A method for predicting individual floods in data - scarce regions based on underlying surface characteristics and improved Xin'anjiang model

Abstract

The present invention discloses a method for predicting individual floods in data - scarce areas based on underlying surface characteristics and an improved Xin'anjiang model. The method of the present invention directly calculates runoff based on the underlying surface characteristics of the basin, can reproduce the hydrological process of the basin to a greater extent, overcomes the problem of inconsistent basin characteristics during the parameter transplantation process, and effectively reduces numerous runoff parameters involved in the original Xin'anjiang model. This method can solve the problem that it is difficult to optimize hydrological model parameters due to the lack of long - term hydrological data in data - scarce areas. Moreover, the method is simple, has low requirements for hydrological data, and has remarkable effects, which is an important means to solve flood prediction in data - scarce areas. The method of the present invention is simple and effective, has high practical value, and has high application prospects in operational forecasting.

Description

Technical Field

[0001] The present invention relates to a method for predicting individual floods in data - scarce areas based on underlying surface characteristics and an improved Xin'anjiang model, belonging to the field of flood prediction, and is mainly used for the flood prediction and early warning work carried out by water conservancy departments. Background Art

[0002] Due to the impacts of climate change and urban expansion, flood disasters occur more and more frequently, seriously threatening the national economy and the safety of people's lives and property. Studying new flood prediction methods and improving flood prediction capabilities and accuracy are of great significance for deeply understanding the characteristics of basin floods and alleviating flood disasters.

[0003] Mountainous areas and slopes are the main sources of flood disasters. In mountainous basins, due to lush vegetation, short and rapid - flowing rivers, floods rise and fall suddenly, which are extremely likely to trigger mountain floods; dry sloping farmlands are the main crop - growing areas in the transitional zone between mountainous areas and urban plains or hilly areas. They have weak water storage capacity. Coupled with the relatively steep terrain, heavy rain can cause floods, and the floods rise and recede rapidly, seriously threatening the lives and property of residents in the downstream plains or river valleys. However, such areas usually lack long - term hydrological measurement sequences, making it difficult to carry out parameter optimization work for hydrological models. Therefore, studying flood prediction methods for data - scarce or data - deficient areas based on basin underlying surface characteristics can minimize the need for measured hydrological data and provide relatively reliable flood simulation and prediction, which has important guiding significance for flood control and disaster reduction in such areas.

[0004] At present, for flood forecasting in data - scarce areas, domestic and foreign scholars mostly adopt the method of parameter regionalization, that is, using a certain method to infer the model parameters of data - scarce or data - lacking areas by using the model parameters of data - rich areas. The research methods of parameter regionalization mainly include the parameter transplantation method and the regression method. The basis of the parameter transplantation method is that the model parameters of two river basins with similar hydro - meteorological and underlying surface characteristics are also similar. Therefore, the parameters of the data - rich river basin (reference basin) can be transplanted to the data - lacking river basin. The parameter transplantation method commonly uses the proximity method and the attribute similarity method. Although the proximity method has achieved high transplantation accuracy in some studies, it is generally believed that geographical proximity does not completely equal the similarity of rainfall - runoff processes, especially in areas with significant differences in climate, terrain, and geology. The attribute similarity method believes that the parameters of river basins with similar physical and climate attributes are transplantable. The regression method refers to establishing a multiple regression equation representing the relationship between the model parameters and basin attributes of the data - rich river basin, so as to infer the model parameters by using the basin attributes of the data - lacking river basin. The hypothesis premise of the regression method is that there is a certain correlation between the parameters and physical attributes in the data - rich area and the data - lacking area. However, in fact, no model has a clear parameter set that can establish a definite and good relationship with the hydrological process. Therefore, there are certain difficulties in the existing flood forecasting methods for data - scarce areas. Especially when the basin attributes of the data - scarce area are quite different from those of the data - rich area, there are significant problems in parameter transplantation. It is of great significance to study new flood forecasting technologies based on the underlying surface characteristics of data - scarce areas. Summary of the Invention

[0005] To solve the deficiencies of the above - mentioned existing technologies, the purpose of the present invention is to provide a method for forecasting individual - event floods in data - scarce areas based on underlying surface characteristics and an improved Xin'anjiang model, so as to realize flood simulation in data - scarce areas, improve the forecasting accuracy, and provide scientific support for flood control and disaster reduction in river basins.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions:

