A method for selecting a climate model and a hydrological model in runoff prediction

Abstract

The application provides a method for selecting a climate model and a hydrological model in runoff estimation, comprising the following steps: obtaining reference annual runoff, annual precipitation and annual average temperature in a historical period for a target basin; selecting a plurality of climate models; using the climate models to output precipitation, maximum temperature and minimum temperature climate elements; inputting the climate elements into a plurality of hydrological models; obtaining runoff daily sequences of the plurality of hydrological models corresponding to the climate elements output by the plurality of climate models in the historical period of the target basin; obtaining simulated annual runoff; obtaining simulated coordinate points of runoff sensitivity of the simulated annual runoff relative to the annual precipitation and the annual average temperature in each hydrological model; comparing the simulated coordinate points with reference sensitivity coordinate points; finding simulated coordinate points with the smallest distance and corresponding climate models and hydrological models. The application has important significance for reducing the problem of great uncertainty in runoff estimation by selecting appropriate models of the climate model and the hydrological model after analyzing runoff sensitivity.

Description

Technical Field

[0001] This invention relates to the technical field of meteorological hydrology, and in particular to a method for selecting the optimal climate model and hydrological model in runoff prediction. Background Technology

[0002] Global warming alters the global hydrological cycle, causing changes in the spatiotemporal distribution of water resources. The "top-down" analysis method, which uses precipitation and temperature variables from climate models, corrects for biases, and then inputs them into hydrological models to predict watershed runoff, has become an important tool for the scientific community to predict future drought and flood changes in watersheds.

[0003] Existing global geographic information systems consist of multiple modules, none of which provide analytical solutions for calculating various global climate elements. Different models published by different institutions employ different parameterization schemes to address these uncertainties, resulting in varying simulation effects for different variables across different regions. Furthermore, different hydrological models exhibit varying simulation results within the same watershed, leading to significant uncertainties in watershed runoff prediction. Precipitation, minimum and maximum temperatures are the primary input variables used in runoff prediction.

[0004] Therefore, by conducting sensitivity analysis to objectively evaluate the precipitation, temperature, and hydrological models output by climate models in runoff prediction, it is possible to better select appropriate climate and hydrological models for runoff prediction in different regions. However, the current selection of climate model prediction results for runoff prediction usually adopts the averaging of multiple model results, which requires a lot of computational data, and the existing evaluation system for runoff prediction effectiveness has a large degree of uncertainty. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide a method for selecting optimal climate and hydrological models in runoff prediction. By analyzing runoff sensitivity and selecting appropriate climate and hydrological models, the present invention is of great significance in reducing the significant uncertainty in runoff prediction.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0007] A method for selecting optimal climate and hydrological models in runoff prediction includes the following steps:

[0008] For the target watershed, obtain reference annual runoff, annual precipitation, and annual average temperature for historical periods;

[0009] Based on actual needs, several climate models are selected, and each climate model outputs climate elements to multiple hydrological models in the target watershed. The daily runoff series of multiple hydrological models corresponding to the climate elements output by multiple climate models in the target watershed during historical periods are obtained. Among them, the climate elements include precipitation, maximum temperature and minimum temperature.

[0010] The simulated annual runoff is obtained from the daily runoff series, and the simulated coordinate points of the runoff sensitivity relative to annual precipitation and annual average temperature are obtained from multiple hydrological models corresponding to various climate models.

[0011] By comparing the runoff sensitivity coordinates of each hydrological model with the reference sensitivity coordinates, the simulated coordinates that are closest to the reference sensitivity coordinates are found, and the climate model and hydrological model corresponding to the simulated coordinates that are closest are identified.

[0012] Furthermore, the climate elements, after being downscaled and biased, are input into multiple hydrological models of the target watershed.

[0013] Furthermore, the downscaling method is as follows:

[0014] Inverse distance interpolation is used to interpolate low-resolution climate features to target high-resolution climate features.

[0015] Furthermore, the DBC bias correction method was used to correct precipitation, minimum temperature, and maximum temperature.

