A method for reconstructing mid-latitude forest SIF based on EVI and LST

Abstract

The application discloses a mid-latitude forest SIF reconstruction method based on EVI and LST, comprising the following steps: 1) constructing an EVI and LST pixel set of a global forest region; 2) constructing an EVI and LST pixel set of a mid-latitude forest region; 3) constructing a SIF data set; and 4) establishing a SIF prediction model of the mid-latitude forest region. The application compresses the number of prediction factors of the model to two (EVI and LST), all the high-resolution satellite data of the prediction factors cover the whole world, and all the prediction factors can be directly obtained, so that the efficiency and universality of the SIF reconstruction are improved, and the application value of the SIF in regulating the global carbon cycle is improved.

Description

Technical Field

[0001] This invention relates to the field of mid-latitude forest SIF reconstruction, specifically to a method for mid-latitude forest SIF reconstruction based on Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST). Background Technology

[0002] Forests, as major terrestrial carbon-scavenging ecosystems, play a crucial role in regulating the global carbon cycle. SIF (Spectral Intensity Reflectance) is considered the medium of vegetation photosynthesis and helps assess its photosynthetic intensity and carbon-scavenging capacity. Some satellites provide SIF data retrieved from vegetation spectral reflectance. However, SIF retrieved directly from satellite observations suffers from low spatial resolution, discontinuities, or low temporal resolution.

[0003] Traditional SIF reconstruction models have improved in terms of low spatial resolution, continuity, or low temporal resolution, but they still face the following challenges. Traditional models often rely on multiple predictors, such as the vegetation index NDVI, enhanced vegetation index EVI, land surface temperature LST, vapor pressure deficit VPD, and photosynthetically active radiation PAR. However, some predictors (such as vapor pressure deficit VPD and photosynthetically active radiation PAR) suffer from low spatial resolution (the highest spatial resolution of publicly available vapor pressure deficit VPD data is currently only 0.5°), and there are no observations or public goods in certain specific regions. As a result, traditional models are only applicable to a small number of regions where all predictor datasets are available, and the reconstructed solar-induced chlorophyll fluorescence data has low spatial resolution. Summary of the Invention

[0004] This invention aims to address the shortcomings of existing technologies by proposing a mid-latitude forest SIF reconstruction method based on EVI and LST, with the goal of significantly improving the efficiency and universality of SIF reconstruction, thereby enhancing the application value of SIF in regulating the global carbon cycle.

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

[0006] The mid-latitude forest SIF reconstruction method based on EVI and LST of this invention is characterized by the following steps:

[0007] Step 1: Construct a pixel set of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) for global forest regions;

[0008] Step 1.1: Extract forest regions worldwide;

[0009] Obtain land cover imagery from the World Cover dataset and determine the boundaries of global forest regions from them, thereby drawing the boundary vectors (ROIs) of global forest regions. f ;

[0010] Step 1.2: Construct pixel sets of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) for global forest regions;

[0011] Raster data of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) for global forest regions were acquired using MODIS remote sensing satellites.

[0012] The boundary vector ROI of the global forest region f By overlaying raster data of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) in global forest regions, pixel sets of EVI and LST are extracted. in, and These are the enhanced vegetation index and surface temperature of the nth pixel sample in the global forest region, respectively, where N is the number of pixel samples in the pixel set;

[0013] Step 2: Construct a pixel set of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) for forest areas in the mid-latitude region;

[0014] Step 2.1: Extract forest areas within the mid-latitude region;

[0015] Obtain land cover imagery from the World Cover dataset and determine the boundaries of forest areas within the mid-latitude region, thereby drawing the boundary vectors (ROIs) of forest areas within the mid-latitude region. m ;

[0016] Step 2.2: Construct a pixel set containing the Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) of forest areas in the mid-latitude region;

[0017] Raster data of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) in forest areas within the mid-latitude region were acquired using MODIS remote sensing satellites.

