A field water-holding capacity fusion calculation method based on multi-source soil data

Abstract

The application relates to the technical field of soil hydrology, in particular to a field water-holding capacity fusion calculation method based on multi-source soil data. The method can effectively improve the field water-holding capacity estimation precision and reduce errors by comprehensively considering the precision characteristics of multi-source soil data, constructing a field water-holding capacity fusion factor set based on multi-source soil data based on a trained and existing soil transfer function, and determining an optimal fusion scheme, and solves the technical problem that the existing field water-holding capacity estimation method has a large error.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of soil hydrology, in particular to a field water capacity fusion calculation method based on multi-source soil data. BACKGROUND

[0002] The description in this section is only provided for background information related to the present application and can not constitute the prior art.

[0003] Field water capacity is one of the most critical hydrological parameters of soil, and plays an important role in the fields of hydrological simulation, agricultural irrigation, drought monitoring, etc. Field water capacity refers to the soil water content value when the capillary suspended water in the soil reaches the maximum. Field water capacity is the highest amount of water that the soil can hold under field or natural conditions without the influence of groundwater, and is a key parameter for dividing runoff mode in hydrological models. Field water capacity is the upper limit of the available water in soil for crops, and the part of soil water content exceeding the field water capacity will not be retained by the soil but will seep downward as free gravity water, which can be used to determine the amount of agricultural irrigation water. In the national agricultural drought monitoring business standard, the adopted drought monitoring index is the soil relative humidity index, which is the ratio of soil water content to field water capacity. Therefore, accurate estimation of soil field water capacity has important scientific value and guiding significance for the research in the fields of hydrology, agriculture, ecology, etc.

[0004] The current methods for obtaining field water capacity mainly include site measurement method and soil transfer function method. Among them, the site measurement method can be divided into natural precipitation method, enclosure irrigation method, sandbox method and cutting ring method. The cutting ring method uses cutting ring to collect soil samples and measure the field water capacity data in the laboratory conditions, which is suitable for various soil conditions, the measurement results are close to the enclosure irrigation method, and the cost of manpower, material resources and financial resources is low. It is the current business standard method for measuring field water capacity in agricultural and water conservancy departments nationwide.

[0005] The site measurement method is only applicable to site-scale field water holding capacity measurement, and requires huge manpower, material resources and financial resources for large-scale field water holding capacity calculation. The soil transfer function method is a commonly used method for obtaining field water holding capacity at a larger scale. The soil transfer function method was proposed by Bouma et al. in the 1980s. The method mainly establishes a relationship model between soil property data (such as soil physical and chemical property data) and its hydraulic parameters (such as field water holding capacity) by analyzing the existing relatively easy-to-obtain soil property data, and then uses the relationship model to determine the distribution of soil hydraulic parameters. Common relationship model construction methods include multiple regression models, artificial neural network models, and classification regression trees. However, the accuracy of simulating large-scale field water holding capacity by the soil transfer function is affected by the uncertainty of soil data and soil transfer function. When the soil data quality of a certain region is poor or the applicability of the soil transfer function is poor, it will lead to a large error in the estimation of field water holding capacity. SUMMARY

[0006] The purpose of the embodiments of the present application is to provide a field water holding capacity fusion calculation method based on multi-source soil data, which solves the technical problem of large error in estimating field water holding capacity by using the soil transfer function method.

[0007] In a first aspect, a field water holding capacity fusion calculation method based on multi-source soil data is provided, comprising the following steps:

[0008] Step 1, obtaining multiple sets of data source soil data and measured field water holding capacity data measured by the soil moisture station in the target area;

[0009] Step 2, matching the measured field water holding capacity data with the grid cell soil data in any one set or more than two sets of data source soil data based on the nearest neighbor pixel method, extracting the soil parameters in the data source soil data, and performing sensitivity analysis on the soil parameters in the data source soil data by principal component analysis, and selecting the soil parameters affecting the field water holding capacity as the field water holding capacity influencing factors;