[0007] A method for forecasting individual - event floods in data - scarce areas based on underlying surface characteristics and an improved Xin'anjiang model, comprising the following steps:

[0008] Step 1: Collect and organize precipitation, underlying surface, and geographical information data of the research area; the underlying surface data includes land - use type raster data and soil - type raster data;

[0009] Step 2: Extract the river basin water system according to the geographical information data, and analyze and delimit sub - basins by comprehensively considering the river basin water system and land - use type;

[0010] Step 3: Summarize the hydrological soil groups corresponding to the soil type raster data according to soil characteristics, and determine the weighted runoff curve number CN2 of each sub-basin under medium soil moisture conditions by integrating land use types and hydrological soil groups;

[0011] Step 4: Divide the antecedent moisture condition of the soil according to the cumulative precipitation in the previous 5 days, and calculate the weighted runoff curve numbers CN1 and CN3 of each sub-basin under dry and wet soil conditions respectively.

[0012] Step 5: Calculate the runoff depth of the sub-basin according to the weighted runoff curve number and precipitation of each sub-basin;

[0013] Step 6: Replace the runoff generation process of the original Xin'anjiang model with the runoff depth calculation process in Step 5 to obtain an improved Xin'anjiang model, perform overland flow concentration and river network flow concentration on the calculated runoff depth, and calculate the flood discharge at the outlet section of the basin.

[0014] In the above technical solution, further, the precipitation data in Step 1 is publicly fused data or reanalysis data, mainly including the China regional fusion data CLDASv2 publicly shared by the China National Meteorological Administration or the global reanalysis data ERA-5 provided by ECMWF. The soil type raster data is the national data of 227 subclasses with a 30-meter resolution in the China Soil Database, and the land use type raster data is the global data with a 10-meter resolution released by the European Space Agency in 2020; the geographic information data, i.e., digital elevation data, is the publicly available data of ASTER GDEM with a 30M resolution from the Geospatial Data Cloud.

[0015] Further, the precipitation data in Step 1 is a raster dataset, which needs to be further sorted into the areal precipitation data of each sub-basin, and calculated with reference to the Thiessen polygon method. The areal precipitation calculation includes the following steps:

[0016] (1) Input the basin contour map and rainfall rasters (P1, P2, … P i …, P N ) in ArcGIS software, and divide the basin into N squares according to the raster size, where S squares overlap with the sub-basin in area;

[0017] (2) Calculate the area coincidence ratio w i of the square where the i-th rainfall raster P i，j is located in the S squares and the sub-basin, and use it as the contribution value of the rainfall raster P i to the areal precipitation of the sub-basin;

[0018] (3) The formula for the areal precipitation of the sub-basin is:

[0019] Further, for the sub - watershed division in step 2, first, sub - watershed division is carried out for the tributaries directly flowing into the main stream. It is necessary to ensure that the areas of each sub - watershed are not very different (the area difference between any two sub - watersheds does not exceed 50%), and the sub - watersheds cannot cross the main stream to ensure the correctness of flood confluence. Finally, land - use type data is overlaid, and sub - watershed subdivision is carried out again in areas where the underlying surface changes significantly (such as changing from a forest land underlying surface to an urban underlying surface) (such as dividing in the transition zone from forest land to urban area), to ensure that the overall characteristics of the underlying surface within the same sub - watershed are the same or similar (such as mainly forest land or mainly urban area).

[0020] Further, in step 3, according to soil characteristics, the soil - type raster data is summarized and classified into its corresponding hydrologic soil groups. The basis for summarization and classification is the soil texture and infiltration rate. The divided hydrologic soil groups include four groups: A (permeable), B (relatively permeable), C (relatively impermeable), and D (impermeable).

[0021] Further, in step 3, the weighted runoff curve number CN is determined using the raster calculator of ArcGIS based on the raster runoff curve number cn. The resolution of the raster runoff curve number is the same as that of the raster land - use type data. Since the raster resolution of the land - use type is higher than that of the soil - type raster data, the resolution of the raster runoff curve number is set to be the same as that of the raster land - use type to ensure that the soil type and land - use type within each raster correspond one by one, and a unique raster runoff curve number is calculated. The specific calculation method of the weighted runoff curve number CN2 for each sub - watershed under medium soil moisture conditions is as follows: By looking up the runoff curve number manual published by the Natural Resources Conservation Service (NRCS) of the United States, the runoff curve number within each land - use type raster is obtained, and then the weighted runoff curve number for each sub - watershed is calculated through the raster calculator of ArcGIS and used as the runoff curve number under the condition of medium - moist soil state, denoted as CN2.