[0018] The boundary vector ROI of the forest area within the mid-latitude region. m The pixel sets of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) are extracted by overlaying raster data of forest areas in the mid-latitude region. in, and These are the enhanced vegetation index and surface temperature of the nth pixel sample in the forest area within the mid-latitude region, respectively, where N is the number of pixel samples in the pixel set;

[0019] Step 3: Construct a sunlight-induced chlorophyll fluorescence (SIF) dataset;

[0020] The daylight-induced chlorophyll fluorescence value (SIF) of the nth pixel sample is calculated using equation (1). n This yields the chlorophyll fluorescence values ​​of N pixel samples, forming the SIF dataset. <n<N;

[0021]

[0022] In equation (1), Δ L on n ΔLat represents the maximum difference in longitude of the forest region where the nth pixel sample is located; n SIF represents the maximum latitudinal difference of the forest region where the nth pixel sample is located; x,y This represents the photoinduced chlorophyll fluorescence value with relative latitude and longitude coordinates (x, y) in forest areas within global or mid-latitude regions;

[0023] Step 4: Establish a prediction model for sunlight-induced chlorophyll fluorescence (SIF) in mid-latitude forests;

[0024] Step 4.1: Use equation (2) to obtain the sunlight-induced chlorophyll fluorescence (SIF) value of the reconstructed nth pixel sample.

[0025]

[0026] In equation (2), a i b i ε represents the weight value of the i-th enhanced vegetation index and surface temperature. i Let t represent the i-th random error. i Choose a value for the i-th coefficient, and t i ∈{0,1}, where i is the number of optimal solutions; This represents the enhanced vegetation index of the nth pixel sample within a global forest region or a mid-latitude forest region. This represents the surface temperature of the nth pixel sample within a global forest region or a mid-latitude forest region, where s∈{m,f};

[0027] Step 4.2: Using equations (3) and (4), obtain the correlation R between enhanced vegetation index (EVI) and land surface temperature (LST) and sunlight-induced chlorophyll fluorescence (SIF). 2 And error RMSE:

[0028]

[0029]

[0030] Step 4.3: Solve equations (3) and (4) using the least squares method to obtain the correlation R. 2By maximizing the maximum value and minimizing the RMSE (Recovery Mean Squared Error), the optimal coefficient matrix is ​​obtained. in, This represents the optimal weighting value between the i-th enhanced vegetation index and the surface temperature. This represents the i-th optimal error;

[0031] Step 4.4: Obtain the optimal coefficient matrix. Substituting into equation (2), the optimal SIF value of sunlight-induced chlorophyll fluorescence in forest areas within the reconstructed mid-latitude region is obtained. This yields the optimal set of light-induced chlorophyll fluorescence (SIF) in mid-latitude forests.

[0032] The present invention provides an electronic device, including a memory and a processor, wherein the memory is used to store a program that supports the processor in executing the mid-latitude forest SIF reconstruction method, and the processor is configured to execute the program stored in the memory.

[0033] The present invention discloses a computer-readable storage medium on which a computer program is stored, wherein the computer program, when executed by a processor, performs the steps of the mid-latitude forest SIF reconstruction method.

[0034] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0035] 1. Traditional models often rely on multiple predictors, some of which suffer from low spatial resolution. This invention compresses the number of predictors in the model to two (Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST)). All predictors have high-resolution satellite data, which can meet the requirements of high-resolution SIF reconstruction.

[0036] 2. Traditional models rely on multiple predictors, some of which have limited publicly available datasets. This invention relies on only two predictors, and global data for all predictors is readily available, thus improving the model's universality.

[0037] 3. Traditional models often use neural networks to build models, which have high algorithm complexity. The model constructed in this invention is a linear model, which significantly reduces the algorithm complexity and is a fast SIF reconstruction scheme. Attached Figure Description

[0038] Figure 1 This is a flowchart of a mid-latitude forest SIF reconstruction method based on EVI and LST according to the present invention;

[0039] Figure 2a Relationship diagram of SIF reconstruction model in January at a monthly scale;

[0040] Figure 2b Relationship diagram of SIF reconstruction model in February at the monthly scale;

[0041] Figure 2c Relationship diagram of the SIF reconstruction model in December at the monthly scale;

[0042] Figure 3a Correlation diagram of SIF reconstruction model from March to June at a seasonal scale;

[0043] Figure 3b Correlation diagram of SIF reconstruction model from July to November at a seasonal scale;

[0044] Figure 4 This is a result diagram of a mid-latitude forest SIF reconstruction method based on EVI and LST according to the present invention. Detailed Implementation

