Method for reconstructing seasonal change track of water volume for small lakes and reservoirs

Abstract

The invention discloses a small lake and reservoir-oriented water quantity seasonal change trajectory reconstruction method, which comprises the following steps of: selecting a lake and reservoir virtual station position, and acquiring medium-high resolution remote sensing satellite data and SWOT satellite data covering a lake and reservoir for extracting a water area range and water level information of a virtual station region; respectively constructing an SWOT water level time sequence and a virtual station area time sequence of the virtual station, performing data pairing on the SWOT water level time sequence and the virtual station area time sequence, and constructing a water level-area curve based on the paired data; and reconstructing the water level with the corresponding time loss by using the curve, obtaining an encrypted water level time sequence, and calculating water volume change time sequence information of the virtual station region in combination with the area time sequence. The invention develops an extensible remote sensing monitoring framework for small lakes and reservoirs which are not covered by traditional altimetry satellites, and realizes refined time sequence reconstruction of water level and water volume changes of the small lakes and reservoirs by using SWOT satellite data products in combination with existing free open medium-high resolution images.

Description

Technical Field

[0001] The invention relates to the field of lake and reservoir hydrology and remote sensing science and technology, and in particular to a method for reconstructing seasonal water volume change trajectories for small lakes and reservoirs. Background Art

[0002] As important surface water storage carriers, lakes and reservoirs are an important part of the surface hydrological cycle and the main source of agricultural irrigation, industrial water and domestic water. In addition, lakes and reservoirs play an important supporting role in flood prevention and disaster reduction and maintaining the stability of the ecosystem. Changes in lakes and reservoirs will affect human water use and ecological environment stability. Therefore, continuous and intensive monitoring of lake and reservoir water resources and analysis of their changing patterns are particularly important for understanding the response mechanism of lakes under climate change and the timely response mode of reservoirs under human regulation, which is conducive to the effective management of water resources.

[0003] Generally speaking, changes in lakes and reservoirs can be directly observed through hydrological monitoring stations, but this is not only time-consuming and labor-intensive, but also has obvious timeliness. On the other hand, hydrological monitoring stations are established for large and medium-sized lakes and reservoirs. In fact, the dynamic changes of small lakes and reservoirs are more sensitive. Therefore, to meet the needs of global and regional water resources management, ground monitoring methods alone are completely insufficient. The integrated application and development of optical remote sensing images and altimetry satellite data have made it possible to obtain long-term change information of lakes and reservoirs on a large scale. Its high efficiency and low-cost observation advantages have promoted the further development of hydrological research. However, optical sensors are easily affected by weather conditions and clouds, and the frequency of observations in rainy and monsoon areas is greatly reduced. The altimetry mission faces the problems of limited spatial coverage, low resolution and long revisit period, and there are many small lakes and reservoirs that are not covered by altimetry. For example, global water level monitoring research based on the current best-performing ICESat-2 laser altimetry satellite is limited to 10 km 2 Above the lake reservoir.

[0004] In summary, on the one hand, there are few existing hydrological stations on small lakes and reservoirs, and it is difficult to obtain measured water level data. Traditional altimetry satellites have limited spatial coverage and lack the ability to monitor the water levels of many small lakes and reservoirs. On the other hand, the service period of a single satellite is limited, and the temporal and spatial resolutions cannot be taken into account. The use of a single data source is difficult to meet the needs of local high-frequency and continuous observations. Therefore, the integration of multi-source and multi-platform remote sensing to improve the temporal frequency and spatial coverage of surface water observations such as small lakes and reservoirs is an important trend in current hydrological dynamic monitoring. Against this background, the SWOT satellite, which was successfully launched in December 2022, realized the wide-swath altimetry mission for the first time by carrying a Ka-band radar interferometer (KaRIn), and broke through the kilometer-scale limit of the radar altimeter through phase processing, achieving simultaneous acquisition of an area of ​​0.0625 km 2The inundation range and water level of the above lakes and reservoirs can significantly improve the spatial resolution and coverage of traditional water regime monitoring.

