Statistical downscaling method based on mountain satellite remote sensing rainfall data

Abstract

The invention discloses a statistical downscaling method based on mountain satellite remote sensing rainfall data, and the method comprises the steps: collecting target region data and auxiliary data, and carrying out the preprocessing of the target region data and the auxiliary data, and obtaining the processed data; constructing an addition and multiplication fusion model, and integrating the processed data based on the addition and multiplication fusion model to obtain integrated data; and performing comparison and error analysis on the integrated data through rainfall actual measurement station data, and performing evaluation and correction according to an error analysis result to obtain a final correction result. According to the method, downscaling research is carried out on mountain satellite precipitation through a geographically weighted regression method, so that the precision of the satellite precipitation data TRMM can be improved to the greatest extent, and an accurate, high-precision and high-resolution precipitation grid data set is obtained; the method plays an important foundation role in scientific researches such as further research on soil moisture content, reservoir regulation and biomass estimation of a mountain system.

Description

Technical Field

[0001] The invention belongs to the field of remote sensing technology, and in particular relates to a statistical downscaling method based on mountain satellite remote sensing precipitation data. Background Art

[0002] Under the background of global warming, mountain systems are sensitive to climate change, and precipitation is the main part of mountain ecosystems, which is easily affected by complex terrain. At present, traditional precipitation observation technology mainly relies on meteorological observation stations to obtain it. High temporal and spatial resolution precipitation raster data is the input data of various hydrological models and ecological models, but mountainous areas often cannot truly reflect the accurate temporal and spatial distribution of precipitation due to the shortcomings of uneven distribution, small number and time discontinuity of meteorological stations. With the development of remote sensing technology, this gap has been filled. Remote sensing technology has the advantages of high temporal and spatial continuity and wide observation range. It is currently widely used in meteorology, hydrology and climate change research. Satellite images can obtain large-scale precipitation data relatively quickly and comprehensively. The current satellite precipitation downscaling method is mainly limited to the vegetation index NDVI and altitude factors for the selection of terrestrial environmental factors. However, the spatiotemporal distribution of precipitation is related to many geographical environmental factors. Existing technologies lack accurate simulation of mountain precipitation remote sensing products. At the same time, whether there are errors between satellite data and measured precipitation data, how to obtain accurate and high spatiotemporal resolution mountain precipitation raster data through multivariate data fusion is of great significance for future regional climate simulation, ecological protection and restoration, and disaster warning. Summary of the invention

[0003] In order to solve the above technical problems, the present invention provides a statistical downscaling method based on mountain satellite remote sensing precipitation data, comprising:

[0004] Collecting target area data and auxiliary data, and preprocessing the target area data and the auxiliary data to obtain processed data;

[0005] An additive and multiplicative fusion model is constructed, and the processed data is integrated based on the additive and multiplicative fusion model to obtain integrated data; and the integrated data is compared and error analyzed through precipitation measurement site data, and evaluation and correction are performed according to the error analysis results to obtain the final correction result.

[0006] Preferably, the process of collecting target area data and auxiliary data includes:

[0007] Set the time span according to the needs, and collect satellite remote sensing precipitation TRMM data, digital elevation model data, meteorological observatory / station precipitation measured data, climate factors and terrain factors within the time span;

[0008] Wherein, the climate factors include but are not limited to temperature, wind speed, and humidity;

[0009] The terrain factors include but are not limited to altitude, slope, and slope direction.

[0010] Preferably, the process of preprocessing the target area data and the auxiliary data includes:

[0011] The TRMM satellite remote sensing precipitation data was subjected to noise reduction, and the climate factors were spatially interpolated and resampled;

[0012] Using ArcGIS platform, the digital elevation model data of the target area is converted into slope and aspect data, and the data is resampled;

[0013] After comparing the satellite remote sensing precipitation TRMM data with the measured precipitation data of the meteorological observatory / station, the satellite remote sensing precipitation TRMM data is downscaled by using the geographically weighted regression method to obtain the processed data.

[0014] Preferably, the process of downscaling the satellite remote sensing precipitation TRMM data by using the geographically weighted regression method comprises:

[0015] In the process of simulating mountain precipitation, the geographically weighted regression method is used to add the climatic factors and topographic factors that affect mountain precipitation to downscale the annual, seasonal and monthly precipitation data in the mountain area, and obtain a high-resolution mountain precipitation raster dataset.

