A method for identifying large area freshwater and saltwater lake types

Abstract

The application discloses a method for identifying large-area fresh water and salt water lake types, which comprises the following steps: binarizing lakes according to the measured salt and fresh water types of the lakes; obtaining attribute data within a research area range; extracting time series values of attribute data of each lake basin within the research area range, and calculating the average value of each attribute data; counting the attribute differences between salt water and fresh water lakes, and selecting the attribute with significant difference as a key attribute; for each key attribute, searching for an optimal segmentation threshold of salt water and fresh water lakes within the value range of the key attribute; taking the optimal identification precision of each key attribute as a weight, and taking the salt and fresh water type identification result of each key attribute under the optimal segmentation threshold as a value to perform weighted average calculation, rounding off the weighted average calculation result, and identifying salt water and fresh water lakes according to the final calculation result. The application can realize high-precision identification of salt and fresh water lakes in a large area.

Description

Technical Field

[0001] This invention relates to the fields of limnology, multi-source product application, and lake water environment prediction. It mainly analyzes various ecological and environmental attributes of lakes and watersheds, selects key attributes to identify the types of freshwater (≤ 2,000 μs / cm) and saline (> 2,000 μs / cm) in lakes, then determines the optimal segmentation threshold for identifying freshwater and saline water in lakes by step-by-step searching, and finally constructs a method for identifying the types of freshwater and saline water in large-area lakes by integrating the identification accuracy and segmentation threshold of all key attributes. Background Technology

[0002] Lake salinity refers to the concentration of dissolved salts and minerals in water bodies, and is usually quantitatively characterized by electrical conductivity (μS / cm) and practical salinity units (‰). Changes in lake salinity can have wide-ranging impacts on ecosystems, water resources, water quality, human health, agriculture, climate, and the carbon cycle. Therefore, careful monitoring and management of lake salinity are necessary to ensure ecosystem health and sustainable social development.

[0003] Lake salinity is typically determined by chemical analysis of conductivity or salinity, and lakes are classified as freshwater or saline water based on conductivity ≤ 2,000 μs / cm. However, this method can only identify the freshwater type of a limited number of surveyed lakes, leaving the freshwater type of the vast majority of unsurveyed lakes unknown. Therefore, there is an urgent need to develop a method for identifying freshwater and saline water types in lakes applicable to large areas. Summary of the Invention

[0004] The purpose of this invention is to provide a method for identifying the types of freshwater and saltwater lakes in a large area.

[0005] To achieve the above technical objectives, the present invention adopts the following solution:

[0006] A method for identifying the types of freshwater and saline lakes over a large area includes:

[0007] The lakes are binarized based on the measured freshwater and brackish water types.

[0008] Acquire attribute data within the study area, including meteorological data, geographical data, and human data;

[0009] Extract time series values ​​of attribute data for each lake basin within the study area, and calculate the average value of each attribute data;

[0010] The differences in various properties between saline and freshwater lakes were statistically analyzed, and the properties with significant differences were selected as key properties.

[0011] For each key attribute, find the optimal segmentation threshold for saline and freshwater lakes within its value range;

[0012] The optimal identification accuracy of each key attribute is used as the weight, and the lake saline-freshwater type identification result based on the optimal segmentation threshold of each key attribute is used as the value to perform a weighted average calculation. The weighted average calculation result is rounded to the nearest integer, and the identification of saline-freshwater lakes is performed based on the final calculation result.

[0013] As a preferred implementation method, the measured brackish water type is obtained based on the field survey results of lake conductivity;

[0014] Lakes with an average conductivity ≤ 2,000 μs / cm are classified as freshwater lakes, and lakes with an average conductivity > 2,000 μs / cm are classified as saline lakes.

[0015] In one preferred embodiment, the meteorological data includes monthly average temperature, monthly surface runoff of the watershed, monthly total evapotranspiration, monthly average wind speed, monthly total rainfall, leaf area index of high-type vegetation, and leaf area index of low-type vegetation.

[0016] The human data includes the region's population density and GDP;

[0017] The geographic data includes digital elevation models of the region.

[0018] As a preferred implementation, the extraction of time series values ​​of lake basin attribute data within the study area includes: extracting lake basin boundaries, rasterizing the basins into a grid, and using the grid to extract time series values ​​of lake basin attribute data within the corresponding spatial range.

[0019] Furthermore, for water areas greater than 20 km² 2 The watershed boundaries of each lake were determined based on HydroBASIN and HydroRIVERS data.

