Calculation method and system for large-scale watershed non-point source phosphorus load and flux

Abstract

The invention provides a large-scale watershed non-point source phosphorus load and flux calculation method and system, and the method comprises the steps: carrying out the simulation supplement of missing runoff data in hydrology and water quality data through employing an HSPF model; the method comprises the following steps: firstly, constructing a drainage basin basic database comprising drainage system data, pollution source data, spatial attribute data and hydrological water quality data of a large-scale drainage basin, inputting the data of the drainage basin basic database into an SPARROW model, calculating to obtain the non-point source phosphorus load of the large-scale drainage basin, and finally calculating the phosphorus flux according to the non-point source phosphorus load of the large-scale drainage basin. According to the method, the HSPF model and the SPARROW model are coupled, and the accuracy of large-scale watershed non-point source phosphorus load and flux calculation is effectively improved.

Description

Technical Field

[0001] The invention belongs to the technical field of agricultural non-point source pollution prevention and control, and specifically relates to a method and system for calculating non-point source phosphorus load and flux in a large-scale watershed. Background Art

[0002] Since the 1960s, the study of non-point source pollution has gradually attracted attention at home and abroad. With the enrichment of monitoring data and the in-depth study of mechanisms, since the 1970s, distributed hydrology knowledge has been gradually applied to the comprehensive analysis of the migration and transformation process of pollutants in the basin. A series of hydrological models and pollutant simulation models have been developed by researchers at home and abroad. By the 1990s, basin pollution models began to gradually consider the joint effects of regional space and environmental factors, and developed to comprehensively quantify factors such as regional climate, soil, topography, and land use. Since the 21st century, researchers at home and abroad have developed and improved many basin pollutant output models based on the scale, characteristics, and purposes of different river sections and research areas around the world, including the AGNPS model, ANSWERS model, HSPF model, MIKE-SHE model, SWAT model, SPARROW model, etc.

[0003] Since the characteristics and data availability of the study areas are not exactly the same, the research methods used are also slightly different. Empirical statistical and mechanistic models are usually effective tools for studying the source and migration of pollutants. The general expression of empirical statistical models is a simple linear relationship between river monitoring values ​​and the pollution sources and landscape pattern characteristics of the basin, and the model structure is relatively simple. Widely used empirical models, such as mass balance algorithms, kriging methods, and artificial neural networks, can depict the direct feedback of river outlets to phosphorus pollution sources and easily quantify the errors of model parameters and predicted values. Although this method can be applied to large basins, it often lacks explanations for the sources of pollutants and their migration mechanisms. Many distributed hydrological models, such as the HSPF model, the DWSM model, and the SWAT model, have been widely used to simulate the migration and fate of various pollutants at different spatial and temporal scales. Although these models have established capabilities to reproduce hydrology, nutrient transport, and river paths, they are often too complex and require monitoring data to be applied to any basin except the most intensively monitored basins. Previous studies have also shown that although these mechanistic models are calibrated using water pollution data, which makes the calibration error at the basin outlet very small, the errors caused by the data source and transportation process may be large, resulting in many uncertainties in the simulation effect. The HSPF, AGNPS and SWAT models mentioned above are all models that focus on mechanistic principles. The complex mechanistic principles make the models have higher requirements on the quantity and quality of input data, and also make the models more restricted in terms of applicability. Models that focus on statistics, such as SPARROW, Kriging, and artificial neural networks, have relatively loose requirements on input data, but still show ideal simulation effects in the generation and transmission of pollution loads and analysis of basin management strategies.

[0004] At present, the simulation of pollution output in China focuses on the application of conventional methods and small-scale experiments, and lacks the identification of pollutant loss characteristics in the first-level area of ​​the basin. In addition, the flow monitoring data of the first-level basin in my country still has problems such as missing sub-basin data or the monitoring points cannot fully cover the sub-basins divided by the first-level basin, which makes it difficult to simulate the hydrology and pollutants of the first-level basin. Summary of the invention

[0005] The purpose of the present invention is to address the above-mentioned problems existing in the prior art and to provide a method and system for calculating non-point source phosphorus load and flux in a large-scale watershed, which can realize accurate simulation of pollutants in a primary watershed.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows:

[0007] In the first aspect, the present invention proposes a method for calculating non-point source phosphorus load and flux in a large-scale watershed, comprising:

[0008] S1. Use the HSPF model to simulate and supplement the missing runoff data in the hydrological and water quality data;

[0009] S2. Construct a basin basic database including water system data, pollution source data, spatial attribute data, and hydrological and water quality data of large-scale basins;

[0010] S3, input the data of the basin basic database into the SPARROW model to calculate the non-point source phosphorus load of the large-scale basin;

[0011] S4. Calculate the phosphorus flux based on the non-point source phosphorus load in the large-scale watershed.