[0010] Step 3, training and establishing a training soil transfer function representing the relationship between each soil parameter and the measured field water holding capacity according to the selected field water holding capacity influencing factors; in addition to the training soil transfer function, the corresponding simulated field water holding capacity data is constructed for each grid cell soil data in each set of data source soil data according to the known existing soil transfer function;

[0011] Step 4, selecting soil grid cells with measured field water holding capacity data from the target area as measured grid cells, and matching the corresponding soil data of each measured grid cell as measured grid cell soil data according to different data source soil data;

[0012] calculating a set of simulated field capacity from each set of source soil data: the set of simulated field capacity includes a subset of field capacity data corresponding to each soil transfer function in step 3, and each subset of simulated field capacity includes the simulated field capacity of each measured grid cell calculated by the set of source soil data and the soil transfer function corresponding to the subset of simulated field capacity;

[0013] Step 5, selecting a subset of simulated field capacity data from each set of simulated field capacity to form a set of fusion factors, and fusing the set of fusion factors: taking the simulated field capacity data and the determination coefficient of the measured field capacity as the fusion weight value, and weighting and fusing each fusion factor according to its fusion weight value to obtain the fused field capacity;

[0014] Step 6, repeating step 5 to calculate the fused field capacity under all fusion factor combination conditions;

[0015] Step 7, verifying the accuracy of each fused field capacity corresponding to the measured grid cell according to the measured field capacity of the measured grid cell, determining the optimal fusion scheme, and calculating the fused field capacity of other soil grid cells other than the measured grid cell according to the optimal fusion scheme to form the fused field capacity distribution of the target area.

[0016] In a possible implementation, the formula for fusing each fusion factor in step 5 is:

[0017]

[0018] wherein FC ensemble is the fused field capacity, σ obs is the standard deviation of the measured field capacity, σ i is the standard deviation of the i-th fusion factor in the set of fusion factors, is the determination coefficient of the i-th fusion factor in the set of fusion factors and the measured field capacity, FC i is the i-th fusion factor in the set of fusion factors, is the mean field capacity of the i-th fusion factor in the set of fusion factors, is the mean of the measured field capacity, and △FC is the deviation correction term from the actual value.

[0019] In a possible implementation, after the measured field capacity is measured in step 1, the measured field capacity is compared with the existing field capacity through analysis of the physical properties of the sampled soil, and the historical drought data and reference results are verified.

[0020] In a possible implementation, the multiple sets of data source soil data include FAO soil data, BNU soil data and SG soil data.

[0021] In a possible implementation, step 1 includes converting the spatial resolution of the multiple sets of data source soil data into a uniform size through a scale conversion algorithm.

[0022] In a possible implementation, step 2 includes: based on the principal component analysis result, analyzing the influence degree of different soil parameters on the field moisture capacity by performing correlation analysis on the measured field moisture capacity and multiple site soil characteristic factors; and verifying the rationality of the soil parameters as the field moisture capacity influencing factors by analyzing the soil characteristic influencing factors in the known existing soil transfer function.

[0023] In a possible implementation, in step 3, the soil transfer function is trained based on at least one of a multiple regression method, a neural network method and a support vector machine method.

[0024] In a possible implementation, step 7 includes: taking the measured field moisture capacity as reference data, calculating statistical indicators of each simulated field moisture capacity data subset, the statistical indicators including a standard deviation, a mean value, a variation coefficient, a determination coefficient with the measured field moisture capacity and a root mean square error, and analyzing the error characteristics of the simulated field moisture capacity data subset.

[0025] In a possible implementation, for a desert area, a soil transfer function suitable for simulating the field moisture capacity of the desert is selected, and the field moisture capacity distribution map of the desert area is calculated based on the multiple sets of data source soil data.

[0026] In a possible implementation, the field moisture capacity influencing factors include a sand content, a clay content, a silt content, an organic matter content and a bulk density.