[0045] In this embodiment, addressing the problems of traditional models relying on numerous predictor factors and the low spatial resolution and difficulty in obtaining some predictor factors, a SIF reconstruction method based solely on the Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) as predictor factors is proposed, using mid-latitude forest regions as an example. This method compresses the number of predictor factors to two (EVI and LST). High-resolution satellite data for all predictor factors covers the globe (currently, publicly available EVI data has a spatial resolution of less than 10 meters, and global LST data has a resolution of 1 km), and all are freely available, improving the model's universality. Specifically, for example... Figure 1 As shown, the method is performed according to the following steps:

[0046] Step 1: Construct a pixel set of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) for global forest regions;

[0047] Step 1.1 Determine the predictors;

[0048] According to light energy utilization efficiency (LUE), the sunlight-induced chlorophyll fluorescence (SIF) can be expressed as shown in equation (1):

[0049] SIF = fPAR × PAR × ε f (1)

[0050] In equation (1), PAR represents photosynthetically active radiation; fAPAR represents photosynthetically active radiation; ε f This represents the efficiency of using absorbed radiation during photosynthesis. Equation (1) indicates that SIF, PAR, fAPAR, and ε fThe correlation is directly proportional. Therefore, vegetation conditions, meteorological conditions, and land cover information are likely indispensable factors in predicting SIF. Considering which variables are more readily available, we prioritize using land surface temperature (LST) to characterize meteorological conditions and enhance the vegetation index (EVI) or normalized difference vegetation index (NDVI) to characterize vegetation status. By analyzing the correlation between EVI, NDVI, and SIF, we found that EVI correlates more strongly with SIF than NDVI, thus identifying EVI and LST as predictors.

[0051] Step 1.2: Extract forest regions worldwide;

[0052] Obtain land cover imagery from the World Cover dataset and determine the boundaries of global forest regions from them, thereby drawing the boundary vectors (ROIs) of global forest regions. f ;

[0053] Step 1.3: Construct a pixel set of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) for global forest regions;

[0054] Using the MODIS remote sensing satellite website, the spatial range was selected to include low-latitude forest regions (0°-6°N, 9°-29°E), mid-latitude forest regions (22.6°-30.0°N, 99.0°-111.1°E), and high-latitude forest regions (57°-65°N, 36°-58°E), with the time range set from January to December 2017. Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) raster data for global forest regions were acquired from January to December 2017.

[0055] ROI of global forest regions f By overlaying raster data of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) in global forest regions, pixel sets of EVI and LST are extracted. in, and These represent the enhanced vegetation index and surface temperature of the nth pixel sample in the global forest region, respectively, where N is the number of pixels in the set.

[0056] Step 2: Construct a pixel set of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) for forest areas in the mid-latitude region;

[0057] Step 2.1: Extract forest areas within the mid-latitude region;

[0058] Obtain land cover imagery from the World Cover dataset and determine the boundaries of forest areas within the mid-latitude region, thereby drawing the boundary vectors (ROIs) of forest areas within the mid-latitude region. m ;

[0059] Step 2.2: Construct a pixel set containing the Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) of forest areas in the mid-latitude region;

[0060] Using MODIS remote sensing satellites, the spatial range was selected to include mid-latitude forest areas (30°-36°N, 99.0°-112°E), and the time range was set to January-December 2017. Enhanced vegetation index (EVI) and land surface temperature (LST) raster data for mid-latitude forest areas were acquired from January to December 2017.

[0061] The boundary vector ROI of forest areas in the mid-latitude region. m The pixel sets of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) are extracted by overlaying raster data of forest areas in the mid-latitude region. in, and These are the enhanced vegetation index and surface temperature of the nth pixel sample in the mid-latitude forest region, respectively, where N is the number of pixels in the set.

[0062] Step 3: Construct a sunlight-induced chlorophyll fluorescence (SIF) dataset.