[0012] The S1 includes: inputting meteorological, DEM, land use and river network data into the HSPF model to obtain simulated values ​​of missing runoff data, and evaluating the simulation accuracy by comparing the simulated values ​​with the observed values. If the simulation accuracy cannot meet the requirements, the improved PEST algorithm is used for calibration;

[0013] The improved PEST algorithm determines the sensitivity coefficient of each parameter according to the following formula:

[0014]

[0015] In the above formula, S i is the sensitivity coefficient of the i-th parameter, ω i,j is the weight of the i-th parameter in the j-th sub-basin, Q k,j is the runoff of the jth sub-basin outputted when the HSPF model is run for the kth time, Q0 is the initial value of the calibration result, C k,j is the change ratio of the operating parameter value of the HSPF model during the kth operation relative to the parameter value after calibration, 0<C k,j <1, n is the number of runs of the HSPF model, and m is the number of sub-basins;

[0016] Parameters with sensitivity coefficients below the threshold will be deleted during model calibration.

[0017] In S3, before the data of the basin basic database is input into the SPARROW model, the LOADEST model is used to normalize the non-normal distribution data and censored data in the basin basic database;

[0018] The S4 uses the LOADEST model to calculate the nonpoint source phosphorus flux of a large-scale watershed, and for the water quality data of each monitoring point required by the LOADEST model, the monitoring point data of the main stream and main tributaries are preferentially used, and the data of the monitoring points with more than 2 missing phosphorus concentration monitoring data are screened out.

[0019] The water system data include river length, sub-basin area, river start and end nodes, river section code and river network density;

[0020] The pollution source data include land use data of farmland, construction land, forest land and grassland;

[0021] The spatial attribute data include annual average rainfall, annual average temperature, slope, river network density, runoff depth, population density, base saturation, soil salinity, soil bulk density, clay ratio, silt ratio, and sand ratio;

[0022] The hydrological and water quality data include monitored runoff and phosphorus concentration.

[0023] In S3, the data input into the SPARROW model also includes the annual average phosphorus load Load calculated based on the monitoring data:

[0024]

[0025] In the above formula, c n is the phosphorus concentration monitoring value in the nth month, q n is the runoff in the nth month, d n The day of the nth month.

[0026] In a second aspect, the present invention proposes a large-scale basin non-point source phosphorus load and flux calculation system, including a data preprocessing module, a basic database construction module, a phosphorus load calculation module, and a phosphorus flux calculation module, wherein the data preprocessing module includes a runoff data supplement unit;

[0027] The runoff data supplement unit is used to simulate and supplement the missing runoff data in the hydrological and water quality data using the HSPF model;

[0028] The basic database construction module is used to construct a basin basic database including water system data, pollution source data, spatial attribute data and hydrological and water quality data of a large-scale basin;

[0029] The phosphorus load calculation module is used to input the data of the basin basic database into the SPARROW model to calculate the non-point source phosphorus load of the large-scale basin;

[0030] The phosphorus flux calculation module is used to calculate the phosphorus flux according to the non-point source phosphorus load of a large-scale watershed.

[0031] The operation process of the runoff data supplement unit includes: inputting meteorological, DEM, land use and river network data into the HSPF model to obtain the simulated value of the missing runoff data, and evaluating the simulation accuracy by comparing the simulated value with the observed value. If the simulation accuracy cannot meet the requirements, the improved PEST algorithm is used for calibration;

[0032] The improved PEST algorithm determines the sensitivity coefficient of each parameter according to the following formula:

[0033]

[0034] In the above formula, S i is the sensitivity coefficient of the i-th parameter, ω i,j is the weight of the i-th parameter in the j-th sub-basin, Q k,j is the runoff of the jth sub-basin outputted when the HSPF model is run for the kth time, Q0 is the initial value of the calibration result, C k,j is the change ratio of the operating parameter value of the HSPF model during the kth operation relative to the parameter value after calibration, 0<C k,j <1, n is the number of runs of the HSPF model, and m is the number of sub-basins;

[0035] Parameters with sensitivity coefficients below the threshold will be deleted during model calibration.