[0020] For water areas less than 20 km² 2 The lakes were defined with their respective HydroBASIN sub-basins (level 12) as the lake basins, and their boundaries were extracted.

[0021] As a preferred implementation, the climate average is calculated by arithmetic mean for each attribute data.

[0022] As a preferred implementation, an independent samples t-test is used for each attribute to determine whether the differences between saline and freshwater lakes are significant.

[0023] As a preferred implementation, the value range of each key attribute is determined, and the value range is traversed based on a preset step size. The recognition accuracy of each traversed value is calculated as a segmentation threshold, and the optimal recognition accuracy is obtained. The corresponding segmentation threshold is the optimal segmentation threshold.

[0024] As a preferred implementation method, saltwater and freshwater lakes are identified according to the following formula:

[0025] (1)

[0026] In the formula, Type represents the lake's brackish water type identification result, which is a binary numerical value; P i T represents the optimal identification accuracy for attribute i; i The result is the identification of freshwater / saline water type of lake based on the optimal segmentation threshold of attribute i, and is a binary numerical value; N is the number of attributes.

[0027] As a preferred embodiment, the method further includes obtaining lakes in any region through the HydroLAKES dataset, extracting the average values ​​of key lake-watershed attributes, and establishing a weighted average calculation formula to identify saline and freshwater lakes.

[0028] Based on large-area lake salinity and freshwater property survey data and multi-source data on lake-watershed ecological environment, this invention identifies key lake-watershed attributes for identifying freshwater and saline lake types through comparative analysis. Then, based on these key attributes, a method for identifying large-area freshwater and saline lake types is constructed for the first time.

[0029] The principle of this invention is as follows:

[0030] Lake salinity variation is a complex systemic process, influenced by a combination of natural and anthropogenic factors within the lake-watershed system. These factors can be broadly categorized into three types:

[0031] (1) Precipitation and evaporation are the main controlling factors of lake salinity. Lakes with watershed precipitation greater than evaporation are usually freshwater lakes, while lakes with evaporation exceeding precipitation will evolve into saltwater lakes due to the concentration of minerals through evaporation. Moreover, watershed vegetation cover, temperature and wind speed can also affect the freshwater and saltwater type of lakes by changing the lake-watershed evaporation.

[0032] (2) Inflow and outflow of water bodies also affect their salinity. Watersheds with high surface runoff can discharge a large amount of fresh water into lakes, while large outflow of water bodies can also discharge dissolved salts from lakes. In other words, both inflow and outflow of water bodies are conducive to making lakes freshwater.

[0033] (3) The geological conditions of the lake-watershed and human activities also affect lake salinity. Groundwater and infiltration water may transport dissolved salts into the lake, and human activities may also discharge dissolved salt substances such as chemical fertilizers, industrial wastewater and waste into the lake.

[0034] Based on the above-mentioned factors affecting salinity, this invention proposes that the identification of freshwater and saline lake types in large areas can be achieved based on the multi-source lake-watershed attributes.

[0035] However, some technical challenges need to be addressed before implementing the above technical process, such as:

[0036] (1) Selection of key lake-watershed attributes. There are hundreds of lake-watershed attributes, and the impact and availability of different attributes on lake salinity vary greatly, requiring the selection of key attributes. To address this technical challenge, this invention selects multiple attributes related to watershed water cycle and regional characteristics, and then determines the key attributes of lake salinity through independent samples t-test.

[0037] (2) Determination of the optimal segmentation threshold for key attributes. The selection of the segmentation threshold for each attribute greatly affects the identification accuracy of freshwater and saline water types in lakes, and it is necessary to determine the optimal threshold. To address this technical challenge, this invention determines the optimal threshold for identifying freshwater and saline water for each attribute factor by increasing the value range of each attribute factor from the minimum to the maximum by one percent.

[0038] (3) Comprehensive judgment of freshwater and saline water types. When identifying the freshwater and saline water types of lakes using only a single attribute, there may be issues such as low identification accuracy and the results being unstable due to attribute errors. To address this technical challenge, this invention constructs a comprehensive freshwater and saline water type identification method that integrates multiple attributes of lakes and watersheds, based on the identification accuracy and optimal threshold of each attribute factor.