[0016] Furthermore, in each hydrological model, the slope of the linear function fitting of simulated annual runoff with respect to annual precipitation is used as the abscissa, and the slope of the linear function fitting of simulated annual runoff with respect to annual average temperature is used as the ordinate, serving as the simulated coordinate points for the runoff sensitivity of simulated annual runoff with respect to annual precipitation and annual average temperature.

[0017] Furthermore, the relative sensitivity coefficient S of runoff to climate factors C The calculation formula is:

[0018]

[0019] In the formula, ΔY represents the change in simulated annual runoff, and ΔC represents the change in meteorological elements during historical periods, which are precipitation or temperature.

[0020] Furthermore, the simulated coordinate point that has the smallest straight-line distance from the reference sensitive coordinate point is the simulated coordinate point that has the smallest distance from the reference sensitive coordinate point.

[0021] Furthermore, the climate model was selected from CMIP5 or CMIP6 published by the IPCC.

[0022] Furthermore, the reference annual runoff is the average daily runoff at the outlet station of the target basin, the annual precipitation is the average annual precipitation within the target basin, and the annual average temperature is the average daily temperature within the target basin.

[0023] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0024] This invention proposes a method for optimizing climate and hydrological models in runoff prediction. The combination of runoff prediction is determined by the simulation results of the corresponding optimized climate and hydrological models. Sensitivity analysis is used to obtain the simulation effect of the closest reference runoff relative to precipitation and temperature, thereby determining the optimal climate and hydrological models for runoff simulation. This overcomes the problems of cumbersome processes and high uncertainties in runoff prediction, which are usually performed by averaging multiple climate and hydrological models. Compared with existing calculation methods, this method requires less computation, saves computational resources, and reduces the uncertainty of future runoff prediction in the watershed by optimizing the climate and hydrological models. Attached Figure Description

[0025] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and are intended to explain the invention, but do not constitute an undue limitation thereof. In the drawings:

[0026] Figure 1 This is a flowchart of a method for selecting the optimal climate model and hydrological model in runoff prediction according to the present invention.

[0027] Figure 2 In this embodiment of the invention, the models in CMIP5 and CMIP6 of the Huaihe River Huaibin control basin runoff sensitivity are numbered sequentially. Among them, 1-21 and 23-43 correspond to 1-21 in CMIP5 and CMIP6, respectively, and 22 and 44 are the average results of all models in CMIP5 and CMIP6 (wherein, the selected hydrological models are GR4J, HMET, SIMHYD and XAJ, respectively).

[0028] Figure 3 In this embodiment of the invention, the models in the runoff sensitivity CMIP5 and CMIP6 of the Yangtze River Hankou Control Basin are numbered sequentially. Among them, 1-21 and 23-43 correspond to 1-21 in CMIP5 and CMIP6, respectively, and 22 and 44 are the average results of all models in CMIP5 and CMIP6 (wherein, the selected hydrological models are GR4J, HMET, SIMHYD and XAJ, respectively).

[0029] Figure 4In this embodiment of the invention, the models in the Pearl River Wuzhou control basin runoff sensitivity CMIP5 and CMIP6 are numbered sequentially. Among them, 1-21 and 23-43 correspond to 1-21 in CMIP5 and CMIP6, respectively, and 22 and 44 are the average results of all models in CMIP5 and CMIP6 (wherein, the selected hydrological models are GR4J, HMET, SIMHYD and XAJ, respectively).

[0030] Figure 5 This is the relative sensitivity of annual runoff to annual precipitation in the Huaihe River-Huaibin control basin in this embodiment of the invention (Note: The first four rows of the first column correspond to the four hydrological models of the preferred climate model in CMIP5; CMIP5-A is the average of all GCMs in CMIP5; the next four rows correspond to the four hydrological models of the preferred climate model in CMIP6; CMIP6-A is the average of all GCMs in CMIP6; 'Observations' are observation data.)

[0031] Figure 6 This is the relative sensitivity of annual runoff to annual precipitation in the Yangtze River Hankou control basin in this embodiment of the invention (Note: The first 4 rows of the first column correspond to the 4 hydrological models of the preferred climate model in CMIP5; CMIP5-A is the average of all GCMs in CMIP5; the next 4 rows correspond to the 4 hydrological models of the preferred climate model in CMIP6; CMIP6-A is the average of all GCMs in CMIP6; 'Observations' are observation data.)