[0022] Further, the basis for the division of the initial soil moisture state AMC in step 4 is as follows in the table:

[0023] Table 1 Classification table of soil moisture state levels

[0024]

[0025] Among them, AMC I is for dry soil conditions, AMC II is for medium soil moisture conditions, and AMC III is for wet soil moisture conditions.

[0026] Further, the calculation method of the weighted runoff curve numbers CN1 and CN3 for each sub - basin under dry and wet soil conditions is obtained by using the raster calculator of ArcGIS based on the raster runoff curve number; among them, the calculation formulas for the raster runoff curve numbers cn1 and cn3 under dry and wet soil conditions are:

[0027] cn1 = cn2 / (2.334 - 0.01334cn2)

[0028] cn3 = cn2 / (0.4036 + 0.0059cn2)

[0029] Among them, cn2 is the raster runoff curve number under medium - humidity soil conditions.

[0030] Further, the runoff depth R(T) of the sub - basin in step 5 is the difference between the cumulative runoff depth R C (T) at time T and the cumulative runoff depth R C (T - 1) at the previous time: R(T)=R C (T)-R C (T - 1); the cumulative runoff depth R C (T) is calculated based on the cumulative rainfall TP(T) and the potential maximum retention S:

[0031]

[0032] Among them, S is the potential maximum retention S (mm) before the start of runoff, The value of I in CNI is 1, 2, 3; λ is the initial loss rate, and its value range is 0 - 0.3. According to the dry - wet zoning in China, the value in the arid area north of the Qinling - Huaihe River is relatively large, with a range of about 0.2 - 0.3; the value range in the humid area south of it is 0.1 - 0.2, and in the extremely humid area along the southeast coast of China, it can take 0 - 0.15.

[0033] Further, in step 6, for the obtained runoff depth, slope - area concentration is carried out. The calculation method of the flow rate QT(T) of slope - area concentration at time T is the linear reservoir method, and the calculation formula is:

[0034] QT(T)=CI×QT(T - 1)+(1 - CI)×R(T)×U

[0035] Among them, CI is the subsoil flow retreat coefficient of the improved Xin'anjiang model, QT(T-1) is the flow of slope confluence at time T-1; R(T) is the runoff depth at time T; U is the unit conversion coefficient, and the calculation formula is U=A / (3.6*24), and A is the sub-basin area. The original Xin'anjiang model has a water source calculation, which divides the basin runoff into surface runoff, subsoil flow and underground runoff. The surface runoff does not pass through the slope report and directly enters the river network, which can be regarded as CI=0. Subsoil flow and underground runoff enter the river network through linear reservoirs. Since the present invention does not have a water source calculation process, it is necessary to improve the Xin'anjiang model, and the runoff depth R(T) calculated according to the underlying surface characteristics is used as the total runoff depth to enter the river network through linear reservoirs. Generally speaking, the richer the subsoil flow, the closer the parameter CI value is to 0.9, and the value is generally 0.5-0.65 in humid areas. For flood events, the total runoff is mainly composed of surface runoff and subsoil flow, and the surface runoff recession coefficient is equivalent to 0. Therefore, the present invention recommends that the CI value be smaller than the parameter of the original Xin'anjiang model, and the value range is about 0.3 to 0.4.

[0036] Furthermore, in step 6, the flow QT(T) after the slope confluence is calculated by using a hysteresis algorithm, and the flow Q(T) after the river network confluence at time T is calculated as follows:

[0037] Q(T)=CS×Q(T-1)+(1-CS)×QT(TL)

[0038] Among them, QT(TL) is the flow after the slope confluence at time TL; Q(T-1) is the flow after the river network confluence at time T-1; the parameters CS and L represent the river network recession coefficient and the river network confluence hysteresis time respectively. The value of L is related to the area. According to experience, the basin area in mountainous areas is less than 50km 2 The hysteresis time L is 0; the area is 200~300km 2 The value is 2 to 3 hours. Generally speaking, the greater the average height difference of the basin, the more concentrated the confluence, which is manifested as a smaller basin storage capacity and a smaller corresponding CS. The larger the basin area, the higher the corresponding river network level, the stronger the flood storage capacity, and the larger the corresponding CS. The empirical formula for determining CS is:

[0039]

[0040] in is the average height difference of the basin, which can be obtained by subtracting the elevation of the basin outlet point from the average elevation of the basin; A is the basin area; is the average slope, which can be expressed as:

[0041]

[0042] Where ΔH is the height interval; l0, l1, l2..., lm is the contour length.