[0016] Preferably, the process of integrating the processed data includes:

[0017] According to the corresponding time points, the precipitation data, longitude and latitude of the meteorological observatories / stations in the target area, the digital elevation model data of the corresponding locations, and the satellite remote sensing precipitation data of the corresponding locations are extracted and integrated.

[0018] Preferably, the process of integrating the processed data based on the addition and multiplication fusion model includes: merging the satellite inversion data and the ground rain gauge data under the addition and multiplication fusion model to obtain high-precision integrated data.

[0019] Preferably, the process of comparing and error analyzing the integrated data using the precipitation measured site data includes:

[0020] Comparing and error analyzing the integrated data through accuracy evaluation indicators;

[0021] The accuracy evaluation indicators include correlation coefficient, relative error BIAS, and root mean square error RMSE.

[0022] Preferably, the process of evaluating and correcting the error analysis results includes:

[0023] The corrected precipitation values ​​are calculated using the geographically weighted regression method using the measured precipitation data of meteorological stations on different time scales, including annual, seasonal and monthly precipitation, satellite remote sensing precipitation TRMM data, digital elevation model data, climate factors and terrain factors. The corrected precipitation values ​​are then compared with the actual precipitation values ​​of the meteorological stations / observatories in the target area to conduct error analysis and evaluate the correction effect.

[0024] Compared with the prior art, the present invention has the following advantages and technical effects:

[0025] The present invention studies the downscaling method of mountain satellite precipitation through the geographically weighted regression model. The downscaling model after adding the natural geographical factors affecting mountain precipitation - "climate factors and terrain factors" can improve the accuracy of satellite precipitation data TRMM to the greatest extent, and obtain an accurate, high-precision and high-resolution precipitation raster dataset, which plays an important basic role in further scientific research on soil moisture conditions, reservoir scheduling and biomass estimation in mountain systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0027] Figure 1 A comparison chart of TRMM data and measured precipitation data in mountainous areas from 2013 to 2015 according to an embodiment of the present invention;

[0028] Figure 2 It is a spatial distribution map of annual precipitation in mountainous areas from 2013 to 2015 according to an embodiment of the present invention;

[0029] Figure 3 It is a spatial distribution map of the accuracy assessment of TRMM data and downscaled data in mountainous areas from 2013 to 2015 according to an embodiment of the present invention;

[0030] Figure 4 A comparison diagram of TRMM data and statistical downscaled precipitation data sets in an embodiment of the present invention;

[0031] Figure 5 It is a bar chart of different downscaling models and TRMM data in mountainous areas from 2013 to 2015 according to an embodiment of the present invention;

[0032] Figure 6 Taylor diagrams of different mountain downscaling models and TRMM data in an embodiment of the present invention;

[0033] Figure 7 The spatial distribution map of the average annual and seasonal precipitation in mountainous areas from 2002 to 2015 according to the embodiment of the present invention;

[0034] Figure 8 This is a spatial distribution map of average monthly precipitation in mountainous areas from 2002 to 2015 according to an embodiment of the present invention;

[0035] Fig. 9 The figure is a schematic diagram of a method flow of an embodiment of the present invention. DETAILED DESCRIPTION

[0036] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0037] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0038] like Figure 1-9 As shown, this embodiment provides a statistical downscaling method based on mountain satellite remote sensing precipitation data, including:

[0039] Collecting target area data and auxiliary data, and preprocessing the target area data and the auxiliary data to obtain processed data;

[0040] Construct an additive and multiplicative fusion model, integrate the processed data based on the additive and multiplicative fusion model to obtain integrated data; compare and analyze the integrated data through precipitation measurement station data, and evaluate and correct the error based on the error analysis results to obtain the final correction result.

[0041] Furthermore, the process of collecting target area data and auxiliary data includes:

[0042] The time span is set according to the needs, and satellite remote sensing precipitation TRMM data, digital elevation model data, meteorological observatory / station precipitation measured data, climate factors and terrain factors are collected within the time span; the target area data time span of this embodiment is 3 years, from 2013 to 2015.

[0043] Among them, climate factors include but are not limited to temperature, wind speed, and humidity;

[0044] Topographic factors include but are not limited to altitude, slope, and aspect.

[0045] Furthermore, the process of preprocessing the target area data and the auxiliary data includes:

[0046] The TRMM satellite remote sensing precipitation data were denoised, and the climate factors were spatially interpolated and resampled to 1 km;

[0047] The ArcGIS platform was used to generate slope and aspect data from the digital elevation model data of the target area, and the data were resampled to 1 km;

[0048] according to Figure 1 The comparison of TRMM data and measured precipitation data in mountainous areas from 2013 to 2015 shows that, compared with the measured precipitation data, the original TRMM data significantly overestimated the measured precipitation data in 2014 and 2015, and underestimated the measured precipitation data in 2013. Through analysis, it can be seen that there are errors in the satellite data to the measured data. Therefore, the annual precipitation in mountainous areas from 2013 to 2015 was downscaled.