[0036] The data preprocessing module also includes a standard processing unit;

[0037] The standard processing unit is used to use the LOADEST model to perform standardization processing on the non-normal distribution data and censored data in the basin basic database before inputting the data in the basin basic database into the SPARROW model;

[0038] The phosphorus flux calculation module uses the LOADEST model to calculate the non-point source phosphorus flux of a large-scale watershed, and for the water quality data of each monitoring point required by the LOADEST model, the monitoring point data of the main stream and main tributaries are preferentially used, and the data of the monitoring points with more than 2 missing phosphorus concentration monitoring data are screened out.

[0039] The water system data include river length, sub-basin area, river start and end nodes, river section code and river network density;

[0040] The pollution source data include land use data of farmland, construction land, forest land and grassland;

[0041] The spatial attribute data include annual average rainfall, annual average temperature, slope, river network density, runoff depth, population density, base saturation, soil salinity, soil bulk density, clay ratio, silt ratio, and sand ratio;

[0042] The hydrological and water quality data include monitored runoff and phosphorus concentration.

[0043] The data input into the SPARROW model also includes the annual average phosphorus load Load calculated based on monitoring data:

[0044]

[0045] In the above formula, c n is the phosphorus concentration monitoring value in the nth month, q n is the runoff in the nth month, d n The day of the nth month.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] 1. The present invention provides a method for calculating the non-point source phosphorus load and flux of a large-scale watershed. The HSPF model is first used to simulate and supplement the runoff data missing from the hydrological and water quality data, and then a basic watershed database including the water system data, pollution source data, spatial attribute data, and hydrological and water quality data of the large-scale watershed is constructed. The data of the basic watershed database is then input into the SPARROW model to calculate the non-point source phosphorus load of the large-scale watershed, and finally the phosphorus flux is calculated according to the non-point source phosphorus load of the large-scale watershed. The method couples the HSPF model with the SPARROW model, and compared with a single model, the accuracy of the non-point source phosphorus load and flux calculation of the large-scale watershed is effectively improved.

[0048] 2. In the process of simulating the missing runoff data in the hydrological and water quality data using the HSPF model, if the simulation accuracy cannot meet the requirements, the improved PEST algorithm is used for calibration. The improved PEST algorithm proposes a method for parameter screening based on the sensitivity coefficient. Parameters with low sensitivity coefficients are deleted when the model is calibrated. This method effectively improves the efficiency of parameter screening and the accuracy of simulation results in the large-scale watershed simulation process. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flow chart for extracting the river network in the Yellow River Basin in Example 1.

[0050] Figure 2 It is a comparison of the measured value and the predicted value of the total phosphorus load in Example 1 and the residual diagram.

[0051] Figure 3 This is the phosphorus load and phosphorus flux distribution diagram obtained in Example 1.

[0052] Figure 4 The structure diagram of the system of the present invention is shown in FIG. DETAILED DESCRIPTION

[0053] The present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0054] In recent years, there have been few studies on the simulation and calculation of phosphorus load transmission models in large-scale watersheds in my country. The main reason is that the mechanistic model has high requirements on the quantity and quality of input data, and needs to use water pollution data for calibration at the outlet of the watershed. Although the calibration error at the outlet of the watershed is small, there may be large errors in estimating the pollution sources and transmission processes in large-scale watersheds. Although the empirical model can be applied to large watersheds, it cannot quantitatively analyze the spatial transmission process of phosphorus. Therefore, for large-scale watersheds, considering the influence of model complexity, data availability, user goals and cost-benefit analysis, the present invention adopts the mechanism HSPF and empirical LOADEST models in data preprocessing, and couples the SPARROW model in data simulation to collaboratively construct a calculation framework for non-point source phosphorus compliance and flux in large-scale watersheds. The results are helpful to provide a scientific basis for the prevention and control of non-point source pollution in large-scale watersheds in China and the promotion of models.