[0027] Advantages of the present application: the present application proposes a field moisture capacity fusion method based on multiple source soil data, which, on the basis of comprehensively considering the accuracy characteristics of the multiple source soil data, constructs a field moisture capacity fusion factor set based on multiple source soil data based on the trained and existing soil transfer function, determines an optimal fusion scheme, can effectively improve the field moisture capacity estimation accuracy, reduce errors, and solves the technical problem that the existing field moisture capacity estimation method has a large error. BRIEF DESCRIPTION OF DRAWINGS

[0028] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some of the embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor.

[0029] Figure 1 Fig. 1 is a schematic diagram of a soil data statistical downscaling algorithm according to an embodiment of the present application;

[0030] Figure 2 Fig. 2 is a schematic diagram of a multi-source field water capacity fusion framework according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the following will combine the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0032] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0033] According to the first aspect of the present application, a multi-source soil data based field water capacity fusion calculation method is first provided, including the following steps:

[0034] Step 1, data collection and processing: obtaining multiple sets of data source soil data and measured field water capacity data measured by the soil moisture station in the target area. The multi-source soil data in the present application refers to multiple sets of soil data from different data sources.

[0035] Specifically, the target area in the present embodiment is the national region. The measured field water capacity measurement method in step 1 is collected by the cutting ring method. After returning to the laboratory, the soil sample is artificially wetted to saturation, and then the soil moisture content at the time when the soil sample is naturally dehydrated to the error of the measured mass before and after is less than the set value is measured, which is the measured field water capacity. After calculating the measured field water capacity, the reliability of the measured data is ensured by analyzing the physical properties of the sampled soil, comparing with the existing field water capacity, and verifying with historical drought data and reference results. The detailed content of step 1 is as follows:

[0036] Step 1.1, measured field water capacity data

[0037] The measured field moisture content data collected from the soil moisture station of the Ministry of Water Resources Information Center are distributed in 2388 provinces and cities across the country.

[0038] The measured field moisture content measurement method is the cutting ring method. After returning to the laboratory, the soil sample is artificially moistened to saturation, and then the soil sample is naturally dehydrated to an error of less than 0.5g in mass measurement, and the soil moisture content at this time is determined as the field moisture content.

[0039] For all sampling points, the field moisture content of 0-10 cm, 10-20 cm, and 20-40 cm depth is measured, and the single station average field moisture content is calculated by weighted average of soil layer depth, as shown in equation (1):

[0040]

[0041] wherein represents the field moisture content of the soil sample, θ i represents the field moisture content measured by the i depth soil sample, h i represents the thickness of the soil sample, which is 10 cm, 10 cm, and 20 cm respectively, and H represents the total thickness of the soil layer, which is 40 cm.

[0042] After calculating the measured field moisture content, the measured field moisture content is compared with the existing field moisture content from the analysis of the physical properties of the sampled soil, and the historical drought data and reference results are verified to ensure the reliability of the measured data.

[0043] Step 1.2, obtain multi-source soil data

[0044] Multi-source soil data refers to multiple sets of data source soil data. In this embodiment, three sets of data source soil data are used, which are:

[0045] 1) FAO soil data

[0046] The United Nations Food and Agriculture Organization (FAO) developed a global 10km grid soil database based on the world soil emission potential library. The data includes soil sand content, clay content, silt content, organic matter content, soil bulk density and other soil property data, divided into upper and lower layers, 0-30 cm as the upper layer, and 30-100 cm as the lower layer.

[0047] 2) BNU soil data

[0048] Dai et al. (2018) from Beijing Normal University (BNU) derived soil physical and chemical properties at 1 km grid cells across China, including sand content, clay content, silt content, organic matter content, soil bulk density, etc., with 7 layers in vertical direction, depths are 0-4.5 cm, 4.5-9.1 cm, 9.1-16.6 cm, 16.6-28.9 cm, 28.9-49.3 cm, 49.3-82.9 cm, 82.9-138.3 cm, respectively, based on 1:1,000,000 China soil map and 8595 soil profiles from the Second National Soil Survey. Inconsistent soil profiles were reclassified into appropriate soil types, considering sample spacing, size, and soil classification, using a polygon-linkage method.