[0039] The present invention has the following beneficial effects:

[0040] ① It can achieve high-precision identification of fresh and salt water in large-area lakes, with an identification accuracy of up to 94%;

[0041] ② By integrating multiple indicators of the lake-watershed ecological environment, the accuracy of identifying the fresh and salt water properties of lakes is robust and not easily affected by the accuracy of a single property;

[0042] ③ It can be applied to all types of lakes in a large area and can integrate various background information. Detailed Implementation

[0043] This invention details a specific implementation method for constructing a large-area freshwater and saline lake type identification method, as follows:

[0044] (1) Binarization of freshwater and saline water types in lakes. Based on the results of multi-period field surveys of electrical conductivity of lakes across the country, this invention calculates the average electrical conductivity of each lake separately, and then classifies lakes with an average electrical conductivity ≤ 2,000 μs / cm as freshwater lakes and assigns a value of 1; and classifies lakes with an electrical conductivity > 2,000 μs / cm as saline lakes and assigns a value of 2.

[0045] (2) Preparation of lake-basin attribute data. Monthly mean air temperature at 2 m elevation, monthly surface runoff, total monthly evapotranspiration, monthly mean wind speed at 10 m elevation, total monthly precipitation, leaf area index of high-type vegetation and leaf area index of low-type vegetation ("low-type" refers to ground vegetation, including crops and mixed crops, irrigated crops, short grass, tall grass, tundra, semi-desert, swamps and marshes, evergreen shrubs, deciduous shrubs and water-land mixtures; "high-type" refers to trees, including green trees, deciduous trees, mixed forests / woodlands and discontinuous forests) with a spatial resolution of 0.1° and a time span of 2010-2019 were obtained from the Resource and Environmental Science and Data Center of the Chinese Academy of Sciences. Population density and GDP data for the region were obtained from 2015 with a spatial resolution of 1,000 m. A digital elevation model (DEM) of the Chinese region was obtained from NASA in 1999, with a spatial resolution of 30 m.

[0046] (3) Lake-watershed boundaries and gridding. For water areas greater than 20 km² 2 For lakes, the watershed boundaries were determined sequentially by manual identification based on HydroBASIN (levels 1-12) and HydroRIVERS (15 arcseconds) data. For lakes with a water area less than 20 km²... 2 The study used the 12th-order HydroBASIN sub-basins where the lakes were located as lake basins. Large and small lake basins were coded separately using numbers, and non-lake basins were coded as 0. Then, all basins in the Chinese region were rasterized into a grid with a spatial resolution of 1,000 m.

[0047] (4) Extraction of multi-source lake-watershed attributes. Using the rasterized watershed grid data from step (3), computer programming was used to extract the time series values ​​of different lake-watershed attribute data within the corresponding spatial range, and the average values ​​for different watersheds were calculated. As can be seen from step (2), the time resolution of different attribute data is different, and lake salinity is the result of the long-term effects of various attributes. Therefore, this study calculates the climate average value by arithmetic mean based on the extracted time series attribute results, that is, each lake ultimately obtains only one value for each attribute.

[0048] (5) Selection of key lake-watershed attributes. Combining all attributes extracted in step (4) and the results of the lake brackish water survey in step (1), the attribute differences between brackish and freshwater lakes were statistically analyzed. For each attribute, an independent samples t-test was used to determine whether the differences between brackish and freshwater lakes were significant (p < 0.01). If a certain attribute showed a significant difference between brackish and freshwater lakes, it was identified as a key attribute for identifying brackish water in lakes. Finally, eight key attributes were identified, namely evapotranspiration, leaf area index of high-type vegetation, leaf area index of low-type vegetation, surface runoff, 2 m high temperature, precipitation, wind speed, and DEM. Compared with saline lakes, freshwater lakes had lower values ​​for evapotranspiration, wind speed, and DEM, but higher values ​​for other attributes.

[0049] (6) Determination of the optimal segmentation threshold for key attributes. For each key attribute, the threshold is increased by 1% from the minimum to the maximum value range to determine the identification accuracy of the corresponding threshold for the brackish water lakes investigated in step (1), thereby determining the optimal segmentation threshold for each key attribute. When using single evapotranspiration, leaf area index of high-type vegetation, leaf area index of low-type vegetation, surface runoff, 2 m high temperature, precipitation, wind speed and DEM, the optimal identification accuracy of brackish water lakes is 92.9%, 80.19%, 90.84%, 90.45%, 88.45%, 92.84%, 81.03% and 88.58%, respectively.

[0050] (7) Identification of fresh and salt water in lakes by integrating all key attributes. For each surveyed lake in step (1), the optimal threshold of each attribute is used to segment and determine the fresh and salt water type value (1 or 2). Then, the weighted average is calculated based on the optimal identification accuracy of a single attribute. The corresponding result is rounded to obtain the comprehensive identification result, as shown in equation (1).

