Precipitation estimation method based on DEM and SAR data

Abstract

The invention discloses a precipitation estimation method based on DEM (Digital Elevation Model) and SAR (Synthetic Aperture Radar) data. The method is characterized by comprising the following steps: acquiring and preprocessing data; data registration: performing accurate geographic registration on the multi-temporal dual-polarized SAR image and the DEM data to ensure that the two kinds of data are aligned under the same geographic coordinate system; surface water body extraction: carrying out image enhancement on the SAR image, and selecting a threshold value to carry out water body extraction; and surface change detection: detecting the range change of the surface water body by using a change detection technology, and identifying the change of the surface water body by comparing SAR image water body images before and after rainfall. In the change detection, two SAR images in different periods are analyzed, and the pixel difference of the SAR images is calculated to detect the surface change; and carrying out precipitation estimation in combination with DEM data. And establishing a mathematical model between the precipitation and the surface deformation by using topographic information provided by the DEM data, and estimating the precipitation. The method can accurately and quickly provide precipitation information of severe terrain areas in extreme weather.

Description

Technical Field

[0001] The invention belongs to the technical field of remote sensing technology and hydrological monitoring, and in particular relates to a precipitation estimation method based on DEM and SAR data. Background Art

[0002] There are many methods for precipitation statistics at present. Among them, rain gauges are the most commonly used ground measurement tools for collecting and measuring precipitation. Its advantages are: it can be measured directly and the data is highly accurate. However, its limitations include: limited spatial coverage, and multiple rain gauges need to be set up in different locations to obtain regional precipitation data; high maintenance and operation costs; and difficulty in deployment in remote or dangerous areas. Numerical weather forecast models simulate atmospheric processes through computer models to predict precipitation conditions. Its advantages are that it can provide future precipitation forecasts, which is helpful for disaster prevention and mitigation; it can be combined with other observational data to improve prediction accuracy. However, there are also limitations: prediction accuracy depends on the accuracy of the model and initial conditions; the model is complex, the calculation is large, and high-performance computing resources are required; there are certain limitations in the simulation of local small-scale weather systems.

[0003] The use of satellite remote sensing images to extract information such as water area, geometry, water quality and water ecological environment has been applied in water resources survey, macro-monitoring of water resources, environmental protection, natural disaster forecasting, land classification and other fields. Water body information extraction has become one of the important branches of remote sensing research and is the premise for building various hydrological models and conducting hydrological simulations in different scenarios. The traditional way to obtain water body and hydrological information is mainly manual field measurement and data acquisition based on hydrological monitoring stations. Although manual measurement of area, length, width, flow velocity and flow can obtain high-precision information, the field measurement cycle is long, dangerous, time-consuming and labor-intensive, and is not suitable for the acquisition and real-time monitoring of water resources information with high temporal and spatial frequencies; remote sensing imaging has a short cycle and strong real-time performance, and high-resolution remote sensing images can be used to quickly, accurately and timely obtain water body information to make up for the shortcomings of manual field measurement. In complex mountainous areas and uninhabited areas that are difficult for humans to enter, there are almost no hydrological monitoring stations that measure water level, flow, precipitation and water ecology. For complex mountainous areas, uninhabited areas, and dangerous environments where natural disasters frequently occur, extracting water body information based on remote sensing data has advantages that other technologies cannot match.

[0004] Using synthetic aperture radar to monitor precipitation on the earth's surface has the advantages of wide coverage, especially suitable for monitoring precipitation in oceans and remote areas, and providing long-term global precipitation data. Its characteristics of using active sensors to send electromagnetic waves to the earth's surface and receive reflected signals ensure all-day and all-weather observation of the ground, and have a certain surface penetration ability. The backscatter coefficients and phase characteristics of horizontally polarized waves and vertically polarized waves on the ground or the ocean are different. Therefore, in addition to increasing the information content of remote sensing through multiple bands, the accuracy of target recognition can also be improved through different polarizations. By measuring and analyzing the polarization characteristics of radar targets and ground clutter, classification and identification of different targets can be achieved. However, due to the side-type geometry of the SAR sensor, the ground may be misclassified due to the relative position of natural features such as urban buildings and taller vegetation. The radar shadow is caused by the position of adjacent buildings and the side-type geometry of the SAR sensor. The radar beam cannot illuminate the surface, causing these areas to look dark in the image, but they are misclassified as flood areas in dry conditions, resulting in excessive monitoring of floods. Therefore, this paper uses DEM data to eliminate the image of mountain shadows to improve the accuracy of precipitation statistics. Summary of the invention

