A method of optimizing a vegetation photosynthesis model

Abstract

This invention discloses a method for optimizing vegetation photosynthesis models, belonging to the field of remote sensing inversion and ecosystem productivity simulation technology. This method addresses the problem that existing vegetation photosynthesis models often have fixed optimal temperature parameters for photosynthesis, failing to reflect spatiotemporal dynamic changes and thus limiting the accuracy of total primary productivity (TPP) estimation. The key points are: obtaining a dynamic optimal temperature parameter at the site scale based on the response relationship between light energy utilization efficiency and temperature; using a machine learning model to integrate multi-source feature variables and extrapolate this parameter to a global scale, generating a spatiotemporally continuous dynamic optimal temperature dataset; and replacing the original fixed optimal temperature parameter based on biome settings in the vegetation photosynthesis model with this dataset, thereby optimizing the model's representation of temperature response. This invention improves the accuracy of remote sensing estimation of TPP, and the generated optimal temperature dataset can be used for global carbon cycle simulation and climate change impact assessment.

Description

Technical Field

[0001] This invention belongs to the field of remote sensing inversion and ecosystem productivity simulation technology, and particularly relates to a method for optimizing vegetation photosynthesis models. Background Technology

[0002] Carbon absorption by terrestrial ecosystems is typically quantified using gross primary productivity (GMP). Accurately estimating GMP is crucial for understanding climate feedback and carbon flux between land and the atmosphere. Most current mainstream GMP estimation methods based on remote sensing data rely on light energy use efficiency (SEE) models, where the temperature response function is a key constraint. In these models, the optimum temperature for photosynthesis is a core parameter, significantly influencing how the model characterizes the temperature-limited nature of vegetation photosynthesis. Widely used models, such as vegetation photosynthesis models, usually set the optimum temperature as a static constant defined by biomes. However, significant uncertainties remain in characterizing the response of GMP to temperature on a large scale, posing a fundamental obstacle to the comparability and transferability of different models across multiple climate zones and across different years. This approach of treating the optimum temperature as a fixed value essentially ignores its spatiotemporal dynamics under climate change and in different geographical environments, causing models to fail to accurately capture the adaptive response of ecosystem photosynthesis to temperature changes, thus introducing systematic errors when simulating long-term changes or applying to complex geographical regions.

[0003] To overcome this limitation, existing technologies have attempted to develop methods for dynamically estimating optimum temperatures. A common approach is to directly utilize flux observation data, defining the temperature corresponding to the peak of total primary productivity (TPP) as the optimum temperature. However, this method faces fundamental difficulties in practice because the peak of TTP is often the result of the simultaneous optimization of multiple environmental conditions, such as photosynthetically active radiation, soil moisture, and phenology, rather than being solely driven by temperature; this phenomenon is known as "co-peaking." Therefore, the optimum temperature determined based on the peak of TTP is essentially an "apparent optimum temperature" confounded by multiple factors, and cannot truly reflect the intrinsic physiological optimum temperature of vegetation photosynthesis; its physical meaning is ambiguous and its universality is poor. Another improvement approach is to use vegetation indices as proxy indicators, such as using the peak value of enhanced vegetation indices to indirectly estimate the optimum temperature. While this method improves spatial applicability to some extent, since vegetation indices mainly reflect canopy greenness and structural information, they may lag or decouple from instantaneous photosynthetic physiological processes in terms of phenology. Therefore, it is still an indirect and non-mechanistic estimation scheme, which is difficult to accurately characterize the direct impact of temperature on photosynthetic biochemical processes.

[0004] In summary, current technologies, whether employing fixed optimum temperature parameters based on biological communities or dynamic estimation methods based on peak total primary productivity (TPP) or peak vegetation index, fail to provide a dynamic optimum temperature acquisition scheme that accurately reflects the physiological optimum temperature while possessing good spatiotemporal scalability, based on the core physiological mechanisms of vegetation photosynthesis. This deficiency directly restricts the simulation accuracy and reliability of light energy utilization models and various ecosystem productivity models at different spatiotemporal scales. Therefore, how to construct a photosynthetic optimum temperature estimation method based on physiological mechanisms that can avoid interference from environmental factors and achieve large-scale dynamic inversion has become a key technical problem that urgently needs to be solved to improve the accuracy of remote sensing estimation of TPP. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention proposes a method for optimizing vegetation photosynthesis models, thereby improving the accuracy of remote sensing estimation of total primary productivity of vegetation.

[0006] To achieve the above objectives, the present invention provides a method for optimizing a vegetation photosynthesis model, comprising: Obtain the optimal temperature parameters for vegetation photosynthesis in a spatiotemporal dynamic manner; The spatiotemporal dynamic optimal temperature parameters for vegetation photosynthesis are input into the vegetation photosynthesis model, replacing the original fixed optimal temperature parameters set based on the biological community in the vegetation photosynthesis model. Based on the vegetation photosynthesis model with replaced parameters, the total primary productivity is estimated.

[0007] Optionally, obtaining the optimal temperature parameters for vegetation photosynthesis in spatiotemporal dynamics includes: Based on the response relationship between light energy utilization efficiency and temperature, the optimal temperature parameters at the station scale are obtained by inversion from the station observation data.