[0063] The sunlight-induced chlorophyll fluorescence value (SIF) of the nth pixel sample is calculated using equation (2). n This yields the chlorophyll fluorescence values ​​of N pixel samples, forming the SIF dataset. <n<N;

[0064]

[0065] In equation (2), Δ L on n ΔLat represents the maximum difference in longitude of the forest region where the nth pixel sample is located; n SIF represents the maximum latitudinal difference of the forest region where the nth pixel sample is located; x,y This represents the photoinduced chlorophyll fluorescence value with relative latitude and longitude coordinates (x, y) in forest areas within global or mid-latitude regions;

[0066] Step 4: Establish a prediction model for sunlight-induced chlorophyll fluorescence (SIF) in mid-latitude forests;

[0067] Step 4.1: Use equation (3) to obtain the sunlight-induced chlorophyll fluorescence (SIF) value of the reconstructed nth pixel sample.

[0068]

[0069] In equation (3), a i bi ε represents the weight value of the i-th enhanced vegetation index and surface temperature. i Let t represent the i-th random error. i Choose a value for the i-th coefficient, and t i ∈{0,1}, where i is the number of optimal solutions; This represents the enhanced vegetation index of the nth pixel sample within a global forest region or a mid-latitude forest region. This represents the surface temperature of the nth pixel sample within a global forest region or a mid-latitude forest region, where s∈{m,f};

[0070] Step 4.2: Using equations (4) and (5), obtain the correlation R between enhanced vegetation index (EVI) and land surface temperature (LST) and sunlight-induced chlorophyll fluorescence (SIF). 2 And error RMSE:

[0071]

[0072]

[0073] Step 4.3: Solve equations (4) and (5) using the least squares method to obtain the correlation R. 2 By achieving the maximum value and the minimum RMSE value, the five mid-latitude forest daylight-induced chlorophyll fluorescence (SIF) reconstruction models with the best fitting effect and the optimal coefficient matrix were obtained. in, This represents the optimal weighting value between the i-th enhanced vegetation index and the surface temperature. This represents the i-th optimal error, where i = 5. The five mid-latitude forest daylight-induced chlorophyll fluorescence (SIF) reconstruction models are: the global monthly-scale model for January, R... 2 The value is 0.73, and the RMSE is 0.082. Figure 2a As shown; the global monthly-scale model for February, R 2 The value is 0.61, and the RMSE is 0.119. Figure 2b As shown; the global monthly-scale model for December, R 2 The value is 0.71, and the RMSE is 0.076. Figure 2c As shown; seasonal-scale model of the study area from March to June, R 2 The value is 0.60, and the RMSE is 0.129. Figure 3a As shown; seasonal-scale model of the study area from July to November, R 2 The value is 0.58, and the RMSE is 0.138. Figure 3b As shown.

[0074] Step 4.4: Obtain the optimal coefficient matrix. Substituting into equation (3), we obtain equation (6);

[0075]

[0076] In equation (6), This represents the optimal set of solutions for sunlight-induced chlorophyll fluorescence (SIF) in forest areas within the mid-latitude region after reconstruction. For SIF reconstruction in January, let t1 = 1, t... i =0 (i∈N) * |2≤i≤5); If it is a SIF reconstruction in February, let t2=1, t i =0 (i∈N) * |i=1∪3≤i≤5); If it is a SIF reconstruction from March to June, let t3=1, t i =0 (i∈N) * |1≤i≤2∪4≤i≤5); If it is a SIF reconstruction from July to November, let t4=1, t i =0 (i∈N) * |1≤i≤3∪i=5); If it is a SIF reconstruction in December, let t5=1, t i =0 (i∈N) * |1≤i≤4); Optimal coefficient matrix

[0077] N Pixels and their corresponding latitude and longitude coordinates constitute Among them, Lon n Let Lat be the longitude coordinate of the nth pixel. n Let be the latitude coordinates of the nth pixel. Construct a sunlight-induced chlorophyll fluorescence SIF array. in, The latitude and longitude coordinates are (Lon n Lat n )of Values. The corresponding pixels in the daylight-induced chlorophyll fluorescence SIF array are marked with color to plot a reconstructed SIF map of the mid-latitude forest region throughout the year, as shown below. Figure 4 As shown.

[0078] In this embodiment, an electronic device includes a memory and a processor. The memory stores a program that supports the processor in executing the above-described method, and the processor is configured to execute the program stored in the memory.

[0079] In this embodiment, a computer-readable storage medium stores a computer program, which is executed by a processor to perform the steps of the above method.