[0005] The purpose of the present invention is to design a precipitation estimation method based on DEM and SAR data, which can provide high-precision and high-resolution surface information and precipitation assessment results, and provide strong support for environmental monitoring, disaster warning and land management.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] A precipitation estimation method based on DEM and SAR data, characterized in that: the method is as follows:

[0008] Step 1: Data acquisition and preprocessing;

[0009] Step 2: Data registration: Accurately geo-register the multi-temporal dual-polarization SAR images and DEM data to ensure that the two data are aligned in the same geographic coordinate system;

[0010] Step 3: Surface water extraction: Enhance the SAR image and select a threshold to extract water.

[0011] Step 4: Surface change detection: Use change detection technology to detect changes in the range of surface water bodies. By comparing SAR images of water bodies before and after precipitation, changes in surface water bodies can be identified. Change detection detects surface changes by analyzing two SAR images from different periods and calculating their pixel differences.

[0012] Step 5: Estimate precipitation using DEM data: Use the terrain information provided by DEM data to estimate precipitation.

[0013] Furthermore, the data in step 1 include SAR image data at multiple time points before and after precipitation, and high-resolution DEM data of the study area are obtained.

[0014] Furthermore, the data preprocessing in step 1 includes performing radiation correction, geometric correction and denoising on the SAR data, and performing necessary interpolation and resampling on the DEM data to match the SAR image data in terms of spatial resolution and coordinate system.

[0015] Furthermore, in step 3, the SAR image is enhanced, and the threshold is selected for water extraction, including: using the polarization information of the SAR image to analyze the change of the surface water body, using the water body index method SDWI to extract the threshold for water body identification, multiplying the VV and VH polarization images and multiplying them by 10 to expand the difference between the water body and other ground objects, and then using the natural logarithm as the function formula,

[0016]

[0017] Where: and is the backscatter coefficient / dB, and the empirical constant 8 is an integer. The purpose is to attribute the threshold of the SDWI method to near 0. When the SDWI value is greater than 0, it is a water body, and when it is less than 0, it is a non-water body. After being processed by the natural logarithm function, the bimodal separation of the image histogram between the polarization images is improved, which is more conducive to the selection of the threshold.

[0018] Furthermore, the surface change detection includes: generating a difference map, which is actually calculating the difference in pixel values ​​between two consecutive grids.

[0019] Further, precipitation estimation based on DEM data includes:

[0020] S1: Establish a mathematical model between precipitation and surface deformation;

[0021] S2: According to the water body area in the SAR image, it can be mapped to the contour surface in the landmark model constructed by DEM. The volume constructed by the SAR plane, the original water body surface and the surface model is the precipitation;

[0022] S3: Precipitation estimation: Apply the above model, based on the water range change information obtained from SAR data and DEM data, estimate the precipitation in the study area. The estimated precipitation V is as follows:

[0023] V 1 =∑(h 灾后水体 -h灾前水体 )

[0024] V 2 =∑(h 灾后水体 -h 灾前地表 )

[0025] V=V 2 +V 1

[0026] Among them, ∑ represents the sum of the elevation values ​​corresponding to all pixels in the range, V is the precipitation, V 1 is the volume of the water body before the disaster, V 2 V is the volume of the part without water before the disaster. When a certain area has no water before the disaster and is flooded, V 1 is 0; h is the DEM elevation value corresponding to pixel i in the water body range of the current SAR image.

