in the competent departments at all levels. It is linked to the automatic monitoring equipment through the network and issues query and control instructions to it.

5. The method for risk assessment and prevention of farmland non-point source pollution monitoring based on multi-scale according to claim 1, characterized in that: The InVEST model results were corrected using field-scale and watershed-scale observation data. The model output error was evaluated by adjusting the Borselli K parameter and calculating the goodness of fit under different Borselli K parameters. The goodness of fit statistic was the determination coefficient R 2 , R 2 The maximum value is 1, R 2 The closer the value is to 1, the smaller the error of the model simulation value is than the observed value; In the formula, y 观测 is the observed value; is the average value of the observations; y 模拟 is the simulated value; is the average of the simulated values.

6. The multi-scale farmland non-point source pollution monitoring risk assessment and prevention and control method according to claim 5 is characterized in that: The InVEST model works as follows: First, the nitrogen and phosphorus loss loads are distributed to the surface and subsurface according to the proportion, and the calculation formula is: load surf,p =(1-proportion subsurface )×load_ βp load sub,p =proportion subsurface ×load_ βp Where β is the nutrient; p is a single pixel; load subsurf,p is the subsurface nutrient load; The surface nutrient transfer rate is calculated as: Where, NDR 0,p is the nutrient transfer rate retained by downstream pixels; IC j is the terrain index; IC0 and k are calibration parameters; The formula for calculating the subsurface nutrient transfer rate is: Where, NDR sub,p is the underground nutrient transfer rate; eff subs To achieve maximum nutrient retention efficiency in underground flow; subs is the interception length of groundwater flow; that is, the distance at which the soil is assumed to retain nutrients at its maximum capacity, and l is the distance from the pixel to the flow; The formula for calculating nutrient output is: U exp,p =load surf,p ×NDR surf,p +load subs,[ ×NDR subs,p Where U exp,p Load is the nutrient output load of each grid cell; surf,p is the surface nutrient load; NDR surf,p is the surface nutrient transfer rate; load subs,p is the subsurface nutrient load; NDR subs,p is the subsurface nutrient transfer rate; is the total nutrient load exported for the sub-basin.

7. The method for risk assessment and prevention of farmland non-point source pollution monitoring based on multi-scale according to claim 1, characterized in that: The monitoring and early warning platform includes a remote intelligent control platform, an early warning platform, and a nitrogen and phosphorus non-point source loss monitoring platform, which can realize online monitoring and early warning of nitrogen and phosphorus loss from farmland non-point sources; Based on the IoT hardware monitoring system of nitrogen and phosphorus non-point source pollution in farmland at the watershed scale, supplemented by field experiments to obtain basic data on nitrogen and phosphorus loss in different crops, InVEST parameter correction and optimization verification are carried out to predict and evaluate the water, nitrogen and phosphorus material cycles in regional farmland ecosystems, build a monitoring and early warning platform, realize remote online intelligent monitoring of non-point source pollution in farmland, and issue real-time early warnings for indicator data that exceed the standard. The constructed network data are mutually verified and correlated, and gradually radiate to other areas.

8. The method for risk assessment and prevention of farmland non-point source pollution monitoring based on multiple scales according to claim 1 is characterized in that: Through nitrogen and phosphorus loss assessment, remote sensing, drones and meteorological stations are used to obtain farmland distribution, terrain slope and meteorological data in the basin. The NDR module of the InVEST model is used to combine the observation data at the field scale and the basin scale to calibrate and optimize the InVEST parameters, and finally simulate the output of nitrogen and phosphorus nutrients at the basin scale.

9. The method for risk assessment and prevention of farmland non-point source pollution monitoring based on multiple scales according to claim 1, characterized in that: Through regional farmland nonpoint source assessment, machine learning models are used and professional survey software "Survey Assistant" is introduced to obtain basic information on the target area's planting area, soil type, farming method, soil nutrients, fertilization method, land use method, terrain slope and historical meteorological data; After collecting various types of agricultural non-point source pollution data, we used Geodatabase as the database based on the ArcGIS platform, and established agricultural data database, meteorological data database and geospatial database for agricultural non-point source pollution. We revised the InVEST model through the measured data at the field scale and watershed scale, and converted the watershed scale simulation into a macro-scale simulation through the machine learning model to establish the farmland non-point source model for the target area. Combined with ArcGIS hot and cold spot analysis, we output the distribution characteristics of farmland non-point source pollution load, key source areas and prevention and control strategies in the target area.

10. The method for risk assessment and prevention of farmland non-point source pollution monitoring based on multiple scales according to claim 9, characterized in that: The principles of ArcGIS hot and cold spot analysis are as follows: ArcGIS hot spot analysis is based on Getis-OrdG i * The index divides hot and cold spots into different types; Getis-OrdG i * The index is a statistical method for identifying clusters of high and low values. It is used in geographical research to identify and explain spatial distribution patterns. It considers highly significant hot spots at a 99% confidence level to be key pollution source areas. The calculation formula is: In the formula, G i * is the Getis-Ord index; V j is the nitrogen and phosphorus emission intensity of the jth study unit; is the mean value of nitrogen emission intensity; W oj is the spatial weight coefficient of regions o and j; N total is the total number of research units.

Images