[0055] Embodiment 1:

[0056] This example uses the Yellow River Basin as a research object to implement a large-scale basin non-point source phosphorus load and flux calculation method described in the present invention. The specific steps are as follows:

[0057] 1. Since the runoff monitoring points in the Yellow River Basin cannot completely cover the 60 sub-basins, it is necessary to supplement the data for the basins without runoff data. This embodiment uses the HSPF model to simulate and supplement the missing runoff data. The specific process is as follows:

[0058] Flow simulation requires processing land use data, DEM data, river data and meteorological data in the BASINS 4.5 environment. First, convert the meteorological data into wdm data format through the WDMutil plug-in. Each type of data corresponds to a DSN code. Then input these data into the HSPF model, overlay DEM, land use and river network data, generate HSPF project files, and then open the HSPF project files in the WinHSPF 3.1 environment. After running, you can get the simulation value of the missing flow data.

[0059] For the evaluation of the model simulation effect, this embodiment compares the simulated value with the observed value, and uses relative error, Nash efficiency coefficient and correlation coefficient to perform the evaluation in turn. The specific formula is as follows:

[0060]

[0061]

[0062] In the above formula, δ, NSE, and R are relative error, Nash efficiency coefficient, and correlation coefficient, respectively. The closer δ is to 0, the better the simulation effect. In this embodiment, the simulation effect is good when δ is less than 10, the simulation effect is good when NSE is greater than 0.7, and the simulation effect is good when R is greater than 0.75. Q si , Q oi are respectively the simulated and observed values ​​of runoff, and Q sia , Q oia are the averages of the simulated and observed values, respectively.

[0063] If the simulation effect evaluated by relative error, Nash efficiency coefficient and correlation coefficient is not good, the improved PEST algorithm is used for automatic calibration. The improved PEST algorithm determines the sensitivity coefficient of each parameter according to the following formula:

[0064]

[0065] In the above formula, S i is the sensitivity coefficient of the i-th parameter, ω i,j is the weight of the i-th parameter in the j-th sub-basin, Q k,j is the runoff of the jth sub-basin outputted when the HSPF model is run for the kth time, Q0 is the initial value of the calibration result, C k,j is the change ratio of the operating parameter value of the HSPF model during the kth operation relative to the parameter value after calibration, 0<C k,j <1, n is the number of runs of the HSPF model, and m is the number of sub-basins;

[0066] Parameters with sensitivity coefficients below the threshold will be deleted during model calibration.

[0067] 2. Construct a basin basic database, which includes water system data, pollution source data, spatial attribute data of large-scale basins, and hydrological and water quality data supplemented by HSPF model simulation.

[0068] The water system data include river length, sub-basin area, river start and end nodes, river segment code and river network density. These data are derived from the digital elevation model (DEM) of the Yellow River Basin obtained from the Resource and Environment Data Cloud Platform. Using ArcGIS and Archhydro Tools, the DEM data is subjected to boundary correction, depression filling, flow direction extraction, cumulative confluence calculation, river network generation and sub-basin division, and the topological structure and attributes of the Yellow River network are extracted, such as Figure 1As shown in the figure. The data of the river is in the form of node connection coding, that is, each section of the river has two attributes: the starting point and the end point. The numbers of these two attributes represent the specific river connection. The main extracted data include river length, sub-basin area, river start and end nodes, river section code and river network density. Finally, the Yellow River Basin was divided into 60 sub-basins.

[0069] The pollution source data include land use data of farmland, construction land, forest land and grassland. Considering that the total phosphorus load in the Yellow River Basin mainly comes from fertilizer application, sewage discharge and animal husbandry corresponding to land use, this embodiment selects the land use data of farmland, construction land, forest land and grassland as the pollution source input, with area as the unit. The land use data comes from the resource and environment data cloud platform. In order to obtain the land use area of ​​each sub-basin, the land use raster map of the Yellow River Basin and the sub-basin division shp file are input into ArcGIS, and the "Spatial Attribute Tabulation" function under the "Reclassification and Spatial Analysis Tools" is used to output the area distribution of different land uses in each sub-basin.