[0027] The following beneficial effects can be obtained through the above technical solution:

[0028] The precipitation estimation method based on DEM and SAR data makes full use of multi-temporal dual-polarization SAR images and DEM data, and can quickly and timely obtain the disaster-stricken area and rainfall in the disaster-stricken area under extreme weather and harsh environment. Among them, the use of multi-temporal SAR images for change detection can effectively eliminate the influence of the original water body, and combined with DEM data, it can remove the influence of mountain shadows and increase the accuracy of precipitation assessment. This application can accurately and quickly provide precipitation information in areas with harsh terrain under extreme weather. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a method flow chart.

[0030] Figure 2 It is a schematic diagram of data fusion and surface model construction. DETAILED DESCRIPTION

[0031] The present invention will be further described below in conjunction with the accompanying drawings:

[0032] like Figure 1 As shown, a precipitation estimation method based on DEM and SAR data is characterized in that: the method is as follows:

[0033] Step 1: Data acquisition and preprocessing;

[0034] Step 2: Data registration: Accurately geo-register the multi-temporal dual-polarization SAR images and DEM data to ensure that the two data are aligned in the same geographic coordinate system;

[0035] Step 3: Surface water extraction: Enhance the SAR image and select a threshold to extract water.

[0036] Step 4: Surface change detection: Use change detection technology to detect changes in the range of surface water bodies. By comparing SAR images of water bodies before and after precipitation, changes in surface water bodies can be identified. Change detection detects surface changes by analyzing two SAR images from different periods and calculating their pixel differences.

[0037] Step 5: Estimate precipitation using DEM data: Use the terrain information provided by DEM data to estimate precipitation.

[0038] The data in step 1 include SAR image data at multiple time points before and after precipitation, such as using data from radar satellites such as Sentinel-1 and RADARSAT-2 to obtain high-resolution DEM data of the study area, such as SRTM (Shuttle Radar Topography Mission) and ASTER GDEM.

[0039] The data preprocessing in step 1 includes radiation correction, geometric correction and denoising of SAR data to ensure the quality and accuracy of SAR images; necessary interpolation and resampling of DEM data to match it with SAR image data in spatial resolution and coordinate system.

[0040] In step 3, the SAR image is enhanced, and the threshold is selected for water body extraction, including: using the polarization information of the SAR image (such as HH and VV polarization) to analyze the change of surface water bodies, using the water body index method SDWI to extract the threshold for water body identification. The SDWI method has higher accuracy than the single polarization threshold method and has been widely used in China. The VV and VH polarization images are multiplied and multiplied by 10 to expand the difference between water bodies and other land objects, and then the natural logarithm is used as the function.

[0041]

[0042] Where: and is the backscatter coefficient / dB, and the empirical constant 8 is an integer. The purpose is to attribute the threshold of the SDWI method to near 0. When the SDWI value is greater than 0, it is a water body, and when it is less than 0, it is a non-water body. After being processed by the natural logarithm function, the bimodal separation of the image histogram between the polarization images is improved, which is more conducive to the selection of the threshold.

[0043] Surface change detection includes: difference map generation, which actually calculates the pixel value difference between two continuous grids (such as temperature grids or multi-band images). The purpose of generating a difference map is to preliminarily distinguish between the unchanged class and the changed class in the two SAR images.

[0044] Precipitation estimation based on DEM data includes:

[0045] S1: Establish a mathematical model between precipitation and surface deformation;

[0046] S2: As attached Figure 2 As shown in the figure, according to the water body area in the SAR image, it can be mapped to the contour surface (mapping surface) in the landmark model constructed by DEM. The volume constructed by the SAR plane (upper surface), the original water body surface (lower surface 1) and the surface model (lower surface 2) is the precipitation.

[0047] S3: Precipitation estimation: Apply the above model, based on the water range change information obtained from SAR data and DEM data, estimate the precipitation in the study area. The estimated precipitation V is as follows:

[0048] V 1 =∑(h 灾后水体 -h 灾前水体 )

[0049] V 2 =∑(h 灾后水体 -h 灾前地表 )

[0050] V=V 2 +V 1

[0051] Among them, ∑ represents the sum of the elevation values ​​corresponding to all pixels in the range, V is the precipitation, V 1 is the volume of the water body before the disaster, V 2 V is the volume of the part without water before the disaster. When a certain area has no water before the disaster and is flooded, V 1 is 0; h is the DEM elevation value corresponding to pixel i in the water body range of the current SAR image.