[0049] 3) SG soil data

[0050] The Global Soil Information Dataset (SG) is constructed by the International Soil Reference and Information Centre based on global soil profile and environmental variable data. The dataset is driven by globally distributed soil profile data, related to environmental covariates, and global soil property and classification maps drawn using geostatistics and machine learning algorithms. In 2015, the dataset was based on geostatistics to produce a 1.0 version of the grid cell soil dataset, providing 1 km soil physical properties (such as sand content, clay content, silt content, bulk density, coarse particle content, etc.), soil biochemical properties (such as SOM, soil pH, cation exchange capacity), and soil classification maps. In 2016, the 2.0 version based on machine learning was released, with a resolution further improved from 1 km to 250 m, and the data accuracy was further improved. The dataset divides the soil into 6 layers from top to bottom, with layer depths of 0-5 cm, 5-15 cm, 15-30 cm, 30-60 cm, 60-100 cm, and 100-200 cm, respectively.

[0051] Step 1.3, statistical downscaling

[0052] Since the spatial resolutions of FAO soil data, BNU soil data, and SG soil data are different, they are 10 km, 1 km, and 250 m, respectively, in order to realize the uniformity of spatial resolution, the method of statistical downscaling is used for spatial scale matching.

[0053] For FAO soil data and BNU soil data, the statistical downscaling procedure is as follows, taking the BNU soil data downscaling as an example:

[0054] The soil grid cells in the 250 m SG soil data are upscaled to 1 km soil grid cells using the grid cell average method.

[0055] The nearest neighbor method is used to match the soil grid cells in the 1 km SG soil data with the soil grid cells in the 1 km BNU soil data.

[0056] The grid cell data in the 1 km BNU soil data is down-scaled to the 250 m grid cell data by using the following formula (2).

[0057] BNU_250m = SG_250m x BNU_1km / SG_1km (2)

[0058] wherein BNU_250m represents the down-scaled BNU soil data, SG_250m represents the original 250 m resolution SG soil data, BNU_1km represents the original 1 km BNU soil data, and SG_1km represents the up-scaled SG soil data.

[0059] Figure 1 The color in the figure from dark to light represents the soil attribute data value from large to small, and the soil attribute data can be sand content, clay content, silt content, organic matter, bulk density, and other physical and chemical attribute parameters.

[0060] The above formula is based only on the SG soil data, changes the distribution of the soil attribute elements in the BNU_1km grid cell, and ensures that the mean value is unchanged.

[0061] The down-scaled BNU soil data can be obtained by substituting the sand content, clay content, organic matter content, and other elements into the formula, respectively.

[0062] For the 10 km FAO soil data, the above down-scaling method is also used for scale conversion.

[0063] In other embodiments, soil data of the same spatial resolution can also be selected, and in this case, scale conversion is not required.

[0064] Step 2, analysis of field moisture capacity influencing factors: based on the nearest neighbor method, the measured field moisture capacity data is matched with the grid cell soil data in the data source soil data, the soil parameters in the data source soil data are extracted, the principal component analysis method is used for sensitivity analysis of the soil parameters in the data source soil data, and the soil parameters affecting the field moisture capacity are selected as the field moisture capacity influencing factors. The specific method is as follows:

[0065] The size of the field moisture capacity is affected by many factors. On the one hand, it is the soil physical and chemical properties, including sand content, clay content, silt content, and other factors. On the other hand, the organic matter content and soil structure of the soil are also affected by multi-year precipitation, air temperature, vegetation replacement, and other factors, which indirectly act on the field moisture capacity.

[0066] Based on the nearest neighbor method, the measured field moisture content data and the grid cell soil data are matched, the relevant physical, chemical and biological parameters in the soil data are extracted as the influencing factors of the field moisture content, and the principal component analysis method is used for sensitivity analysis.