[0051] (1)

[0052] In the formula, P i The optimal identification accuracy of attribute i obtained in step (6); T i T represents the lake brackish water type identification result based on the optimal segmentation threshold of attribute i. i = 1 represents a freshwater lake, Ti = 2 indicates a saline lake; the round function rounds the calculated result to the nearest integer. Ultimately, the accuracy of identifying the freshwater / saline water type of 1552 lakes nationwide was 94%.

[0053] (8) Application of freshwater and saltwater identification in lakes across the country. For the 24,366 lakes in the HydroLAKES dataset within the Chinese region, the key attribute climate average of all lakes-watersheds is extracted according to steps (1) to (4), and then the freshwater and saltwater type of lakes can be identified using formula (1).

Claims

1. A method for identifying the types of freshwater and saline lakes in a large area, characterized in that, include: Based on the measured freshwater and saltwater types of the lakes, the lakes are binarized. The binarization process refers to classifying lakes with an average conductivity ≤ a preset value as freshwater lakes and assigning them a first value, and classifying lakes with an average conductivity > a preset value as saltwater lakes and assigning them a second value. Acquire attribute data within the study area, including meteorological data, geographical data, and human data; Extract time series values ​​of watershed attribute data for each lake within the study area, and calculate the arithmetic mean of each attribute data for each lake; The differences in various properties between saline and freshwater lakes were statistically analyzed, and the properties with significant differences were selected as key properties. For each key attribute, find the optimal segmentation threshold for saline and freshwater lakes within its value range; The optimal identification accuracy of each key attribute is used as the weight, and the lake saline-freshwater type identification result based on the optimal segmentation threshold of each key attribute is used as the value to perform a weighted average calculation. The weighted average calculation result is rounded to the nearest integer, and the identification of saline-freshwater lakes is performed based on the final calculation result.

2. The method according to claim 1, characterized in that, The measured brackish water type was obtained based on the results of a field survey of lake conductivity. Lakes with an average conductivity ≤ 2,000 μs / cm are classified as freshwater lakes, and lakes with an average conductivity > 2,000 μs / cm are classified as saline lakes.

3. The method according to claim 1, characterized in that, The meteorological data includes monthly average temperature, monthly surface runoff in the watershed, monthly total evapotranspiration, monthly average wind speed, monthly total rainfall, leaf area index of high-type vegetation, and leaf area index of low-type vegetation. The human data includes the region's population density and GDP; The geographic data includes digital elevation models of the region.

4. The method according to claim 1, characterized in that, The extraction of time series values ​​of attribute data for each lake basin within the study area includes: extracting the lake basin boundaries, rasterizing the basins into a grid, and using the grid to extract the time series values ​​of lake basin attribute data within the corresponding spatial range.

5. The method according to claim 4, characterized in that, For water areas greater than 20 km² 2 The watershed boundaries of each lake were determined based on HydroBASIN and HydroRIVERS data. For water areas less than 20 km² 2 The lakes were defined with their respective HydroBASIN sub-basins (level 12) as the lake basins, and their boundaries were extracted.

6. The method according to claim 1, characterized in that, The climate average is calculated by arithmetic mean for each attribute data.

7. The method according to claim 1, characterized in that, For each attribute, an independent samples t-test was used to determine whether the differences between saline and freshwater lakes were significant.

8. The method according to claim 1, characterized in that, For each key attribute, determine its value range, traverse its value range based on a preset step size, calculate the recognition accuracy when each traversed value is used as a segmentation threshold, obtain the optimal recognition accuracy, and the corresponding segmentation threshold is the optimal segmentation threshold.

9. The method according to claim 1, characterized in that, The following formula can be used to distinguish between saltwater and freshwater lakes: （1） In the formula, Type represents the lake's brackish water type identification result, which is a binary numerical value; P i The optimal identification accuracy for attribute i; T i The result is the identification of freshwater / saline water type of lake based on the optimal segmentation threshold of attribute i, and is a binary numerical value; N is the number of attributes; The round function rounds the result of a calculation to the nearest integer.

10. The method according to claim 1, characterized in that, It also includes obtaining lakes in any region using the HydroLAKES dataset, extracting the average values ​​of key lake-watershed attributes, and establishing a weighted average calculation formula to identify saline and freshwater lakes.

Images