[0052] In the above scheme, before precipitation and before disaster have the same concept and meaning, and after precipitation and after disaster have the same concept and meaning.

[0053] SAR generates high-resolution surface images by actively emitting microwave signals and receiving echoes. Compared with optical remote sensing, SAR is not restricted by weather and lighting conditions and can obtain data at any time and under any meteorological conditions. Through InSAR (Interferometric SAR) technology, high-precision surface deformation maps and DEMs can be generated by interfering SAR images at different times. For example, the Sentinel-1 satellite provides free SAR data worldwide, which is widely used for surface deformation monitoring and DEM construction.

[0054] DEM is widely used in terrain analysis, hydrological models, land use planning, disaster assessment (such as floods and landslides), etc. The construction of surface models relies on high-precision DEM to accurately reflect terrain features. DEM data can be obtained through aerial photogrammetry, LiDAR, satellite altimetry and other methods. ALOS is the Advanced Land Observing Satellite-1 (ALOS) project of the Japan Aerospace Exploration Agency (JAXA). ALOS-12m terrain data comes from the PALSAR sensor of ALOS. PALSAR has multiple observation modes, including single polarization (FBS), multi-polarization (FBD) and polarization mode (PLR).

[0055] Table 1 PALSAR mode comparison

[0056]

[0057] As can be seen from the table above, the terrain resolution of PALSAR radar can reach 10m in FBS mode and 20m in FBD mode. In addition, through actual testing, the actual resolution of DEM data in FBS mode is about 12.5m. Therefore, the effective resolution of ALOS terrain dataset is about 12.5m.

[0058] SAR data can be used to monitor surface changes caused by precipitation, especially soil moisture and flood monitoring. For example, NASA's GPM (Global Precipitation Measurement) combined with SAR data can provide high-resolution precipitation estimates. DEM can be used to analyze the distribution and flow path of precipitation on the terrain, which helps to understand the impact of precipitation on watersheds and groundwater.

[0059] Using near-real-time SAR data and known DME data can achieve more comprehensive terrain and precipitation analysis and improve the accuracy and reliability of the model. By regularly acquiring SAR data, it is possible to dynamically monitor the impact of surface changes and precipitation, provide real-time environmental change information, and achieve the purpose of dynamic monitoring.

[0060] Example 1: Figure 1 As shown in Figure 1, a precipitation estimation method based on DEM and SAR data, the specific steps are:

[0061] 1. Dual-polarization SAR satellite image preprocessing

[0062] Before extracting water bodies from SAR images, the SAR image data needs to be preprocessed. In the PIE-SAR software, the SAR images are sequentially subjected to complex data conversion, multi-view processing, filtering, geocoding, radiometric calibration, and DB conversion.

[0063] The specific steps of SAR data preprocessing are as follows:

[0064] (1) Complex data conversion

[0065] Intensity / amplitude feature is one of the most important features of SAR images. Based on SAR intensity / amplitude images, ground object target information can be extracted. Therefore, it is necessary to convert SAR complex data into SAR intensity / amplitude data. Complex data conversion can convert complex format data into intensity, amplitude, phase, real part, imaginary part and other types.

[0066] (2) Multi-view processing

[0067] In order to improve the visual effect of the image and the estimation accuracy of the backscatter of each pixel, multi-view processing is required. The purpose of multi-view processing is to suppress speckle noise at the cost of reducing resolution. Custom view number: Set the view number in azimuth and range. The multi-view view number can be calculated based on the incident angle, range resolution and azimuth resolution in the imported data (xml).