[0067] On the one hand, based on the principal component analysis results, the influence of different soil parameters on the field moisture content is analyzed by correlating the measured field moisture content with multiple soil characteristic factors at multiple sites, so as to avoid information redundancy in modeling.

[0068] On the other hand, the rationality of selecting soil parameters is verified by analyzing the soil characteristic influencing factors in the existing soil transfer function.

[0069] Finally, through analysis, five soil parameters including sand content, clay content, silt content, organic matter content (SOM) and bulk density (BD) are selected as soil influencing factors for field moisture content simulation.

[0070] Step 3: According to the selected field moisture content influencing factors, a training soil transfer function is trained to represent the relationship between soil parameters and measured field moisture content.

[0071] Specifically, for each set of data source soil data, the nearest neighbor method is used to match the site observed field moisture content and the data source soil data. Based on multivariate regression, neural network, support vector machine and other methods, a training soil transfer function is trained.

[0072] Based on the simulation accuracy, model stability and complexity of the training soil transfer function, the optimal training soil transfer function corresponding to each set of data source soil data is selected.

[0073] Finally, three sets of optimal training soil transfer function models are determined as the training soil transfer functions corresponding to the three sets of soil data.

[0074] It is worth noting that different regions in the country can be regressed separately to obtain different parameterization schemes.

[0075] According to the existing soil transfer functions established by scholars, the simulated field moisture content data for each grid cell soil data in each set of data source soil data is constructed. That is, the known existing soil transfer functions are selected to construct the simulated field moisture content data corresponding to each grid cell soil data in the data source soil data.

[0076] Specifically, in addition to the training soil transfer function, the existing soil transfer functions established at home and abroad are also selected, and the simulated field moisture content data for each set of data source soil data is calculated respectively.

[0077] For all the 2388 observed sites, the nearest neighbor method was used to match the soil data from the three data sources, and three sets of soil data for all the sites were obtained.

[0078] Then, based on the three sets of soil data for each site, 21 sets of simulated field capacity data were calculated using the seven sets of existing soil transfer functions (PTFs) that had been established.

[0079] The established existing soil transfer functions are as follows:

[0080] FC_PTFs1 = (0.043 + 0.004 x clay) / (0.471 + 0.00411 x clay) (3)

[0081]

[0082] FC_PTFs4 = 0.2081 + 0.0045 x clay + 0.0013 x silt - 0.0595 x BD (6)

[0083] FC_PTFs5 = 0.1183 + 0.0096 x clay - 0.00008 x clay 2 (7)

[0084] FC_PTFs6 = 0.04046 + 0.00426 x silt + 0.00404 x clay (8)

[0085]

[0086] Wherein, FC represents field capacity, sand is sand content, clay is clay content, silt is silt content, SOM is organic matter content, and BD is soil bulk density.

[0087] In addition to the three sets of simulated field capacity data subsets constructed by the trained soil transfer functions, 24 sets of simulated field capacity data subsets are formed.

[0088] In other embodiments, the number of existing soil transfer functions can be increased or decreased as needed, such as selecting any two, any three, etc. of the above existing soil transfer functions.

[0089] Step 4: Selecting soil grid cells with observed field capacity data from the target area as observed grid cells, and matching corresponding soil data for each of the observed grid cells as observed grid cell soil data according to different data source soil data.

[0090] calculating a simulated field capacity set for each set of data source soil data: the simulated field capacity set includes a subset of field capacity data corresponding to each soil transfer function in step 3, and each simulated field capacity subset includes the simulated field capacity of each measured grid cell calculated by the set of data source soil data and the soil transfer function corresponding to the simulated field capacity subset.

[0091] The simulated field capacity data subset is composed of the simulated field capacity data of each measured grid cell soil data of a set of data source soil data and a soil transfer function; the simulated field capacity set is composed of multiple simulated field capacity data subsets of each soil transfer function in step 3 for each measured grid cell soil data of a set of data source soil data.