[0043] By adopting the above technical means, the beneficial effects of the present invention are as follows:

[0044] (1) The method of the present invention for forecasting individual floods in data - scarce areas based on underlying surface characteristics and improved Xin'anjiang model can solve the problem that it is difficult to optimize hydrological model parameters due to the lack of long - term hydrological data in data - scarce areas. The method of the present invention is simple, has low requirements for hydrological data, and has remarkable effects, and is an important means to solve flood forecasting in data - scarce areas.

[0045] (2) The present invention directly calculates runoff based on the underlying surface characteristics of the basin, can reproduce the hydrological process of the basin to a greater extent, overcomes the problem of inconsistent basin characteristics during parameter transplantation, and effectively reduces the numerous runoff parameters involved in the original Xin'anjiang model. The method of the present invention is simple and effective, has high practical value, and has high application prospects in operational forecasting. Description of the Drawings

[0046] Figure 1 is the overall implementation flowchart of a method for forecasting individual floods in data - scarce areas based on underlying surface characteristics and improved Xin'anjiang model of the present invention;

[0047] Figure 2 are the land use types (a), soil types (b), rainfall distribution (c) and elevation water system map (d) of the study area;

[0048] Figure 3 are the sub - basin division of the study area and its distribution relationship with the water system (a), land use and soil types (b);

[0049] Figure 4 are the distribution maps of grid cn2 (a) and weighted CN2 (b) values;

[0050] Figure 5 is the calculation of sub - basin areal rainfall based on the principle of Thiessen polygon;

[0051] Figure 6 is the contour map of sub - basin 11 when ΔH takes the value of 50m;

[0052] Figure 7 is the comparison of the simulation results of the present invention and the calibrated traditional Xin'anjiang model for the 2019 Lekima flood. Detailed Embodiments

[0053] The present invention will be further described below in conjunction with the drawings and specific embodiments. Taking the Datian Plain in Taizhou City, Zhejiang Province as an example for specific implementation, the following steps are included:

[0054] Step 1: Collection and collation of basin meteorological data, land use, soil type, and basin geographic information data. The meteorological data uses the hourly integrated precipitation data of CLDASv2 provided by the China Meteorological Data Center, with a spatial resolution of 0.06°×0.06°. The land use data is the global land use (WorldCover) data released by the European Space Agency in 2020, with a spatial resolution of 10 meters. The soil type data is the national soil type raster data of 227 subclasses with a resolution of 30 meters in the China Soil Database. The basin geographic information data mainly consists of the ASTER GDEM digital elevation data with a resolution of 30M provided by the Geospatial Data Cloud. The specific results are as Figure 2 shown. According to the overlapping area ratio of the rainfall grid and each sub-basin, the weighted average precipitation can be calculated. The Thiessen polygon method can be referred to calculate the areal rainfall of each sub-basin ( Figure 5 ), and the specific calculation method is as follows:

[0055] (1) Input the basin contour map and rainfall rasters (P1, P2,...P i ..., P N ) into the ArcGIS software. The basin is divided into N squares with the grid position as the center and the grid size as the side length. Among them, S squares overlap with the calculated sub-basin in area;

[0056] (2) Calculate the area overlapping ratio w i of the i-th rainfall raster P i，j in the S squares with the sub-basin as the contribution value of the rainfall raster P i to the areal rainfall of this sub-basin;

[0057] (3) The calculation formula for the areal rainfall of this sub-basin is:

[0058] Step 2: Sub-basin division of the basin

[0059] Use ArcGIS to extract and process the geographic information data of the target basin to obtain the basin contour and the basin catchment area. Extract the digital elevation data within the basin catchment area, use ArcGIS to further obtain the water system distribution, divide the sub-basins according to the water system distribution and river network classification, and further fine-tune and subdivide on the basis of land use, as Figure 3 shown. It can be seen from Figure 3 that the finally divided sub-basins not only maintain the integrity of the water system, but also the land use types within each sub-basin are relatively similar.