[0005] Although existing studies have carried out monitoring of water volume changes in lakes and reservoirs by constructing storage curves by combining SWOT-like data and water body area, since SWOT-like data are mainly simulated data generated by referring to radar altimetry data products, such studies are similar to traditional altimetry methods and can only focus on large lakes and reservoirs, and fail to evaluate the monitoring ability of the real SWOT data products for small lakes and reservoirs. On the other hand, there are inherent differences in imaging spectral characteristics and spatial resolution between different satellite sensors. Although they can be ignored when extracting the area of large lakes and reservoirs, it is necessary to perform systematic bias calibration when integrating multi-source water body area time series for small lakes and reservoirs. Therefore, the present invention uses the published SWOT satellite data products as the core to truly evaluate its ability to monitor the water levels of small lakes and reservoirs not covered by traditional altimetry; at the same time, based on consistency evaluation and systematic bias calibration, it integrates the area time series from multi-source satellite image data, and then constructs a high-precision water level-area relationship model to encrypt the water level time series. This provides an extensible solution for monitoring the water level / water volume changes of small lakes and reservoirs. In the future, with the release of more medium and high-resolution satellite data products, daily remote sensing monitoring of the water levels of small lakes and reservoirs can be realized based on this method, providing important data support for analyzing the seasonal variation laws of lake water volume and capturing extreme hydrological events in a short time. By achieving comprehensive monitoring in data-scarce areas, this study not only provides support for improving water resources management, but also contributes scientific knowledge to revealing the impact of human activities on freshwater resources. Summary of the Invention

[0006] The purpose of the present invention is to propose a method for reconstructing the seasonal change trajectory of water volume for small lakes and reservoirs in view of the significant dynamic changes of many small lakes and reservoirs in the prior art but the lack of existing radar / laser altimetry satellite monitoring. With the advantage of the latest SWOT satellite's full spatial coverage for hydrological observations, it can significantly improve the ability of remote sensing technology in the refined monitoring of the water levels of small lakes and reservoirs. Combining the advantages of the dense frequency of publicly available and free medium and high-resolution satellite joint observations (Landsat8 / 9, Sentinel-1, Sentinel-2), by constructing a high-precision lake area-water level storage curve model, it is possible to synchronously reconstruct the long-time series, high-time frequency water level and water volume change information of small lakes and reservoirs.

[0007] To achieve the above technical objectives, the present invention adopts the following technical solutions:

[0008] A method for reconstructing the seasonal change trajectory of water volume for small lakes and reservoirs, comprising:

[0009] Select the virtual station locations for lakes and reservoirs, and obtain medium- and high-resolution remote sensing satellite data and SWOT satellite data covering the lakes and reservoirs;

[0010] Extract the water surface elevation of the lakes and reservoirs based on the SWOT satellite data, extract the water level information of all the pixel points of the SWOT satellite data images within the virtual station area, and construct the SWOT water level time series of the virtual station;

[0011] Extract the water area range of the virtual station based on the medium- and high-resolution remote sensing satellite data, and obtain the time series of the virtual station area;

[0012] Pair the SWOT water level time series with the time series of the virtual station area, and construct a water level-area curve based on the paired data;

[0013] Based on the time series of the virtual station area and the water level-area relationship model, reconstruct the missing water levels at corresponding times, obtain the encrypted water level time series, and calculate the time series information of the water volume change in the virtual station area using the encrypted water level time series and the area time series.

[0014] As a preferred implementation manner, the medium- and high-resolution remote sensing satellite data is sourced from multi-source medium- and high-resolution remote sensing satellites.

[0015] Furthermore, for multi-source medium- and high-resolution remote sensing satellites, calculate the mean value of the water area range extraction results of different images on the same day as the virtual station area for that day.

[0016] As a preferred implementation manner, adopt the method of edge detection combined with Otsu threshold segmentation to extract the water area range of the virtual station.

[0017] As a preferred implementation manner, the optical remote sensing satellite data covering the lakes and reservoirs is the optical remote sensing satellite data covering the largest water area range of the lakes and reservoirs.