[0092] Specifically, in this embodiment, there are three sets of data source soil data, eight soil transfer functions, including seven existing soil transfer functions and one trained soil transfer function, and 24 simulated field capacity data subsets. There are eight simulated field capacity data subsets in the simulated field capacity set corresponding to a set of data source soil data.

[0093] According to the source of soil data, the simulated field capacity data subsets are FC_FAO, FC_BNU and FC_SG, so there are three groups of simulated field capacity sets, which are:

[0094] FC_FAO group: FC_FAO_1, FC_FAO_2……FC_FAO_8;

[0095] FC_BNU group: FC_BNU_1, FC_BNU_2……FC_BNU_8;

[0096] FC_SG group: FC_SG_1, FC_SG_2……FC_SG_8.

[0097] Step 5, selecting a set of simulated field capacity data subsets from each simulated field capacity set to form a group of fusion factors, and fusing the group of fusion factors: taking the simulated field capacity data and the determination coefficient of the measured field capacity as the fusion weight value, and weighting and fusing each fusion factor according to its fusion weight value to obtain the fused field capacity.

[0098] Specifically, the following steps are included:

[0099] Step 5.1, fusion factor statistical index analysis: taking the measured field capacity as the reference data, the statistical indexes of each simulated field capacity data subset are calculated respectively, including standard deviation, mean value, variation coefficient, determination coefficient with measured value, root mean square deviation, and error characteristics of each simulated field capacity data subset are analyzed.

[0100] Step 5.2, fusion method based on mathematical statistics: based on the statistical index characteristics of each set of data, a fusion method based on mathematical statistics is proposed, and the fusion is carried out according to the weight value of the certainty coefficient of the fusion factor and the measured field moisture content. The greater the certainty coefficient, the greater the weight obtained. In addition, the scale calibration is carried out on the fusion factor, so as to ensure that there is no systematic deviation in the overall fusion result. The fusion formula is shown in formula (10) and (11):

[0101]

[0102] Wherein, FC ensemble is the fusion field moisture content, σ obs is the standard deviation of the measured field moisture content, σ i is the standard deviation of the i-th fusion factor in the group of fusion factors, is the certainty coefficient of the i-th fusion factor in the group of fusion factors and the measured field moisture content, FC i is the i-th fusion factor in the group of fusion factors, is the mean value of the field moisture content of the i-th fusion factor in the group of fusion factors, is the mean value of the measured field moisture content, and △FC is the deviation correction term of the actual value.

[0103] Step 6, repeat step 5, and calculate the fusion field moisture content under all fusion factor combination conditions.

[0104] Specifically, according to the algorithm loop, the fusion field moisture contents under all FC_FAO, FC_BNU and FC_SG combination conditions are calculated.

[0105] Step 7, according to the measured field moisture content of the measured grid unit, the accuracy of each fusion field moisture content corresponding to the measured grid unit is verified, and according to the accuracy verification result, the optimal fusion scheme is determined; according to the optimal fusion scheme, the fusion field moisture contents of other soil grid units except the measured grid unit are calculated, and the fusion field moisture content distribution of the target area is formed.

[0106] Select FC_FAO, FC_BNU and FC_SG based on FAO, BNU and SG respectively, and estimate the error based on the measured field moisture content data FC_ measured.

[0107] It should be noted that in the fusion process, only 2388 soil grid units containing the measured station are calculated, and the fusion result verification is obtained by comparing the measured field moisture content with the fusion field moisture content distribution based on the 2388 soil grid units. After determining the optimal fusion scheme, the fusion scheme is used to fuse the field moisture content of all soil grid units in the country.

[0108] For example, in the first cycle combination process, FC_FAO_1, FC_BNU_1 and FC_SG_1 are selected as fusion factors to participate in fusion calculation. Through step 5.1 statistical calculation result, the standard deviation, determination coefficient, mean value and other statistical indexes corresponding to FC_FAO_1, FC_BNU_1 and FC_SG_1 are obtained respectively, and are substituted into formula (10) and formula (11), so as to calculate the fusion field water capacity data of the measured grid unit under the fusion scheme.