[0032] Figure 7 This is the relative sensitivity of annual runoff to annual precipitation in the Pearl River Wuzhou control basin in this embodiment of the invention (Note: The first 4 rows of the first column correspond to the 4 hydrological models of the preferred climate model in CMIP5; CMIP5-A is the average of all GCMs in CMIP5; the next 4 rows correspond to the 4 hydrological models of the preferred climate model in CMIP6; CMIP6-A is the average of all GCMs in CMIP6; 'Observations' are observation data.) Detailed Implementation

[0033] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0034] In related technologies, current methods for selecting climate model prediction results in runoff forecasting typically involve averaging multiple model results, which requires a significant amount of computational data. Furthermore, existing evaluation systems for runoff forecasting effectiveness exhibit considerable uncertainty. This invention proposes a method for optimizing climate and hydrological models in runoff forecasting. This method enables the selection of appropriate climate and hydrological models based on runoff sensitivity analysis, which is of great significance in reducing the significant uncertainty in runoff forecasting.

[0035] This invention provides a method for selecting optimal climate and hydrological models in runoff prediction, such as... Figure 1 As shown, it includes the following steps:

[0036] Step S1: For the target watershed, obtain the reference annual runoff, annual precipitation, and annual average temperature for historical periods;

[0037] Step S2: Based on actual needs, select several climate models and use the climate elements output by each climate model to multiple hydrological models in the target watershed. Obtain the daily runoff series of multiple hydrological models corresponding to the climate elements output by multiple climate models in the historical period of the target watershed. Among them, the climate elements include precipitation, maximum temperature and minimum temperature.

[0038] Step S3: Obtain the simulated annual runoff based on the daily runoff series, and acquire the simulated coordinate points of the runoff sensitivity of the simulated annual runoff relative to annual precipitation and annual average temperature in multiple hydrological models corresponding to each climate model.

[0039] Step S4: Compare the runoff sensitivity coordinates of each hydrological model with the reference sensitivity coordinates, find the simulated coordinates that are closest to the reference sensitivity coordinates, and find the climate model and hydrological model corresponding to the simulated coordinates that are closest.

[0040] The runoff prediction method of this invention is determined by the simulation results of corresponding preferred climate models and hydrological models. Sensitivity analysis is used to obtain the simulation effect of the closest reference runoff relative to precipitation and temperature, thus identifying the optimal climate model and hydrological model for runoff simulation. This overcomes the problems of cumbersome processes and high uncertainties in runoff prediction typically achieved by averaging multiple climate models and hydrological models. Compared with existing calculation methods, this method requires less computation, saves computational resources, and reduces the uncertainty of future runoff prediction in the watershed by selecting the preferred climate model and hydrological model.

[0041] Furthermore, current methods for optimizing climate models typically evaluate their simulation performance using multiple spatiotemporal indicators based on variables such as precipitation, maximum or minimum temperatures output by the climate model, before inputting these indicators into a hydrological model. This invention directly uses runoff sensitivity coefficients to analyze the performance of climate and hydrological models in runoff simulation, optimizing the climate or hydrological model based on its simulation effectiveness. This approach is more targeted, and the selected climate and hydrological models effectively reduce uncertainties in runoff simulation.

[0042] In step S1 of this invention, the climate model is selected from CMIP5 or CMIP6 published by IPCC.

[0043] The calculation process for the daily runoff sequence is as follows:

[0044] Step S201: Based on the simulation effect of climate models on precipitation, maximum and minimum temperatures in the watershed, select an appropriate number of climate models and use the multi-year daily series data of precipitation, maximum and minimum temperature variables output by the climate models.

[0045] Step S202: For the target watershed, calibrate the selected hydrological models using daily series data of precipitation, maximum and minimum temperatures and runoff from historical periods.

[0046] Step S203: Input the precipitation, maximum and minimum temperature data output by the selected climate model into multiple hydrological models to obtain the daily runoff series data calculated for the historical period of the target watershed.