[0070] The spatial attribute data include annual average rainfall, annual average temperature, slope, river network density, runoff depth, population density, base saturation, soil salinity, soil bulk density, clay ratio, silt ratio, and sand ratio. The transmission parameters selected in the initial stage of the model include annual average rainfall, annual average temperature, slope, river network density, runoff depth, clay ratio, silt ratio, and sand ratio. The meteorological and soil attribute data are all derived from the resource and environmental data cloud platform with an accuracy of 1km×1km. The river network density is the ratio of the river length to the basin area in the basic data of the water system, and the runoff depth is the ratio of the annual average flow of the basin to the basin area. The selection of spatial attribute data will directly affect the operation results of the model. When the model fitting, model estimation coefficients, and residual graphs reach the optimal model specifications, the final model parameter set is obtained.

[0071] The hydrological and water quality data include the monitored runoff volume and phosphorus concentration, and the monitoring points cover the main stream and major tributaries.

[0072] 3. Preprocess the data of the basin basic database, including using the LOADEST model to standardize the non-normal distribution data and censored data.

[0073] 4. Input the pre-processed data of the basin basic database and the annual average phosphorus load into the SPARROW model to calculate the non-point source phosphorus load of the large-scale basin. Among them, the annual average phosphorus load data is an important reference object for model parameter correction, which is mainly calculated through phosphorus concentration monitoring data and flow data. The specific calculation formula is:

[0074]

[0075] In the above formula, Load is the annual average phosphorus load, c n is the phosphorus concentration monitoring value in the nth month, q n is the runoff in the nth month, d n The day of the nth month.

[0076] 5. Based on the non-point source phosphorus load of the large-scale watershed, the LOADEST model is used to calculate the non-point source phosphorus flux of the large-scale watershed. In the process of parameter estimation, the model uses the phosphorus concentration monitoring data of the river and the calculated annual average phosphorus load data as the response variable, and gives priority to the monitoring point data of the main stream and main tributaries, and screens out the data of the monitoring points with more than 2 missing phosphorus concentration monitoring data to ensure the accuracy of the data input into the LOADEST model.

[0077] Model Validation:

[0078] (1) The SPARROW model has a large number of built-in parameters, and the initial values ​​and value ranges are set by referring to the SPARROW user manual. After the setting is completed, all the above model data are converted into SAS table data format, input into the model, and the data system of the Yellow River Basin is built. The model parameters are calibrated according to the SAS interface operation log information, and simulation equations with high simulation accuracy, good parameter sensitivity and values ​​consistent with the physical and chemical behavior of pollutants are selected (as shown in Table 1).

[0079] Table 1 Evaluation of SPARROW model parameters in the Yellow River Basin

[0080]

[0081] In order to verify the robustness of the model, the non-parametric bootstrap method was used to determine the mean and confidence interval of the parameters. The parameters and sampling estimation results of the model are shown in Figure 2 shown.

[0082] Through Table 1 and Figure 2 As can be seen, most of the phosphorus pollution source and export coefficients are very significant (p < 0.05), indicating that most variables are important in describing phosphorus loads. These coefficients are stable (standard errors are small, 95% confidence intervals are narrow, and the average bootstrap estimate is within 5% of the model coefficient estimate). The simulated loads of the SPARROW model show a good correlation with the monitored loads, with r 2=0.84, p<0.01, indicating that the model can accurately estimate the phosphorus load in the basin. The hypothesis t test and residual analysis were performed using SPSS software, and the final t test result p=0.001, showing a high degree of significance. In the residual distribution diagram, the residual values ​​are all in a fixed range of -2 to 2, indicating that the input data of the model and the simulated values ​​of the corresponding parameters are valid and can reflect the differences in phosphorus load in different regions. In general, the simulation results of SPARROW can better reflect the phosphorus load and its spatial differences in the basin, and can be used for the simulation analysis of phosphorus load differences in the Yellow River Basin.