[0018] As a preferred implementation manner, the selection method of the virtual station location is as follows: select the shoreline part where the difference between the boundary of the largest water body and the boundary of the permanent water body of the lake or reservoir exceeds a meters and the average slope of the surrounding terrain is less than b as the virtual station location; a and b are preset thresholds. Preferably, select the shoreline part where the annual change of the water area boundary of the lake or reservoir is greater than 100 m and the average slope of the surrounding terrain is less than 5° as the virtual station location. The water level and water surface area of the lakes and reservoirs in the virtual station area are less affected by the steep terrain, which can more accurately depict the linear relationship between the co-variation of the water level and the area. Moreover, the significant expansion and contraction changes of the lakes in the flat shoreline area are more conducive to constructing a high-precision water level-area model.

[0019] As a preferred embodiment, when extracting the water level information of all the pixel points of the SWOT satellite data images within the virtual station area, if there is SWOT data that does not directly cover the virtual station area, the elevation difference between each area and the virtual station is calculated using the monthly water surface slope of the lake or reservoir with full coverage, and the water levels of other areas are converted to the water level benchmark of the virtual station.

[0020] As a preferred embodiment, the time deviation between the paired data of the SWOT water level time series and the virtual station area time series is ≤ 3 days.

[0021] As a preferred embodiment, the number of paired data is > 8 pairs, and the R value of the constructed water level - area relationship model 2 > 0.8.

[0022] As a preferred embodiment, using the encrypted water level time series and area time series, the water volume change in the virtual station area is calculated based on the following formula:

[0023]

[0024] In the formula, is the water volume change of the lake or reservoir at time t 1 and t 2 , A 1 is the area at time t 1 , A 2 is the area at time t 2 , H 1 is the water level at time t 1 , and H 2 is the water level at time t 2 .

[0025] As a preferred embodiment, the spatial resolution of the medium - high - resolution remote sensing satellite data is ≤ 30m.

[0026] As a preferred embodiment, the SWOT satellite data is uniformly calibrated to the EGM2008 / WGS84 reference system after pre - processing to facilitate matching with the subsequent lake or reservoir area. The pre - processing includes format conversion, spatial projection, etc.

[0027] The present invention has the following beneficial effects:

[0028] (1) The present invention develops an extensible remote sensing monitoring framework for small lakes and reservoirs not covered by traditional altimetry satellites, and uses SWOT satellite data products combined with existing freely available medium - high - resolution images to achieve refined temporal reconstruction of the water levels and water volume changes of small lakes and reservoirs, which has important application significance for quantifying regional water resources and analyzing the seasonal variation laws of lake and reservoir water volumes.

[0029] (2)The present invention uses SWOT satellite data for water level extraction, which can continuously monitor the water levels of lakes and reservoirs with an area of 0.0625 km 2 or more, making up for the defect that traditional altimetry satellites cannot cover small lakes and reservoirs, and significantly improving the ability of remote sensing technology in the refined monitoring of lake and reservoir water levels. On the other hand, through the joint observation of freely available medium and high-resolution satellites (Landsat8 / 9, Sentinel-1, Sentinel-2), the present invention can greatly increase the monitoring frequency of water area, and can synchronously reconstruct the long time series, high time frequency water level and water volume change information of small lakes and reservoirs.

[0030] (3)The method of the present invention is simple to implement and has low requirements for data, providing an extensible solution for the monitoring of water level / water volume changes in small lakes and reservoirs. In the future, with the release of more medium and high-resolution satellite data products, daily remote sensing monitoring of the water levels of small lakes and reservoirs can be achieved based on this method, which has important application significance for analyzing the seasonal variation law of water volume in large-scale lakes and reservoirs and capturing extreme hydrological events in a short time. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The drawings are not intended to be drawn to scale. In the drawings, each identical or approximately identical component shown in each figure may be denoted by the same reference numeral. For the sake of clarity, not every component is labeled in each figure. Now, each step of the present invention will be described through research cases and with reference to the drawings, where:

[0032] Figure 1 is the distribution map of the water body frequency change and the virtual station area in the Beijiao Reservoir in Embodiment 1 of the present invention.