[0047] The reference annual runoff is the average daily runoff at the outlet station of the target basin, the annual precipitation is the average annual precipitation within the target basin, and the annual average temperature is the average daily temperature within the target basin.

[0048] In step S2 of this invention, the daily runoff is averaged using daily runoff data to obtain the simulated annual runoff. Annual precipitation and average annual temperature are calculated by averaging the daily precipitation and temperature variables output from various climate models.

[0049] In step S2 of this invention, climate elements are input into multiple hydrological models of the target watershed after being downscaled and biased.

[0050] Specifically, the downscaling method is as follows:

[0051] Inverse distance interpolation is used to interpolate low-resolution climate features to target high-resolution climate features.

[0052] Precipitation correction using the DBC bias correction method includes the following steps:

[0053] For a given month out of 12 months, determine a precipitation threshold such that the probability of the climate model output variable exceeding the precipitation threshold in that month is equal to the probability of the reference variable in that month.

[0054] The precipitation threshold is used in the model prediction data for the corresponding future months to correct the probability of precipitation in the future period;

[0055] Calculate the values ​​of the 100th quantile of the empirical distribution of accumulated rainfall days in the simulated precipitation series and the measured precipitation series for the baseline period, and calculate the ratio of the simulated and measured quantile values ​​at these 100th quantiles as the precipitation correction factor.

[0056] The distribution of precipitation in the simulated precipitation sequence is corrected using a precipitation correction factor;

[0057] The precipitation correction factor is applied to future time periods to correct the precipitation distribution in the simulated precipitation sequence for future time periods.

[0058] Correcting the highest or lowest temperature using the DBC deviation correction method includes the following steps:

[0059] For a given month out of 12 months, first determine the maximum or minimum temperature threshold, such that the probability of the climate model output variable for that month exceeding the maximum or minimum temperature threshold is equal to the probability of the reference variable for that month.

[0060] The highest or lowest temperature threshold is used in the model prediction data for the corresponding month in the future to correct the probability of the highest or lowest temperature occurring in the future period;

[0061] The difference between the quantile of the simulated temperature series for a given month and the quantile of the measured temperature series for each month is used as the maximum or minimum temperature factor.

[0062] The distribution of temperature in the simulated temperature sequence is corrected using the highest or lowest temperature correction factor;

[0063] The highest or lowest temperature correction factor is applied to future time periods to correct the temperature distribution in the simulated temperature series for future time periods.

[0064] In this invention, in step S3, in each hydrological model, the slope of the linear function fitting of simulated annual runoff with respect to annual precipitation is used as the abscissa, and the slope of the linear function fitting of simulated annual runoff with respect to annual average temperature is used as the ordinate, serving as the simulated coordinate points for the runoff sensitivity of simulated annual runoff with respect to annual precipitation and annual average temperature. The simulated coordinate points in each hydrological model of multiple climate models are compared with the reference sensitivity coordinate points to observe the differences between the simulated annual runoff sensitivity with respect to annual precipitation and annual average temperature and the reference sensitivity coordinate points for different climate models and hydrological models.

[0065] Obtain the relative sensitivity coefficient S of runoff to climate factors. C The relative sensitivity coefficient S of runoff to climate factors is used to evaluate the effectiveness of different climate models and hydrological models in simulating runoff. The closer the simulation sensitivity is to the reference data, the better the simulation effect. C The calculation formula is:

[0066]

[0067] In the formula, ΔY represents the change in simulated annual runoff, and ΔC represents the change in meteorological elements during historical periods, which are precipitation or temperature.

[0068] In step S4 of this invention, the simulated coordinate point that has the smallest straight-line distance from the reference sensitive coordinate point is the simulated coordinate point that has the smallest distance from the reference sensitive coordinate point.