[0060] Step 3: Determine the weighted runoff curve number CN2 of each sub-basin under medium soil moisture conditions

[0061] First, extract the raster data of land use and soil type within the basin and output it in ASCII format. Secondly, classify the soil types into hydrological soil groups according to Table 1 (the basis for summarization is the texture and infiltration rate of the soil, and the soil types are divided into four categories: A (permeable), B (relatively permeable), C (relatively impermeable), and D (impermeable)). Then, calculate the runoff curve number cn2 corresponding to each land use raster under medium soil moisture conditions according to Table 2. Finally, use the raster calculator in ArcGIS to calculate the weighted runoff curve number of each sub-basin, as Figure 4 shown. Figure 4 shows the distribution of the raster runoff curve number cn2 values calculated by the present invention under medium moisture soil conditions and the weighted runoff curve number CN2 values of each sub-basin calculated by using the raster calculator. The sub-basin numbers are as Figure 4 shown. It is not difficult to see that the CN2 value shows a gradually increasing trend from forest land to urban area. Among them, for sub-basins 4-7, due to the relatively high proportion of impervious ground, the CN2 value is relatively large.

[0062] Table 1 Soil types and their corresponding hydrological soil group classifications

[0063]

[0064] Table 2 cn2 value table of different soil types and land use types under medium soil moisture conditions

[0065]

[0066]

[0067] Step 4, determine the weighted runoff curve numbers CN1 and CN3 of each sub-basin under dry and wet soil conditions

[0068] Calculate the hourly cumulative rainfall, and judge the previous moisture state of the soil according to the cumulative rainfall in the previous 5 days. The basis for judging the moisture state is Table 3. Since the flood season in China is mainly in summer and autumn, the division criteria of the main crop growth period are mainly referred to. Then, according to the raster runoff curve number cn2 of each sub-basin under medium moisture soil conditions in Step 3, calculate the raster runoff curve number cn1 and cn3 values of each sub-basin under dry and wet soil conditions respectively, and use the raster calculator in ArcGIS to calculate the weighted runoff curve numbers CN1 and CN3 of each sub-basin under dry and wet soil conditions.

[0069] The calculation formulas for the raster runoff curve number cn1 and cn3 values of each sub-basin under dry and wet soil conditions are as follows:

[0070] AMCⅠ: cn1 = cn2 / (2.334 - 0.01334cn2)

[0071] AMC Ⅲ: cn3 = cn2 / (0.4036 + 0.0059cn2)

[0072] Table 3 Soil Moisture State AMC Grade Classification Table

[0073]

[0074] Step 5, calculate the runoff depth of the sub - watershed based on the underlying surface characteristics

[0075] Calculate the hourly cumulative rainfall TP(T) according to the hourly areal average rainfall P of each sub - watershed:

[0076]

[0077] Calculate the potential maximum retention S (mm) before the start of runoff, Cumulative runoff depth R C (T) (mm) is calculated by the formula:

[0078]

[0079] λ is the initial loss rate, with a value range of 0 - 0.3. Considering that the watershed is located in the humid area of the eastern coast of China, the empirical value of λ = 0.15. The hourly runoff depth is R(T) = R C (T) - R c (T - 1).

[0080] Step 6, use the improved Xin'anjiang model for overland flow concentration and river network flow concentration to calculate the flood discharge at the outlet section of the watershed.

[0081] First, determine the overland flow concentration parameter, the subsurface flow recession coefficient CI. According to the characteristics of the hydrological process in humid areas, during floods, subsurface flow and surface runoff are abundant, and the proportion of subsurface flow is smaller during large floods. Considering that the flood events of Typhoon Lekima are super - large floods and mainly surface runoff, the value of CI is taken as 0.3. Input the hourly runoff depth R(T) for overland flow concentration to obtain the runoff volume QT(T) after overland flow concentration in the sub - watershed:

[0082] QT(T) = CI × QT(T - 1)+(1 - CI) × R(T) × U

[0083] U is the unit conversion coefficient, and the calculation formula is U = A / (3.6 * 24), where A is the area of each sub - watershed.