[0083] (2) The spatial distribution of phosphorus load along the entire Yellow River basin and the upper, middle and lower reaches obtained in Example 1 is shown in Figure 3 The total phosphorus load in the Yellow River Basin is 41.76×10 4 t / yr, and the average phosphorus load in the Yellow River sub-basin is 696.01t / yr. The phosphorus load in different regions shows strong spatial heterogeneity, and is ranked as midstream (736.08t / yr)> downstream (549.73t / yr)> upstream (470.32t / yr). The load in Qinghai Province, the source of the Yellow River, is relatively low, and the phosphorus load is at a medium level before entering Gansu Province and reaching Toudaoguai in the Inner Mongolia Autonomous Region, which is the boundary between the upper and middle reaches. The sub-basins at the source of the main stream or the source of the tributaries have low loads due to the lack of runoff accumulation of phosphorus loads, and most of them are below 400t / yr, which is less than 2 / 3 of the average value. The phosphorus load in most tributaries is also at a low level, which is related to the low annual runoff of the tributaries in the Yellow River Basin. The average annual runoff is generally less than 100m 3 / s, when the phosphorus concentration remains the same as that of the main stream, the total phosphorus that the river can carry is limited, so the load is relatively small. The middle reaches of the Yellow River flow through the Loess Plateau, mainly through Shaanxi Province and Shanxi Province, and the phosphorus load has increased significantly. It is at the highest level in the entire basin as a whole, with the highest value reaching 900.38t / yr. The load in the lower reaches of the Yellow River is significantly lower than that in the middle reaches, but the average value is still greater than that in the upper reaches. In addition, from the perspective of the entire basin, due to the small flow of the tributaries, the amount of water and sediment received from the upstream is small, and there is a lack of phosphorus transmission channels, so the phosphorus load of the main stream is much higher than that of the tributaries.

[0084] Since the area of ​​each sub-basin is different, the pollution load obtained by taking the sub-basin as the spatial unit cannot represent the difference in the spatial distribution of the load of different pollution sources. Therefore, on the basis of calculating the sub-basin load, the non-point source pollution load generated per unit area of ​​the basin, that is, the phosphorus flux, is calculated to reflect the output intensity of the pollution load in space. The average phosphorus flux in the Yellow River Basin is 50.89 kg / km 2 ·yr -1 The average phosphorus yield in different regions is as follows: downstream (158.93 kg / km 2 ·yr-1 )>Upstream (69.92kg / km 2 ·yr -1 )>Midstream (6.45kg / km 2 ·yr -1 The three sub-basins with the highest phosphorus flux mainly flow through Henan Province in the downstream, Gansu Province in the upstream, and Ningxia Hui Autonomous Region, reaching 543.673 kg / km 2 ·yr -1 、472.6kg / km 2 ·yr -1 and 348.2kg / km 2 ·yr -1 The sub-basins with medium yields are mainly located in the upper and middle reaches of the basin, while most areas in the middle reach are at low levels. Although the phosphorus load in the upper reaches is low, due to the high phosphorus output of some cities, most of its sub-basins have medium fluxes. However, the phosphorus yield in the high-load midstream is low (10-20kg / km 2 ·yr -1 ), which is related to the large proportion of grassland and forest land use in the middle reaches, reducing the output intensity of the middle reaches. But it also shows that the accumulation and transmission of phosphorus load in the upper reaches have a certain impact on the high load level in the middle reaches.

[0085] Embodiment 2:

[0086] A system for calculating non-point source phosphorus loads and fluxes in large-scale watersheds, such as Figure 4 As shown, it includes a data preprocessing module, a basic database construction module, a phosphorus load calculation module, and a phosphorus flux calculation module. The data preprocessing module includes a runoff data supplement unit and a standard processing unit;

[0087] The runoff data supplement unit is used to simulate and supplement the missing runoff data in the hydrological and water quality data using the HSPF model. Its operation process includes: inputting meteorological, DEM, land use and river network data into the HSPF model to obtain the simulated value of the missing runoff data, and evaluating the simulation accuracy by comparing the simulated value with the observed value. If the simulation accuracy cannot meet the requirements, the improved PEST algorithm is used for calibration. The improved PEST algorithm determines the sensitivity coefficient of each parameter according to the following formula:

[0088]

[0089] In the above formula, S i is the sensitivity coefficient of the i-th parameter, ω i,j is the weight of the i-th parameter in the j-th sub-basin, Q k,jis the runoff of the jth sub-basin outputted when the HSPF model is run for the kth time, Q0 is the initial value of the calibration result, C k,j is the change ratio of the operating parameter value of the HSPF model during the kth operation relative to the parameter value after calibration, 0<C k,j <1, n is the number of runs of the HSPF model, and m is the number of sub-basins;

[0090] Parameters with sensitivity coefficients below the threshold will be deleted during model calibration.