[0033] Figure 2 is the main flow chart of the inventive method in Embodiment 1 of the present invention.

[0034] Figure 3 is a schematic diagram of stepwise extraction of the water surface elevation of the Beijiao Reservoir (a), the water surface elevation of the virtual station area (b), rejection of the water level anomaly value of the virtual station (c), and calculation of the average daily value water level (d) based on SWOT data in Embodiment 1 of the present invention.

[0035] Figure 4 is the accuracy verification result diagram of the water level sequence of the virtual station of the Beijiao Reservoir extracted based on SWOT data in combination with the measured water level of the reservoir in Embodiment 1 of the present invention.

[0036] Figure 5 is the comparison of the virtual station areas of the Beijiao Reservoir extracted based on Sentinel-1 and Landsat-8 / 9 data in Embodiment 1 of the present invention.

[0037] Figure 6It is the area sequence of the virtual station of Beijiao Reservoir after water body extraction based on multi-source image data and quality control in Embodiment 1 of the present invention.

[0038] Figure 7 It is the storage capacity curve model constructed based on the nearly synchronous water levels and areas of the virtual stations of Beijiao Reservoir in Embodiment 1 of the present invention.

[0039] Figure 8 It is the high-frequency water level sequence of the virtual station reconstructed based on the constructed storage capacity curve model and the area sequence of the virtual stations of Beijiao Reservoir in Embodiment 1 of the present invention.

[0040] Figure 9 It is the water volume change sequence of the virtual station area calculated based on the water volume change calculation formula and combined with the area sequence of the virtual station and the reconstructed water level sequence of the virtual station in Embodiment 1 of the present invention. Detailed implementation manners

[0041] The following combines the accompanying drawings and embodiments to further describe in detail the specific implementation manners of the present invention. The following embodiments are used to illustrate the present invention, but are not used to limit the scope of the present invention.

[0042] Embodiment 1

[0043] Embodiment 1 of the present application takes Beijiao Reservoir as the research object, and the spatial distribution of the reservoir is as Figure 1 shown. As one of the important drinking water sources in Xiangyang City, Hubei Province, although the water area of Beijiao Reservoir is less than 3 km 2 , its dynamic changes in hydrology and water resources are related to local drinking water safety. Therefore, it is very important to continuously monitor the water levels and water volume changes of Beijiao Reservoir. And Beijiao Reservoir is not covered by existing traditional altimetry satellites, so it is more suitable as the research object of the present invention.

[0044] As Figure 2 shown, it is the main method flow chart of Embodiment 1. Embodiment 1 includes the following steps:

[0045] Step 1: The present invention first combines the multi-year water body frequency data of the Global Surface Water dataset (JRC Global Surface Water, GSW) and the SRTM (Shuttle Radar Topography Mission) DEM data, and selects the shoreline part where the difference between the maximum water body boundary and the permanent water body boundary of Beijiao Reservoir exceeds 100 meters and the average terrain slope around is less than 5° as the virtual station location, as Figure 1As shown in the figure. The water level and water surface area of Beijiao Reservoir in the virtual station area are less affected by the steep terrain, can better represent the synchronous change trend of water level and area and maintain their linear relationship, are not easily affected by the steep terrain, and the significant expansion and contraction changes of lakes in the flat shoreline area are more conducive to constructing a high-precision water level-area model.

[0046] Next, taking the maximum water area range of Beijiao Reservoir in the GSW dataset (i.e., the water area range included in the maximum water body boundary) as the spatial constraint, Landsat-8 / 9, Sentinel-2 optical image data, and Sentinel-1 SAR image data covering the virtual station area of Beijiao Reservoir were screened and sorted on the GEE platform.

[0047] Finally, the SWOT WaterMask Raster Image Level-2 data product with a spatial resolution of 100 m was selected and downloaded from the data website (https: / / podaac.jpl.nasa.gov / dataset) of the Jet Propulsion Laboratory (JPL) of the National Aeronautics and Space Administration (NASA) of the United States, which improved the data processing efficiency while ensuring the accuracy of water level data. The SWOT satellite data product can cover lakes and reservoirs with an area of more than 0.0625 km 2 Combined with the SWOT data period, the research period of this Example 1 was set from July 1, 2023 to June 30, 2024.