[0084] Then, conduct river network flow concentration on the above results. The calculation formula for the flow rate Q(T) after river network flow concentration at time T is:

[0085] Q(T) = CS × Q(T - 1)+(1 - CS) × QT(T - L)

[0086] It is necessary to determine the river network recession coefficient CS of this sub-basin, and this coefficient is calculated through the empirical formula:

[0087]

[0088] where is the average elevation difference of the basin, which is the difference between the average elevation of the basin and the elevation of the basin outlet point; A is the basin area; is the average slope, which can be expressed as:

[0089]

[0090] In the formula, ΔH is the contour interval; l0, l1, l2,..., l m are the contour lengths. The calculation example is Figure 6 as shown. Taking sub-basin 11 as an example, the contour lines with a contour interval of 50 meters are drawn for calculating the average slope of the basin.

[0091] Finally, according to the size of the basin area, confirm the river network confluence lag time L. When the basin area is less than 50 km 2 the lag time L is taken as 0; when the area is 50 - 125 km 2 it is taken as 1 hour, when it is 125 - 200 km 2 it is taken as 2, when it is greater than 200 and less than 300 km 2 it is 3 hours, and so on. Finally, superimpose the river network confluence results of all sub-basins to obtain the flood discharge at the basin outlet section. The specific results are Figure 7 as shown. Figure 7 This is the comparison between the flood processes of sub-basins 14 and 15 calculated by the present invention and the calculation results of the optimized Xin'anjiang model. It is not difficult to see that the calculation results of the present invention are relatively close to the optimized Xin'anjiang model based on hydrological data, and have good flood simulation ability.

Claims

1. A method for predicting individual floods in data - scarce regions based on underlying surface characteristics and improved Xin'anjiang model, characterized in that, It includes the following steps: Step 1: Collect and organize precipitation, underlying surface, and geographic information data of the study area; the underlying surface data includes land use type raster data and soil type raster data; Step 2: Extract the watershed water system based on the geographic information data, and analyze and delimit sub-watersheds by comprehensively considering the watershed water system and land use type; Step 3: Summarize the corresponding hydrological soil groups of the soil type raster data according to soil characteristics, and determine the weighted runoff curve number of each sub-watershed under medium soil moisture conditions by comprehensively considering land use type and hydrological soil groups; Step 4: Divide the initial moisture state of the soil according to the cumulative precipitation in the previous 5 days, and calculate the weighted runoff curve number of each sub-watershed under dry and wet soil conditions respectively; Step 5: Calculate the runoff depth of the sub-watershed according to the weighted runoff curve number and precipitation of each sub-watershed; Step 6: Replace the runoff generation process of the original Xin'anjiang model with the runoff depth calculation process in Step 5 to obtain an improved Xin'anjiang model, and perform overland flow concentration and river network flow concentration on the calculated runoff depth to calculate the flood discharge at the outlet section of the watershed.

2. The method for forecasting individual floods in data - scarce regions based on underlying surface characteristics and improved Xin'anjiang model according to claim 1, wherein, The data involved in Step 1 are all publicly available datasets in China or other countries and regions in the world. Among them, the precipitation data is the China Regional Integrated Data CLDASv2 publicly shared by the China National Meteorological Administration or the global reanalysis data ERA-5 provided by ECMWF; the soil type raster data is the national data of 227 subclasses with a 30-meter resolution in the China Soil Database, and the land use type raster data is the global data with a 10-meter resolution released by the European Space Agency in 2020; the geographic information data is the publicly available data of ASTER GDEM with a 30M resolution from the Geospatial Data Cloud.

3. The method for forecasting individual floods in data - scarce regions based on underlying surface characteristics and improved Xin'anjiang model as claimed in claim 1, wherein, The precipitation data in Step 1 is a raster dataset and needs to be further organized into the areal precipitation data of each sub-watershed; the areal precipitation is calculated using the Thiessen polygon method. The specific method is as follows: (1) Input the basin contour map and rainfall rasters (P1, P2, … P i …, P N ) in the ArcGIS software. Divide the basin into N squares according to the raster size, where S squares have area overlap with the sub-basin; (2) Calculate the overlapping ratio w of the square where the i-th rainfall grid P in S squares is located with the area of the sub-basin i as the contribution value of the rainfall grid P i,j to the areal rainfall of the sub-basin; i ​ (3) The calculation formula for the areal rainfall of this sub-watershed is as follows:

4. The method for forecasting individual floods in data - scarce regions based on underlying surface characteristics and improved Xin'anjiang model according to claim 1, wherein, In Step 2, the sub-watershed delimitation method is completed based on ArcGIS. By comprehensively considering land use type and river network distribution, it is necessary to ensure that the land use type is the same within the same sub-watershed as much as possible.