[0049] according to Figure 2 From the spatial distribution diagram of mountain TRMM data, measured precipitation data and precipitation raster data downscaled by Geographically Weighted Regression (GWR) from 2013 to 2015, we can see that the spatial distribution of the three types of data is basically consistent. The precipitation range is from 350mm to 1200mm, with high-value areas appearing on the south slope and low-value areas concentrated on the north slope. Figure 2 TRMM, measured precipitation and COR_GWR in the figure represent the original TRMM data, precipitation interpolation data and GWR downscaled data, respectively.

[0050] according to Figure 3 The spatial distribution map of the precision test index of mountain precipitation data shows that the TRMM data shows higher precision in the western part of the southern slope of the mountain and poorer precision in the southern slope. The statistical downscaling results show higher precision in the southern slope of the mountain, but poorer precision in the northeastern part of the mountain. Figure 3 TRMM and COR_GWR represent the original TRMM data and GWR downscaled data, respectively.

[0051] After comparing the satellite remote sensing precipitation TRMM data with the measured precipitation data of meteorological observatories / stations, the satellite remote sensing precipitation TRMM data were downscaled using the geographically weighted regression method to obtain the processed data.

[0052] Furthermore, the process of downscaling satellite remote sensing precipitation TRMM data by using the geographically weighted regression method includes:

[0053] In the process of simulating mountain precipitation, the geographically weighted regression method is used to add the climatic factors and topographic factors that affect mountain precipitation to downscale the annual, seasonal and monthly precipitation data in the mountain area, and obtain a high-resolution mountain precipitation raster dataset.

[0054] Furthermore, the process of integrating the processed data includes:

[0055] According to the corresponding time points, the precipitation data, longitude and latitude, digital elevation model data of the corresponding location, and satellite remote sensing precipitation data of the meteorological stations in the target area are extracted and integrated. As shown in Table 1, the maximum, minimum and average comparison table of TRMM data, measured precipitation data and statistical downscaling precipitation result data, it can be seen from Table 1 that from 2013 to 2015, compared with TRMM data, the maximum, minimum and average values ​​of statistical downscaling data are closer to the measured precipitation data. For example, in 2014, the maximum value of TRMM data was 1020.7 mm, and the measured precipitation data and statistical downscaling data were 947.9 mm and 931.8 mm, respectively. In general, the statistical downscaling method can improve the overestimated precipitation values ​​caused by TRMM data.

[0056] In order to further quantitatively evaluate the accuracy of the downscaling results, three accuracy evaluation indicators were selected as shown in Table 2 to evaluate the accuracy of the TRMM data and statistical downscaled raster datasets in mountainous areas from 2013 to 2015. It can be seen from Table 2 that the statistical downscaling method can significantly improve the accuracy of TRMM data, with the correlation coefficient r increasing from 0.71 to 0.9, the relative error BIAS decreasing from -3.6 to -2.8, and the root mean square error RMSE decreasing from 99.3 to 93.2. Figure 4 The scatter plots shown are from top to bottom and from left to right for TRMM data, measured precipitation data, and statistical downscaled precipitation datasets in 2013, 2014, and 2015. Figure 4 It can be seen that the statistical downscaling dataset results are closer to the 1:1 line, and the TRMM data are relatively scattered. The results show that the raster dataset results obtained by the statistical downscaling method are more accurate. In 2013, statistical downscaling increased the correlation coefficient r from 0.89 to 0.94, especially in 2014 and 2015, statistical downscaling increased the correlation coefficient from 0.56 to 0.77 and from 0.67 to 0.87, respectively.

[0057] Table 1

[0058]

[0059] Table 2

[0060]

[0061] Furthermore, the process of integrating the processed data based on the additive and multiplicative fusion model includes: under the additive and multiplicative fusion model, merging the satellite inversion data and the ground rain gauge data to obtain high-precision integrated data.