[0048] Step 2: Perform preprocessing operations such as variable extraction, format conversion, and spatial projection on all the downloaded SWOT data, and uniformly calibrate them to the EGM2008 / WGS84 reference system, and further generate water surface elevation data within the complete water area of lakes and reservoirs during all effective periods, as shown in Figure 3 In (a) is the water surface elevation distribution of Beijiao Reservoir on July 29, 2023. Relevant attribute information such as water level and water level error of all SWOT pixels in the virtual station area was extracted. Considering that the water level elevation near the shoreline is easily contaminated with some land information, the data within 100 m from the reservoir boundary was excluded. Here, the water level error refers to the standard deviation of the water levels of all pixels and represents the uncertainty of the current water level.

[0049] In some embodiments, when extracting water levels, the data is divided into two groups for water level extraction. For SWOT data directly covering the virtual station area, the virtual station water level can be directly derived. For SWOT data not directly covering the virtual station area, the elevation difference between each location and the virtual station is calculated using the monthly water surface slope of the reservoir with full coverage, so that the water levels in other areas can be converted to the virtual station water level reference. The monthly water surface slope refers to the monthly water surface elevation difference relative to the virtual station area, that is, the SWOT data that completely covers the maximum water body boundary of the reservoir every month is selected. Finally, the water level time series of the two groups are integrated to obtain a complete SWOT-based reservoir water level time series, which can effectively increase the SWOT observation water level frequency. In this embodiment, since the area of the Beijiao Reservoir involved is small, all SWOT data completely cover the virtual station area, and the virtual station water level can be directly derived by excluding outliers.

[0050] Preferably, quality control is performed on the extracted water level data to ensure the accuracy and reliability of the extracted time series, specifically including: removing outliers from the water level data of the same day based on the iterative two-standard-deviation statistical denoising method ( Figure 3 in (c)), and calculating the average daily value water level based on the denoised data to obtain the SWOT water level time series of the virtual station of the Beijiao Reservoir after quality control, with a total of 27 observations, as Figure 3 shown in (d). Finally, combined with the measured water level data of the lake and reservoir, the water levels of the same date are selected to verify the accuracy of the virtual station water level sequence obtained based on SWOT data, and evaluate its feasibility for subsequent research, as Figure 4 shown.

[0051] Step 3: On the GEE platform, for the Landsat-8 / 9 and Sentinel-2 optical images screened in Step 1, identify low-quality observation pixels covered by clouds, snow, or ice through quality control bands, remove them, then calculate the Modified Normalized Difference Water Index (MNDWI), and use the Otsu maximum inter-class variance threshold method to perform threshold segmentation on the water index image to extract the virtual station water area range; for the Sentinel-1 SAR image, extract the water area range through the method of Gamma-MAP filtering, edge detection combined with Otsu threshold segmentation. Combining the virtual station range selected in Step 1, select the water area range extraction results where the lake and reservoir boundaries in the virtual station area are complete or can be corrected by manual vector editing. Then, based on the original remote sensing images, through visual inspection, manually correct and edit the extraction results with individual errors. Finally, perform multi-temporal integration and area calculation of the virtual station water area range, and conduct consistency evaluation on the areas obtained from different image sources, as follows Figure 5It can be clearly seen that there are certain systematic biases in the virtual station areas extracted based on Sentinel-1 and Landsat-8 / 9 satellites. Calibration is performed based on the average relative deviation, and the mean values of the extraction results of different images on the same day are calculated to obtain the time series of the virtual station area of the Beijiao Reservoir, as Figure 6 shown.