5. The method for predicting individual floods in data - scarce regions based on underlying surface characteristics and improved Xin'anjiang model according to claim 1, wherein, In Step 3, summarize the corresponding hydrological soil groups of the soil type raster data according to soil characteristics. The basis for summarization is the texture and infiltration rate of the soil. The divided hydrological soil groups include four groups: A (permeable), B (relatively permeable), C (relatively impermeable), and D (impermeable); The weighted runoff curve number CN2 of each sub-watershed under medium soil moisture conditions is determined using the raster calculator of ArcGIS based on the raster runoff curve number cn2. The resolution of the raster runoff curve number is the same as that of the land use type raster data.

6. The method for predicting the flood events in data - scarce areas based on underlying surface characteristics and improved Xin'anjiang model as claimed in claim 1, wherein In Step 4, the initial moisture state of the soil is divided into three states: dry AMCⅠ, medium wet AMCⅡ, and wet AMCⅢ. The specific basis for division is as follows in the table: Table 1 Classification Table of Soil Moisture State Levels 7. The method for forecasting individual floods in data - scarce regions based on underlying surface characteristics and improved Xin'anjiang model as claimed in claim 6, wherein In step 4, the calculation methods of the weighted runoff curve numbers CN1 and CN3 for each sub-basin under dry and wet soil conditions are obtained by using the raster calculator of ArcGIS based on the raster runoff curve number. Among them, the calculation formulas for the raster runoff curve numbers cn1 and cn3 under dry and wet soil conditions are as follows: cn1 = cn2 / (2.334 - 0.01334cn2) cn3 = cn2 / (0.4036 + 0.0059cn2) Among them, cn2 is the raster runoff curve number under medium humidity soil conditions.

8. The method for forecasting individual floods in data - scarce regions based on underlying surface characteristics and improved Xin'anjiang model as claimed in claim 7, wherein, In step 5, the runoff depth R(T) of the sub-watershed is the cumulative runoff depth R C (T) at time T minus the cumulative runoff depth R C (T - 1) at the previous time: R(T) = R C (T) - R C (T - 1); the cumulative runoff depth R C (T) is calculated based on the cumulative rainfall TP(T) and the potential maximum retention S: where S is the potential maximum retention amount S (mm) before the start of runoff, I in CNI takes values of 1, 2, or 3; λ is the initial loss rate, with a value range of 0 to 0.

3.

9. The method for forecasting the flood events in data - scarce regions based on underlying surface characteristics and improved Xin'anjiang model according to claim 8, wherein, In step 6, the overland flow concentration is carried out on the obtained runoff depth. The calculation method of the overland flow concentration flow rate QT(T) at time T is the linear reservoir method, and the calculation formula is: QT(T) = CI × QT(T - 1) + (1 - CI) × R(T) × U Among them, CI is the interflow recession coefficient of the improved Xin'anjiang model, QT(T - 1) is the overland flow concentration flow rate at time T - 1; R(T) is the runoff depth at time T; U is the unit conversion coefficient, and the calculation formula is U = A / (3.6 * 24), where A is the sub-basin area.

10. The method for forecasting individual floods in data - scarce regions based on underlying surface characteristics and improved Xin'anjiang model as claimed in claim 9, wherein, In step 6, the lag algorithm is used to carry out the river network concentration on the overland flow concentration flow rate QT(T). The calculation formula for the river network concentration flow rate Q(T) at time T is: Q(T) = CS × Q(T - 1) + (1 - CS) × QT(T - L) Among them, Q(T - 1) is the river network concentration flow rate at time T - 1; QT(T - L) is the overland flow concentration flow rate at time T - L; the parameters CS and L respectively represent the river network recession coefficient and the river network concentration lag time; the calculation formula for CS is: wherein is the average elevation difference of the basin, which can be obtained by subtracting the elevation of the basin outlet point from the average elevation of the basin; A is the basin area; is the average slope gradient, which can be expressed as: where ΔH is the contour interval; l0, l1, l2…, l m are the lengths of the contour lines.

Images