[0062] More specifically, as shown in Table 3, according to different models, the precipitation satellite TRMM data is downscaled using the geographically weighted regression method; mountain precipitation is affected by multiple climate factors and terrain factors, so this embodiment designs 6 different downscaling models, as shown in Table 3, obtains the output results of the six models and conducts accuracy evaluation. As shown in Tables 4 and Figure 5 As shown in Table 4, the accuracy evaluation table of six different statistical downscaling models from 2013 to 2015 is shown. Figure 5 This is a bar chart of different statistical downscaling models and TRMM data from 2013 to 2015. From top to bottom and from left to right, they are the correlation coefficient r, relative error BIAS, and root mean square error RMSE. Figure 5 It can be seen that the correlation coefficients of model one to model six are 0.85, 0.85, 0.85, 0.85, 0.86 and 0.86 respectively. It shows that the six models have a strong correlation with the measured precipitation data. The relative errors BIAS are -3.23, -3.27, -3.27, -3.15, -2.93 and -2.77 respectively; the root mean square errors RMSE are 94.05, 93.97, 93.84, 93.78, 93.41 and 93.24 respectively. The index evaluation results all show that model six, which takes into account both topographic factors and climate factors, has the best accuracy, with the highest correlation coefficient (0.86), the lowest RMSE (-2.77) and the lowest BIAS (93.24). The results show that the statistical downscaling method can obtain the highest precision precipitation results by considering not only topographic factors but also climate factors, which can affect precipitation in mountainous areas. The statistical downscaling model that takes not only topography but also climate factors into account can obtain a more accurate precipitation raster dataset.

[0063] In order to better illustrate the advantages and disadvantages of statistical downscaling models, the error comparison between different models is compared. Figure 6 , Figure 6 From top to bottom and from left to right, they are Taylor diagrams of different statistical downscaling models and TRMM data from 2013 to 2015. Taylor diagrams can clearly show the correlation coefficient r, standard error SD and root mean square error RMSE in the same figure. The closer the downscaling result is to the measured data, the better the downscaling accuracy is. Figure 6 It can be seen that point H, i.e., model 6, is closest to the measured precipitation point A, especially in 2014, when the standard error SD of the measured precipitation is 90.0 mm, while the SD of model 6 is 77.3 mm, the correlation coefficient is 0.77, and the RMSE also shows the lowest value. This shows that model 6 is the optimal model for obtaining mountain precipitation raster datasets.

[0064] Table 3

[0065]

[0066] Table 4

[0067]

[0068] Furthermore, the process of comparing and error analyzing the integrated data with the precipitation measurement station data includes:

[0069] Compare and analyze the errors of the integrated data through accuracy evaluation indicators;

[0070] Among them, the accuracy evaluation indicators include correlation coefficient, relative error BIAS, and root mean square error RMSE.

[0071] Furthermore, the process of evaluating and correcting the error analysis results includes:

[0072] The calibrated precipitation values ​​were calculated using the geographically weighted regression method using the annual, seasonal and monthly precipitation data from meteorological stations at different time scales from 2002 to 2015, satellite remote sensing precipitation TRMM data, digital elevation model data, climate factors and terrain factors. The calibrated precipitation values ​​were then compared with the actual precipitation values ​​of the meteorological stations in the target area for error analysis and to evaluate the correction effect.

[0073] like Figure 7 As shown in the figure, the long-term precipitation data from 2002 to 2015 are verified and applied. From top to bottom and from left to right, they are the results of spring, summer, autumn and winter. The average annual precipitation in the mountains over the past 14 years ranges from 565.6 to 872.6 mm, and the average precipitation is 759.7 mm. The average precipitation in spring, summer, autumn and winter in the mountains from 2002 to 2015 is 113.6 to 236.4 mm, 233.5 to 433.2 mm, 123.7 to 275.7 mm and 11.9 to 36.2 mm. The average precipitation in the four seasons is summer (366.1 mm) > autumn (221.8 mm) > spring (146.4 mm) > winter (25.3 mm). Figure 8 This is the average monthly precipitation distribution map of the mountain area in the past 14 years. From top to bottom and from left to right, they are January to December. It can be seen from the figure that the mountain area has the most precipitation in July, with precipitation ranging from 88.7 to 180.6 mm and an average precipitation of 141.9 mm. The precipitation is particularly abundant in the southern slope area.

[0074] In order to further verify the scientificity and accuracy of the mountain precipitation raster data over the past 14 years, the results are verified. The results are shown in Tables 5 and 6. It can be seen from the tables that the mountain precipitation raster dataset from 2002 to 2015 is relatively close to the measured meteorological station data and has high accuracy, indicating that the raster dataset has high credibility and can be used as an input parameter for hydrological models and ecological models.