[0109] For the fusion field water capacity under this fusion scheme, based on the measured field water capacity data, the accuracy of the fusion field water capacity of the measured grid unit is verified, and the data is analyzed and evaluated.

[0110] Further start the next cycle combination, such as selecting FC_FAO_2, FC_BNU_1 and FC_SG_1 as fusion factors for fusion calculation. And according to the previous scheme, data fusion and accuracy evaluation are carried out.

[0111] According to the optimal accuracy verification result, the final optimal fusion scheme is determined, and the final fusion field water capacity distribution is determined.

[0112] After determining the optimal fusion scheme, for all grid units in China, the fusion scheme is used to select fusion factors and determine fusion weights, so as to calculate the 250m grid unit scale fusion field water capacity distribution in China.

[0113] It should be noted that the fusion scheme can also be further divided according to different regions. The application briefly describes the core principle and steps to improve robustness, and only one set of fusion scheme is used in the whole country.

[0114] The advantage of the above fusion algorithm is that it can reduce the uncertainty caused by the error of individual regional soil data, and also can reduce the situation that the applicability of single soil transfer function is insufficient. In addition, by introducing mean and standard deviation for data processing, the systematic deviation of data is reduced.

[0115] The field water capacity fusion calculation method based on multi-source soil data also includes a special region processing method: for the desert areas such as Taklimakan in the south of Xinjiang, the sand content in desert areas is much higher than that in other areas, and there are few measured field water capacity sites in desert areas. Based on the above fusion method, the construction of field water capacity will cause systematic deviation of field water capacity in desert areas, and it is unreasonable to use the fusion formula established by samples of other soil types to estimate the field water capacity in desert areas.

[0116] Therefore, the desert geographical distribution is extracted by using Maryland land cover data, the most suitable soil transfer function for simulating the field capacity of the desert is selected by referring to the existing research, and the field capacity distribution map of the desert is calculated based on the multi-source soil data set.

[0117] In addition, the null value processing method is used for the main lake, river and other water body covered areas.

[0118] Step 7 also includes field capacity fusion construction and analysis: the fusion factor and fusion weight are determined according to the optimal fusion scheme, the field capacity distribution is constructed based on three sets of soil data sources nationwide, special processing methods are used for the desert and water body parts, and finally the 250m fused field capacity distribution is constructed.

[0119] Table 1 shows the precision improvement of the fused field capacity of the present application compared with the existing field capacity products. It can be seen that the correlation coefficient between the fused field capacity constructed by the present application and the measured field capacity is 0.61, which has a great improvement compared with the simulation precision of the existing research, and the average precision is improved by more than 40%. Compared with the lowest precision of Tomasella field capacity, the precision is improved by 74%. Compared with the highest precision of BNU field capacity, the precision is improved by 33%.

[0120] Table 1 Comparison of precision of fused field capacity and existing field capacity products

[0121]