[0052] Step 4: Data sorting is carried out based on the virtual station SWOT water levels extracted in Step 2 and the time series of the virtual station areas obtained in Step 3. According to the data pairing situation, a pairing standard with an area-time deviation from the water level ≤ 3 days is set. The R² of the storage capacity curve model for the area-water level relationship of the Beijiao Reservoir finally constructed is 0.92, and the number of paired water levels and areas is 13, as Figure 7 shown.

[0053] Step 5: Based on the storage capacity curve model constructed in Step 4 and the time series of the virtual station areas obtained in Step 3, the virtual station water level series of the Beijiao Reservoir is reconstructed ( Figure 8 ), a total of 65 water level observations are encrypted and supplemented, and the time series information of the water volume change in the virtual station area of the Beijiao Reservoir is obtained by combining the following formula (the calculation formula for the water volume change in lakes and reservoirs), as Figure 9 shown.

[0054]

[0055] In the formula, is the water volume change in the lake or reservoir at times t 1 , t 2 , A 1 is the area at time t 1 , A 2 is the area at time t 2 , H 1 is the water level at time t 1 , and H 2 is the water level at time t 2 .

[0056] Through the above method, continuous and dense time series information on the water levels and water volume changes of corresponding small lakes and reservoirs can be obtained, which can be extended to large-region and even global-scale research, and can provide important methodological support for analyzing the seasonal variation laws of lake and reservoir water volumes and predicting extreme hydrological events in a short period.

[0057] Although the present invention illustrates the method with the above cases, it is not intended to limit the present invention. Those with ordinary knowledge in the technical field to which the present invention pertains can make various modifications and refinements without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be determined by what is defined in the claims.

Claims

1. A method for reconstructing seasonal changes in water volume in small lakes and reservoirs, characterized in that: include: Select the location of the lake and reservoir virtual station, and obtain medium- and high-resolution remote sensing satellite data and SWOT satellite data covering the lake and reservoir; Extracting lake surface elevation based on the SWOT satellite data, and extracting water level information of all SWOT satellite data image pixels in the virtual station area, and constructing the SWOT water level time series of the virtual station; Extract the water area of ​​the virtual station based on the medium and high resolution remote sensing satellite data to obtain the time series of the virtual station area; Pairing the SWOT water level time series with the virtual station area time series, and constructing a water level-area curve based on the paired data; Based on the virtual station area time series and the water level-area relationship model, the water level corresponding to the missing time is reconstructed to obtain the encrypted water level time series, and the encrypted water level time series and area time series are used to calculate the time series information of water volume changes in the virtual station area.

2. The method according to claim 1, characterized in that The medium to high resolution remote sensing satellite data are derived from multi-source medium to high resolution remote sensing satellites.

3. The method according to claim 2, characterized in that For medium- and high-resolution remote sensing satellites from multiple sources, the average of the water area extraction results from different images on the same day is calculated as the virtual station area for that day.

4. The method according to claim 1 or 2, characterized in that: The water area of ​​the virtual station is extracted by combining edge detection with Otsu threshold segmentation.

5. The method according to claim 1, characterized in that The optical remote sensing satellite data covering lakes and reservoirs are optical remote sensing satellite data covering the largest water area of ​​lakes and reservoirs.

6. The method according to claim 1, characterized in that The virtual station location is selected in the following manner: the shoreline portion where the difference between the maximum water body boundary of the lake and the permanent water body boundary exceeds a meters and the average slope of the surrounding terrain is less than b is selected as the virtual station location; a and b are preset thresholds.

7. The method according to claim 1, characterized in that The paired data time deviation between the SWOT water level time series and the virtual station area time series is ≤3 days.

8. The method according to claim 1, characterized in that The number of paired data is greater than 8 pairs, and the R value of the constructed water level-area relationship model is 2 >0.

8.

9. The method according to claim 1, characterized in that: Using the encrypted water level time series and area time series, the water volume change in the virtual station area is calculated based on the following formula: ； In the formula, is the change of lake water volume at t1 and t2, A1 is the area at t1, A2 is the area at t2, H1 is the water level at t1, and H2 is the water level at t2.

10. The method according to claim 1, characterized in that The spatial resolution of the medium- and high-resolution remote sensing satellite data is ≤30m.

Images