[0069] The method for selecting the optimal climate model and hydrological model in runoff prediction proposed in this invention is validated, including:

[0070] Step S5: Using the output variables of the climate model with optimized runoff sensitivity, plot a scatter plot of the relative sensitivity analysis of annual runoff to annual precipitation calculated by different hydrological models, with the slope and sensitivity of the linear function fitting of simulated annual runoff to annual precipitation as the abscissa and the slope and sensitivity of the linear function fitting of annual average temperature as the ordinate. Mark the sensitivity point values ​​of simulated annual runoff to annual precipitation and annual average temperature calculated by reference data, and compare them with the reference sensitivity coordinate points. When comparing, mainly observe the distance between the calculated point and the reference point.

[0071] Step S6: Using runoff sensitivity, optimize the climate and hydrological models for the watershed and verify the optimization results of step S4.

[0072] In a specific embodiment of this invention, the following illustrates the interpolation of 42 climate models from CMIP5 and CMIP6 to a resolution of 0.5°×0.5°, using the Yangtze River Hankou control basin, the Huaihe River Huaibin control basin, and the Pearl River Wuzhou control basin as examples. Table 1 shows the climate models. Due to limitations in basin data, the three typical humid basins selected here are: the Yangtze River Hankou control basin (catchment area of ​​stations is [missing data], daily flow data from hydrological stations spans from January 1, 1961 to December 31, 2008, a total of 48 years); the Huaihe River Huaibin control basin (catchment area of ​​stations is [missing data], daily flow data from hydrological stations spans from January 1, 1961 to December 31, 2003, a total of 43 years); and the Pearl River Wuzhou control basin (catchment area of ​​stations is [missing data], daily flow data from hydrological stations spans from January 1, 1961 to December 31, 2008, a total of 48 years).

[0073] Due to limitations in parameter acquisition and hydrological and soil information gathering during hydrological model calculations, generalized hydrological models are currently the most widely used. The hydrological model selected for calculation is primarily based on the characteristics of the watershed and the applicability of the model. Since this example focuses on three watersheds in a humid region, the selected hydrological models are also applicable to humid regions and are four commonly used hydrological models in climate change hydrological response calculations. By optimizing the watershed's hydrological model through runoff sensitivity analysis and comparing the performance of different generalized models, the following models were selected: XAJ, HMETS, GR4J, and SIMHYD.

[0074] The method for selecting the optimal climate model and hydrological model in runoff prediction provided by this invention, such as... Figures 2-4 As shown, taking the Huaihe River Huaibin control basin, the Yangtze River Hankou control basin, and the Pearl River Wuzhou control basin as examples, the daily runoff series of the corresponding hydrological models XAJ, HMETS, GR4J, and SIMHYD for climate elements output by climate models 1-44 are obtained in the historical period of the target basin. The simulated annual runoff is obtained based on the daily runoff series. The simulated coordinate points of the runoff sensitivity of the simulated annual runoff relative to the annual precipitation and annual average temperature in each hydrological model are obtained. The runoff sensitivity coordinate points corresponding to each hydrological model are compared with the reference sensitivity coordinate points. The simulated coordinate point with the smallest distance from the reference sensitivity coordinate point is found, and the climate model and hydrological model corresponding to the simulated coordinate point with the smallest distance are found.

[0075] like Figures 5-7 As shown, based on step S5, using the climate model output variables selected by the runoff sensitivity optimization, a scatter plot of the relative sensitivity analysis of annual runoff to annual precipitation calculated by different hydrological models is made, with the slope and sensitivity of the linear function fitting of simulated annual runoff to annual precipitation as the abscissa and the slope and sensitivity of the linear function fitting of annual average temperature as the ordinate. The sensitivity point values ​​of simulated annual runoff to annual precipitation and annual average temperature calculated by reference data are marked and compared with the reference sensitivity coordinate points. When comparing, the distance between the calculated point and the reference point is mainly observed.

[0076] Table 2 shows the relative sensitivity of the optimal climate model to annual precipitation for each watershed and the average results of the ensemble of 21 models. According to Table 2, the reference values ​​of the multi-model averaged climate model have a larger error than the optimal model for the watershed. For both CMIP5 and CMIP6, the sensitivity of annual runoff simulated by the optimal climate model and hydrological model to annual precipitation or annual temperature is better than that of the averaged simulation of all models, and is closer to the reference data, reducing the uncertainty in the runoff simulation process.