[0091] The basic database construction module is used to construct a basin basic database including water system data, pollution source data, spatial attribute data and hydrological and water quality data of a large-scale basin, wherein the water system data includes river length, sub-basin area, river starting and ending nodes, river section code and river network density; the pollution source data includes land use data of farmland, construction land, forest land and grassland; the spatial attribute data includes annual average rainfall, annual average temperature, slope, river network density, runoff depth, population density, base saturation, soil salinity, soil bulk density, clay ratio, silt ratio and sand ratio; the hydrological and water quality data includes monitored runoff and phosphorus concentration.

[0092] The standard processing unit is used to use the LOADEST model to perform standardization processing on the non-normal distribution data and censored data in the basin basic database before inputting the data into the SPARROW model.

[0093] The phosphorus load calculation module is used to input the data of the basin basic database and the annual average phosphorus load into the SPARROW model to calculate the non-point source phosphorus load of the large-scale basin, wherein the annual average phosphorus load Load is calculated according to the following formula:

[0094]

[0095] In the above formula, c n is the phosphorus concentration monitoring value in the nth month, q n is the runoff in the nth month, d n The day of the nth month.

[0096] The phosphorus flux calculation module is used to calculate the non-point source phosphorus flux of a large-scale watershed according to the non-point source phosphorus load of the large-scale watershed using the LOADEST model. For the water quality data of each monitoring point required by the LOADEST model, the monitoring point data of the main stream and main tributaries are preferably used, and the data of the monitoring points with more than two missing phosphorus concentration monitoring data are screened out.

Claims

1. A method for calculating non-point source phosphorus load and flux in a large-scale watershed, characterized in that: The method comprises: S1. Use the HSPF model to simulate and supplement the missing runoff data in the hydrological and water quality data; S2. Construct a basin basic database including water system data, pollution source data, spatial attribute data, and hydrological and water quality data of large-scale basins; S3, input the data of the basin basic database into the SPARROW model to calculate the non-point source phosphorus load of the large-scale basin; S4. Calculate the phosphorus flux based on the non-point source phosphorus load in the large-scale watershed.

2. The method for calculating the non-point source phosphorus load and flux in a large-scale watershed according to claim 1, characterized in that: The S1 includes: inputting meteorological, DEM, land use and river network data into the HSPF model to obtain simulated values ​​of missing runoff data, and evaluating the simulation accuracy by comparing the simulated values ​​with the observed values. If the simulation accuracy cannot meet the requirements, the improved PEST algorithm is used for calibration; The improved PEST algorithm determines the sensitivity coefficient of each parameter according to the following formula: In the above formula, S i is the sensitivity coefficient of the i-th parameter, ω i,j is the weight of the i-th parameter in the j-th sub-basin, Q k,j is the runoff of the jth sub-basin outputted when the HSPF model is run for the kth time, Q0 is the initial value of the calibration result, C k,j is the change ratio of the operating parameter value of the HSPF model during the kth operation relative to the parameter value after calibration, 0<C k,j <1, n is the number of runs of the HSPF model, and m is the number of sub-basins; Parameters with sensitivity coefficients below the threshold will be deleted during model calibration.

3. The method for calculating the non-point source phosphorus load and flux in a large-scale watershed according to claim 1 or 2, characterized in that: In S3, before the data of the basin basic database is input into the SPARROW model, the LOADEST model is used to normalize the non-normal distribution data and censored data in the basin basic database; The S4 uses the LOADEST model to calculate the nonpoint source phosphorus flux of a large-scale watershed, and for the water quality data of each monitoring point required by the LOADEST model, the monitoring point data of the main stream and main tributaries are preferentially used, and the data of the monitoring points with more than 2 missing phosphorus concentration monitoring data are screened out.