[0068] (3) Adaptive filtering

[0069] Adaptive filtering belongs to the spatial domain filtering algorithm. It takes a sliding window on the image and filters the pixels in the window to obtain the filter value of the center pixel of the window (the current filtering pixel). In a typical local adaptive filter:

[0070] One type is the Lee algorithm, Frost algorithm and Kuan algorithm, all of which use the mean and variance of pixels in the sliding window as parameters and perform filtering according to certain estimation principles;

[0071] The other type is Sigma filter, Weighting filter and improved Sigma method, which takes the average value of all pixels with the same distribution as the center pixel of the window as the filtering value.

[0072] (4) Geocoding

[0073] Based on the attitude and orbit data transmitted by the satellite, the L1-level image data is geometrically positioned, projected on the map and resampled to form an L2-level geocoding product.

[0074] The geocoding function uses a geometric correction processing method based on the RD positioning model, including geocoding ellipsoid correction (GEC) and geocoding terrain correction (GTC).

[0075] (5) Radiation calibration

[0076] The amplitude feature is the feature of SAR image reflecting different backscattering strengths, and its formula is as follows:

[0077]

[0078] Where: A is the amplitude of the image point in the SAR image; r is the real number in the complex number of the image point in the SAR image; i is the imaginary number.

[0079] (6)DB conversion

[0080] The function of converting to DB image is to output the radar backscatter coefficient and calculate the absolute backscatter value of the corresponding ground object.

[0081] (II) Calculation of dual polarization water index (SDWI)

[0082] Multiply the VV and VH polarized images and multiply by 10 to magnify the difference between water bodies and other land objects, and then use the natural logarithm as the function.

[0083]

[0084] Where: and is the backscatter coefficient / dB. The empirical constant 8 is an integer, the purpose of which is to attribute the threshold of the SDWI method to near 0. When the SDWI value is greater than 0, it is a water body, and when it is less than 0, it is a non-water body. After being processed by the natural logarithm function, the bimodal separation of the image histogram between the polarization images is improved, which is more conducive to the selection of the threshold.

[0085] 3. Shadow extraction

[0086] The slope of the study area was calculated using DEM data. Since the slope of water bodies in mountainous areas, such as reservoirs, lakes, and rivers, is small, areas with a slope threshold above 5° are not easy to retain as water bodies and are considered non-water areas. The slope mask area is thus determined.

[0087] (IV) Change Detection

[0088] Change detection technology is used to detect changes in the range of surface water bodies. By comparing SAR images of water bodies before and after precipitation, changes in surface water bodies can be identified. Change detection detects surface changes by analyzing two SAR images from different periods and calculating their pixel differences.

[0089] (V) Generating contour lines using DEM data

[0090] The contour line tool based on ArcGIS software uses DEM data to generate contour lines, and the density of contour lines is consistent with the resolution of SAR images as much as possible.

[0091] (VI) Establish a mathematical model between precipitation and surface deformation, and use the terrain information provided by DEM data to estimate precipitation

[0092] As attached Figure 2 As shown in the figure, according to the water body area in the SAR image, it can be mapped to the contour surface (mapping surface) in the landmark model constructed by DEM. The volume constructed by the SAR plane (upper surface), the original water body surface (lower surface 1) and the surface model (lower surface 2) is the precipitation.

[0093] According to the water body range obtained in step (ii), the water body edge is obtained, and it is put together with the contour map generated in step (v) to obtain the elevation of the water body surface captured by the SAR image, and the water body height h after the disaster is calculated respectively. 灾后水体 and the water height before the disaster h 灾前水体 and the surface model height h 灾前地表 .

[0094] Precipitation estimation: Apply the above model, based on the water range change information obtained from SAR data and DEM data, estimate the precipitation in the study area. The estimated precipitation V is as follows:

[0095] V 1 =∑(h 灾后水体 -h 灾前水体 )

[0096] V 2 =∑(h 灾后水体 -h 灾前地表 )

[0097] V=V 2 +V 1

[0098] Among them, ∑ represents the sum of the elevation values ​​corresponding to all pixels in the range, V is the precipitation, V 1 is the volume of the water body before the disaster, V 2 V is the volume of the part without water before the disaster. When a certain area has no water before the disaster and is flooded, V 1 is 0; h is the DEM elevation value corresponding to pixel i in the water body range of the current SAR image.