[0077] After optimization, as shown in Table 3, the best-performing GCM in the Huaihe River Basin controlled by the Huaihe River Commission is 'IPSL-CM5A-MR', with the corresponding optimal hydrological model being GR4J; the best-performing GCM in the Hankou Basin controlled by the Yangtze River is 'NorESM1-M', with the corresponding optimal hydrological model being XAJ; the best-performing GCM in the Wuzhou Basin controlled by the Pearl River is 'CanESM2', with the corresponding optimal hydrological model being GR4J; the best-performing GCM in the Huaihe River Basin controlled by the Huaihe River Commission is 'EC-Earth3', with the corresponding optimal hydrological model being HMETS; the best-performing GCM in the Hankou Basin controlled by the Yangtze River is 'MRI-ESM2-0', with the corresponding optimal hydrological model being HMETS; and the best-performing GCM in the Wuzhou Basin controlled by the Pearl River is 'INM-CM5-0', with the corresponding optimal hydrological model being XAJ.

[0078] Table 1 GCM Information in CMIP5 and CMIP6

[0079]

[0080] Table 2. Relative Sensitivity of Annual Runoff to Annual Precipitation

[0081]

[0082] Table 3. Optimal climate model and hydrological model for annual runoff relative to annual precipitation and average temperature.

[0083]

[0084] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for selecting optimal climate and hydrological models in runoff prediction, characterized in that, Includes the following steps: For the target watershed, obtain reference annual runoff, annual precipitation, and annual average temperature for historical periods; Based on actual needs, several climate models are selected, and the climate elements output by each climate model are used to multiple hydrological models in the target watershed. The daily runoff series of multiple hydrological models corresponding to the climate elements output by multiple climate models in the target watershed during historical periods are obtained. Among them, the climate elements include precipitation, maximum temperature and minimum temperature. The simulated annual runoff is obtained from the daily runoff series, and the simulated coordinate points of the runoff sensitivity relative to annual precipitation and annual average temperature are obtained from multiple hydrological models corresponding to various climate models. The coordinates of runoff sensitivity corresponding to each hydrological model are compared with the reference sensitivity coordinates. The simulated coordinates that are closest to the reference sensitivity coordinates are found, and the climate model and hydrological model corresponding to the simulated coordinates that are closest are also found. The simulated coordinates that are closest to the reference sensitivity coordinates are the simulated coordinates that are closest to the reference sensitivity coordinates. In each hydrological model, the slope of the linear function fitting of simulated annual runoff with respect to annual precipitation is used as the abscissa, and the slope of the linear function fitting of simulated annual runoff with respect to annual average temperature is used as the ordinate, serving as the simulated coordinate points for the runoff sensitivity of simulated annual runoff with respect to annual precipitation and annual average temperature. The relative sensitivity of runoff to climate factors The calculation formula is: In the formula, To simulate the variation in annual runoff, This refers to the changes in meteorological elements over a historical period, where the meteorological elements are precipitation or temperature.

2. The method for selecting the optimal climate model and hydrological model in runoff prediction according to claim 1, characterized in that: Climate elements, after being downscaled and biased, are input into multiple hydrological models of the target watershed.

3. The method for selecting the optimal climate model and hydrological model in runoff prediction according to claim 2, characterized in that, The downscaling method is to use inverse distance interpolation to interpolate low-resolution climate elements to target high-resolution climate elements.

4. The method for selecting the optimal climate model and hydrological model in runoff prediction according to claim 2, characterized in that: The DBC bias correction method was used to correct precipitation, minimum temperature, and maximum temperature.

5. The method for selecting the optimal climate model and hydrological model in runoff prediction according to claim 1, characterized in that: The climate models were selected from CMIP5 or CMIP6 published by the IPCC.

6. The method for selecting the optimal climate model and hydrological model in runoff prediction according to claim 1, characterized in that: The reference annual runoff is the average daily runoff at the outlet station of the target basin, the annual precipitation is the average annual precipitation within the target basin, and the annual average temperature is the average daily temperature within the target basin.

Images