[0122] The above only describes the preferred embodiments of the present application and is not used to limit the present application. For those skilled in the art, the present application can have various changes and variations. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for calculating field water holding capacity fusion based on multi-source soil data, characterized in that, The method comprises the following steps: Step 1, obtaining multiple sets of soil data sources and measured field moisture capacity data measured by soil moisture stations in the target area; Step 2, matching the measured field moisture capacity data with the grid cell soil data in any one set or more than two sets of soil data sources based on the nearest neighbor pixel method, extracting the soil parameters in the soil data sources, and performing sensitivity analysis on the soil parameters in the soil data sources by principal component analysis, and selecting the soil parameters affecting the field moisture capacity as the field moisture capacity influencing factors; Step 3, training and establishing a training soil transfer function representing the relationship between each soil parameter and the measured field moisture capacity according to the selected field moisture capacity influencing factors; in addition to the training soil transfer function, the corresponding simulated field moisture capacity data is constructed for each grid cell soil data in each set of soil data source according to the known existing soil transfer function; Step 4, selecting soil grid cells with measured field moisture capacity data from the target area as measured grid cells, and matching the corresponding soil data as measured grid cell soil data for each of the measured grid cells according to different data source soil data; For each set of data source soil data, a simulated field moisture capacity set is calculated: the simulated field moisture capacity set includes a subset of field moisture capacity data corresponding to each soil transfer function in step 3, and each simulated field moisture capacity subset includes the simulated field moisture capacity of each measured grid cell calculated from the set of data source soil data and the soil transfer function corresponding to the simulated field moisture capacity subset; Step 5, selecting a set of simulated field moisture capacity data subsets from each simulated field moisture capacity set to form a fusion factor group, and fusing the fusion factor group: taking the determination coefficient of the simulated field moisture capacity data and the measured field moisture capacity as the fusion weight value, and weighting and fusing each fusion factor according to the respective fusion weight value to obtain the fused field moisture capacity; Step 6, repeating step 5 to calculate the fused field moisture capacity under all fusion factor combination conditions; Step 7, verifying the accuracy of each fused field moisture capacity corresponding to the measured grid cell according to the measured field moisture capacity of the measured grid cell, determining the optimal fusion scheme, and calculating the fused field moisture capacity of other soil grid cells except the measured grid cells to form the fused field moisture capacity distribution of the target area.

2. The method of claim 1, wherein the method further comprises: The formula for fusing each fusion factor in step 5 is: where FC ensemble is the fusion field capacity, σ obs is the standard deviation of the measured field capacity, σ i is the standard deviation of the i-th fusion factor in the set of fusion factors, is the determination coefficient of the i-th fusion factor in the set of fusion factors and the measured field capacity, FC i is the i-th fusion factor in the set of fusion factors, is the mean of the field capacity of the i-th fusion factor in the set of fusion factors, is the mean of the measured field capacity.

3. The method according to claim 1 or 2, wherein, In step 1, after measuring the measured field moisture capacity, the measured field moisture capacity is compared with the existing field moisture capacity from the analysis of the physical properties of the sampled soil, and the historical drought data and reference results are verified.

4. The method of claim 1 or 2, wherein the method is characterized by, The multiple sets of data source soil data include FAO soil data, BNU soil data, and SG soil data.

5. The method of claim 1 or 2, wherein the method is characterized by, Step 1 includes converting the spatial resolution of the multiple sets of data source soil data to a uniform size through a scale conversion algorithm.

6. The method of claim 1 or 2, wherein the method is characterized by, Step 2 includes: based on the principal component analysis result, by correlation analysis on the measured field water holding capacity and a plurality of site soil characteristic factors, the influence degree of different soil parameters on the field water holding capacity is analyzed; by analyzing the soil characteristic influence factors in the known existing soil transfer function, the rationality of the soil parameters as the field water holding capacity influence factors is verified.

7. The method of claim 1 or 2, wherein the method further comprises: In step 3, the soil transfer function is trained based on at least one of a multiple regression method, a neural network method and a support vector machine method.

8. The method of claim 1 or 2, wherein the method is characterized by, Step 7 includes: taking the measured field water holding capacity as reference data, statistical indexes of each simulated field water holding capacity data subset are calculated respectively, the statistical indexes include standard deviation, mean value, variation coefficient, determination coefficient and root mean square deviation with the measured field water holding capacity, and error characteristics of the simulated field water holding capacity data subset are analyzed.

9. The method of claim 1 or 2, wherein the method further comprises: For desert areas, a soil transfer function suitable for simulating desert field water holding capacity is selected, and the field water holding capacity distribution map of the desert area is calculated based on a plurality of sets of soil data sources.

10. The method of claim 1 or 2, wherein the method further comprises: The field water holding capacity influence factors include silt content, clay content, silt content, organic matter content and bulk density.

Images