[0075] Table 5

[0076]

[0077] Table 6

[0078]

[0079] This embodiment takes into account that the temporal and spatial distribution of precipitation is related to many geographical environmental factors. Whether the satellite precipitation data can truly reflect the spatial variation of precipitation under the complex terrain of the Qinling Mountains. The statistical downscaling method of this embodiment uses the geographically weighted regression method (GWR). In the process of simulating mountain precipitation, it considers both climate factors (temperature, wind speed and humidity) and terrain factors (altitude, slope and slope aspect), and corrects the annual, seasonal and monthly scale precipitation data of mountain satellite precipitation, aiming to obtain more accurate and high-resolution (1km) precipitation raster data. It provides accurate parameter preparation for the input data of various hydrological models and ecological models in the mountains; it is of great significance to the research on soil moisture, vegetation growth, land cover / utilization patterns, reservoir scheduling and climate change.

[0080] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A statistical downscaling method based on mountain satellite remote sensing precipitation data, characterized in that: include: Collecting target area data and auxiliary data, and preprocessing the target area data and the auxiliary data to obtain processed data; constructing an addition and multiplication fusion model, and integrating the processed data based on the addition and multiplication fusion model to obtain integrated data; The integrated data is compared and error analyzed using actual precipitation measurement site data, and evaluation and correction are performed based on the error analysis results to obtain the final correction result.

2. The statistical downscaling method based on mountain satellite remote sensing precipitation data according to claim 1 is characterized in that: The process of collecting target area data and auxiliary data includes: Set the time span according to the needs, and collect satellite remote sensing precipitation TRMM data, digital elevation model data, meteorological observatory / station precipitation measured data, climate factors and terrain factors within the time span; Wherein, the climate factors include but are not limited to temperature, wind speed, and humidity; The terrain factors include but are not limited to altitude, slope, and slope direction.

3. The statistical downscaling method based on mountain satellite remote sensing precipitation data according to claim 1 is characterized in that: The process of preprocessing the target area data and the auxiliary data includes: The TRMM satellite remote sensing precipitation data was subjected to noise reduction, and the climate factors were spatially interpolated and resampled; Using ArcGIS platform, the digital elevation model data of the target area is converted into slope and aspect data, and the data is resampled; After comparing the satellite remote sensing precipitation TRMM data with the measured precipitation data of the meteorological observatory / station, the satellite remote sensing precipitation TRMM data is downscaled by using the geographically weighted regression method to obtain the processed data.

4. The statistical downscaling method based on mountain satellite remote sensing precipitation data according to claim 3 is characterized in that: The process of downscaling the satellite remote sensing precipitation TRMM data by using the geographically weighted regression method includes: In the process of simulating mountain precipitation, the geographically weighted regression method is used to add the climatic factors and topographic factors that affect mountain precipitation to downscale the annual, seasonal and monthly precipitation data in the mountain area, and obtain a high-resolution mountain precipitation raster dataset.

5. The statistical downscaling method based on mountain satellite remote sensing precipitation data according to claim 1 is characterized in that: The process of integrating the processed data includes: According to the corresponding time points, the precipitation data, longitude and latitude of the meteorological observatories / stations in the target area, the digital elevation model data of the corresponding locations, and the satellite remote sensing precipitation data of the corresponding locations are extracted and integrated.

6. The statistical downscaling method based on mountain satellite remote sensing precipitation data according to claim 1 is characterized in that: The process of integrating the processed data based on the addition and multiplication fusion model includes: merging the satellite inversion data and the ground rain gauge data under the addition and multiplication fusion model to obtain high-precision integrated data.

7. The statistical downscaling method based on mountain satellite remote sensing precipitation data according to claim 1 is characterized in that: The process of comparing and analyzing the errors of the integrated data with the actual precipitation station data includes: Comparing and error analyzing the integrated data through accuracy evaluation indicators; The accuracy evaluation indicators include correlation coefficient, relative error BIAS, and root mean square error RMSE.

8. The statistical downscaling method based on mountain satellite remote sensing precipitation data according to claim 1 is characterized in that: The process of evaluating and correcting according to the error analysis results includes: The corrected precipitation values ​​are calculated using the geographically weighted regression method using the measured precipitation data of meteorological stations on different time scales, including annual, seasonal and monthly precipitation, satellite remote sensing precipitation TRMM data, digital elevation model data, climate factors and terrain factors. The corrected precipitation values ​​are then compared with the actual precipitation values ​​of the meteorological stations / observatories in the target area to conduct error analysis and evaluate the correction effect.

Images