4. The method for calculating the non-point source phosphorus load and flux in a large-scale watershed according to claim 1 or 2, characterized in that: The water system data include river length, sub-basin area, river start and end nodes, river section code and river network density; The pollution source data include land use data of farmland, construction land, forest land and grassland; The spatial attribute data include annual average rainfall, annual average temperature, slope, river network density, runoff depth, population density, base saturation, soil salinity, soil bulk density, clay ratio, silt ratio, and sand ratio; The hydrological and water quality data include monitored runoff and phosphorus concentration.

5. The method for calculating the non-point source phosphorus load and flux in a large-scale watershed according to claim 1 or 2, characterized in that: In S3, the data input into the SPARROW model also includes the annual average phosphorus load Load calculated based on the monitoring data: In the above formula, c n is the phosphorus concentration monitoring value in the nth month, q n is the runoff in the nth month, d n The day of the nth month.

6. A system for calculating non-point source phosphorus load and flux in a large-scale watershed, characterized in that: The system comprises a data preprocessing module, a basic database construction module, a phosphorus load calculation module, and a phosphorus flux calculation module, wherein the data preprocessing module comprises a runoff data supplement unit; The runoff data supplement unit is used to simulate and supplement the missing runoff data in the hydrological and water quality data using the HSPF model; The basic database construction module is used to construct a basin basic database including water system data, pollution source data, spatial attribute data and hydrological and water quality data of a large-scale basin; The phosphorus load calculation module is used to input the data of the basin basic database into the SPARROW model to calculate the non-point source phosphorus load of the large-scale basin; The phosphorus flux calculation module is used to calculate the phosphorus flux according to the non-point source phosphorus load of a large-scale watershed.

7. A large-scale basin non-point source phosphorus load and flux calculation system according to claim 6, characterized in that: The operation process of the runoff data supplement unit includes: inputting meteorological, DEM, land use and river network data into the HSPF model to obtain the simulated value of the missing runoff data, and evaluating the simulation accuracy by comparing the simulated value with the observed value. If the simulation accuracy cannot meet the requirements, the improved PEST algorithm is used for calibration; The improved PEST algorithm determines the sensitivity coefficient of each parameter according to the following formula: In the above formula, S i is the sensitivity coefficient of the i-th parameter, ω i,j is the weight of the i-th parameter in the j-th sub-basin, Q k,j is the runoff of the jth sub-basin outputted when the HSPF model is run for the kth time, Q0 is the initial value of the calibration result, C k,j is the change ratio of the operating parameter value of the HSPF model during the kth operation relative to the parameter value after calibration, 0<C k,j <1, n is the number of runs of the HSPF model, and m is the number of sub-basins; Parameters with sensitivity coefficients below the threshold will be deleted during model calibration.

8. A large-scale basin non-point source phosphorus load and flux calculation system according to claim 6 or 7, characterized in that: The data preprocessing module also includes a standard processing unit; The standard processing unit is used to use the LOADEST model to perform standardization processing on the non-normal distribution data and censored data in the basin basic database before inputting the data in the basin basic database into the SPARROW model; The phosphorus flux calculation module uses the LOADEST model to calculate the non-point source phosphorus flux of a large-scale watershed, and for the water quality data of each monitoring point required by the LOADEST model, the monitoring point data of the main stream and main tributaries are preferentially used, and the data of the monitoring points with more than 2 missing phosphorus concentration monitoring data are screened out.

9. A large-scale basin non-point source phosphorus load and flux calculation system according to claim 6 or 7, characterized in that: The water system data include river length, sub-basin area, river start and end nodes, river section code and river network density; The pollution source data include land use data of farmland, construction land, forest land and grassland; The spatial attribute data include annual average rainfall, annual average temperature, slope, river network density, runoff depth, population density, base saturation, soil salinity, soil bulk density, clay ratio, silt ratio, and sand ratio; The hydrological and water quality data include monitored runoff and phosphorus concentration.

10. A large-scale basin non-point source phosphorus load and flux calculation system according to claim 6 or 7, characterized in that: The data input into the SPARROW model also includes the annual average phosphorus load Load calculated based on monitoring data: In the above formula, c n is the phosphorus concentration monitoring value in the nth month, q n is the runoff in the nth month, d n The day of the nth month.

Images