Claims

1. A method for reconstructing the SIF of mid-latitude forest land based on EVI and LST, characterized in that, The procedure is as follows: Step 1: Construct a pixel set of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) for global forest regions; Step 1.1: Extract forest regions worldwide; Obtain land cover imagery from the World Cover dataset and determine the boundaries of global forest regions from them, thereby drawing the boundary vectors (ROIs) of global forest regions. f ; Step 1.2: Construct pixel sets of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) for global forest regions; Raster data of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) for global forest regions were acquired using MODIS remote sensing satellites. The boundary vector ROI of the global forest region f By overlaying raster data of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) in global forest regions, pixel sets of EVI and LST are extracted. in, and These are the enhanced vegetation index and surface temperature of the nth pixel sample in the global forest region, respectively, where N is the number of pixel samples in the pixel set; Step 2: Construct a pixel set of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) for forest areas in the mid-latitude region; Step 2.1: Extract forest areas within the mid-latitude region; Obtain land cover imagery from the World Cover dataset and determine the boundaries of forest areas within the mid-latitude region, thereby drawing the boundary vectors (ROIs) of forest areas within the mid-latitude region. m ; Step 2.2: Construct a pixel set containing the Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) of forest areas in the mid-latitude region; Raster data of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) in forest areas within the mid-latitude region were acquired using MODIS remote sensing satellites. The boundary vector ROI of the forest area within the mid-latitude region. m The pixel sets of Enhanced Vegetation Index (EVI) and Land Surface Temperature (LST) are extracted by overlaying raster data of forest areas in the mid-latitude region. in, and These are the enhanced vegetation index and surface temperature of the nth pixel sample in the forest area within the mid-latitude region, respectively, where N is the number of pixel samples in the pixel set; Step 3: Construct a sunlight-induced chlorophyll fluorescence (SIF) dataset; The daylight-induced chlorophyll fluorescence value (SIF) of the nth pixel sample is calculated using equation (1). n This yields the chlorophyll fluorescence values ​​of N pixel samples, forming the SIF dataset. <n<N; In equation (1), Δ L on n ΔLat represents the maximum difference in longitude of the forest region where the nth pixel sample is located; n SIF represents the maximum latitudinal difference of the forest region where the nth pixel sample is located; x,y This represents the photoinduced chlorophyll fluorescence value with relative latitude and longitude coordinates (x, y) in forest areas within global or mid-latitude regions; Step 4: Establish a prediction model for sunlight-induced chlorophyll fluorescence (SIF) in mid-latitude forests; Step 4.1: Use equation (2) to obtain the sunlight-induced chlorophyll fluorescence (SIF) value of the reconstructed nth pixel sample. In equation (2), a i b i ε represents the weight value of the i-th enhanced vegetation index and surface temperature. i Let t represent the i-th random error. i Choose a value for the i-th coefficient, and t i ∈{0,1}, where i is the number of optimal solutions; This represents the enhanced vegetation index of the nth pixel sample within a global forest region or a mid-latitude forest region. This represents the surface temperature of the nth pixel sample within a global forest region or a mid-latitude forest region, where s∈{m,f}; Step 4.2: Using equations (3) and (4), obtain the correlation R between enhanced vegetation index (EVI) and land surface temperature (LST) and sunlight-induced chlorophyll fluorescence (SIF). 2 And error RMSE: Step 4.3: Solve equations (3) and (4) using the least squares method to obtain the correlation R. 2 By maximizing the maximum value and minimizing the RMSE (Recovery Mean Squared Error), the optimal coefficient matrix is ​​obtained. in, This represents the optimal weighting value between the i-th enhanced vegetation index and the surface temperature. This represents the i-th optimal error; Step 4.4: Obtain the optimal coefficient matrix. Substituting into equation (2), the optimal SIF value of sunlight-induced chlorophyll fluorescence in forest areas within the reconstructed mid-latitude region is obtained. This yields the optimal set of light-induced chlorophyll fluorescence (SIF) in mid-latitude forests.

2. An electronic device, comprising a memory and a processor, characterized in that, The memory is used to store programs that support the processor in executing the mid-latitude forest SIF reconstruction method of claim 1, and the processor is configured to execute the programs stored in the memory.

3. A computer-readable storage medium storing a computer program, characterized in that, The computer program is executed by the processor to perform the steps of the mid-latitude forest SIF reconstruction method of claim 1.

Images