[0099] The above are all preferred embodiments of the present invention. For ordinary technicians in this technical field, without departing from the principle of the present invention, various equivalent modifications to the present invention belong to the protection scope of the claims attached to this application.

Claims

1. A precipitation estimation method based on DEM and SAR data, characterized by: The method is as follows: Step 1: Data acquisition and preprocessing; Step 2: Data registration: Accurately geo-register the multi-temporal dual-polarization SAR images and DEM data to ensure that the two data are aligned in the same geographic coordinate system; Step 3: Surface water extraction: Enhance the SAR image and select a threshold to extract water. Step 4: Surface change detection: Use change detection technology to detect changes in the range of surface water bodies. By comparing SAR images of water bodies before and after precipitation, changes in surface water bodies can be identified. Change detection detects surface changes by analyzing two SAR images from different periods and calculating their pixel differences. Step 5: Estimate precipitation using DEM data: Use the terrain information provided by DEM data to estimate precipitation.

2. The precipitation estimation method based on DEM and SAR data according to claim 1, characterized in that: The data in step 1 include SAR image data at multiple time points before and after precipitation or disasters, and high-resolution DEM data of the study area are obtained.

3. A precipitation estimation method based on DEM and SAR data according to claim 1 or 2, characterized in that: The data preprocessing in step 1 includes radiation correction, geometric correction and denoising of SAR data, and necessary interpolation and resampling of DEM data to match the spatial resolution and coordinate system of SAR image data.

4. The precipitation estimation method based on DEM and SAR data according to claim 1, characterized in that: In step 3, the SAR image is enhanced and the threshold is selected for water extraction, including: using the polarization information of the SAR image to analyze the change of surface water, using the water index method SDWI to extract the threshold for water identification, multiplying the VV and VH polarization images and multiplying them by 10 to expand the difference between the water body and other ground objects, and then using the natural logarithm as the function. Where: and is the backscatter coefficient / dB, and the empirical constant 8 is an integer. The purpose is to attribute the threshold of the SDWI method to near 0. When the SDWI value is greater than 0, it is a water body, and when it is less than 0, it is a non-water body. After being processed by the natural logarithm function, the bimodal separation of the image histogram between the polarization images is improved, which is more conducive to the selection of the threshold.

5. The precipitation estimation method based on DEM and SAR data according to claim 1, characterized in that: Surface change detection includes: difference map generation, which actually calculates the difference in pixel values ​​between two continuous rasters.

6. The precipitation estimation method based on DEM and SAR data according to claim 1, characterized in that: Precipitation estimation based on DEM data includes: S1: Establish a mathematical model between precipitation and surface deformation; S2: According to the water body area in the SAR image, it can be mapped to the contour surface in the landmark model constructed by DEM. The volume constructed by the SAR plane, the original water body surface and the surface model is the precipitation; S3: Precipitation estimation: Apply the above model, based on the water range change information obtained from SAR data and DEM data, estimate the precipitation in the study area. The estimated precipitation V is as follows: V1=∑(h 灾后水体 -h 灾前水体 ) <h2 style=";text-align:left;direction:ltr">V2 = ∑(h<h2 style=";text-align:left;direction:ltr"> 灾后水体 <h2 style=";text-align:left;direction:ltr"> -h<h2 style=";text-align:left;direction:ltr"> 灾前地表 <h2 style=";text-align:left;direction:ltr"> ) V=V2+V1 Among them, ∑ represents the sum of the elevation values ​​corresponding to all pixels in the range, V is the precipitation, V1 is the volume of the part with water before the disaster, and V2 is the volume of the part without water before the disaster. When a certain area has no water before the disaster and is flooded, V1 is 0; h is the DEM elevation value corresponding to pixel i in the water range of the current SAR image.

7. The precipitation estimation method based on DEM and SAR data according to claim 1 or 2, characterized in that: Before extracting water bodies from SAR images, the SAR image data needs to be preprocessed. In the PIE-SAR software, the SAR images are subjected to complex data conversion, multi-view processing, filtering, geocoding, radiation calibration, and DB conversion operations.

Images