[0008] Optionally, the process of retrieving the optimal temperature parameters at the site scale from site observation data based on the response relationship between light energy utilization efficiency and temperature includes: Calculate the solar energy utilization efficiency of the site; Dates that achieve a preset high percentile in light energy utilization efficiency are selected as candidate dates; From the candidate dates, select dates whose daily average temperature falls within the preset high percentile of the annual temperature distribution of that station. The average temperature of the selected dates is determined as the optimal temperature parameter at the station scale.

[0009] Optionally, the method further includes: Build a machine learning regression model; The model is trained by using multiple annual average characteristic variables related to vegetation physiology and climate as inputs to the machine learning regression model and the optimal temperature parameter at the station scale as the output target. Using the trained machine learning regression model, the optimal temperature parameters at the site scale are extrapolated to a continuous spatial scale to obtain the optimal temperature parameters for vegetation photosynthesis in the spatiotemporal dynamics.

[0010] Optional annual average characteristic variables related to vegetation physiology and climate include: saturated vapor pressure gradient, carbon dioxide concentration, photosynthetically active radiation absorption ratio, evapotranspiration, and downward shortwave radiation.

[0011] Optionally, the machine learning regression model is a random forest model.

[0012] Optionally, the vegetation photosynthesis model is a light energy utilization efficiency model; the replacement refers to replacing the fixed optimal temperature parameter in the temperature response function of the light energy utilization efficiency model with the spatiotemporally dynamic optimal temperature parameter for vegetation photosynthesis.

[0013] Optionally, the method further includes: generating grid data products covering a specified time period and geographical range based on the optimal temperature parameters for vegetation photosynthesis in the spatiotemporal dynamics.

[0014] An electronic device, the electronic device comprising: a processor and a memory storing computer program instructions; The processor implements the method for optimizing a vegetation photosynthesis model when executing the computer program instructions.

[0015] A computer storage medium storing computer program instructions, which, when executed by a processor, implement the method for optimizing a vegetation photosynthesis model.

[0016] Technical Effects of this Invention: This invention discloses a method for optimizing vegetation photosynthesis models, improving the accuracy of remote sensing estimation of total primary productivity of vegetation. By replacing the fixed parameters in the original model with dynamic optimal temperature parameters obtained from light energy utilization efficiency inversion, the model can more accurately characterize the spatiotemporal dynamic response of vegetation photosynthesis to temperature. This invention generates a globally continuous optimal temperature dataset that can be directly transferred and used, possessing both universality and adaptability, providing key parameter support for ecosystem productivity research, vegetation adaptability analysis, and global carbon cycle simulation. This method does not rely on intensive field observations, but mainly utilizes multi-source remote sensing and reanalysis data, achieving simple and efficient large-scale parameter estimation. Furthermore, this invention's paradigm of parameter optimization and extrapolation based on a process mechanism framework combined with machine learning has good scalability, providing a feasible technical approach for improving the simulation of land-climate processes in other Earth system models. Attached Figure Description

[0017] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a schematic diagram illustrating the estimation of optimal temperature using light energy utilization efficiency and air temperature in an embodiment of the present invention; Figure 2 The above is a scatter plot of the actual and estimated values ​​of the embodiments of the present invention, where (A) is the model fitting accuracy and (B) is the prediction accuracy. Figure 3 This is a schematic diagram comparing the observed GPP of the flux tower with the model-estimated GPP in an embodiment of the present invention, where (A) is the estimation result of the original VPM model; and (B) is the estimation result of the VPM model adjusted using Topt-LUE. Figure 4 This is a global Topt-LUE spatial distribution map based on the random forest regression algorithm in an embodiment of the present invention, where (A) is the global Topt-LUE spatial differentiation characteristics; (B) is the latitudinal distribution characteristics of Topt-LUE; and (C) is the Topt-LUE box plot for different vegetation types. Figure 5 This is a schematic diagram illustrating the seasonal variation of photosynthetic variables at an exemplary evergreen coniferous forest (ENF) site of the present invention. Figure 6 The following is a global distribution map of the daily average GPP difference between the adjusted VPM and the original VPM in this embodiment of the invention. (A) shows the spatial distribution differentiation characteristics of the GPP difference before and after the VPM model adjustment; (B) shows the latitudinal variation characteristics of the GPP difference between the two models. Figure 7 This is a flowchart illustrating a method for optimizing a vegetation photosynthesis model according to an embodiment of the present invention. Detailed Implementation

[0018] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0019] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0020] like Figure 7 As shown, this embodiment provides a method for optimizing a vegetation photosynthesis model, including: Obtain the optimal temperature parameters for vegetation photosynthesis in a spatiotemporal dynamic manner; The spatiotemporal dynamic optimal temperature parameters for vegetation photosynthesis are input into the vegetation photosynthesis model, replacing the original fixed optimal temperature parameters set based on the biological community in the vegetation photosynthesis model. Based on the vegetation photosynthesis model with replaced parameters, the total primary productivity is estimated.

[0021] Furthermore, obtaining the optimal temperature parameters for vegetation photosynthesis in a spatiotemporal dynamic manner includes: Based on the response relationship between light energy utilization efficiency and temperature, the optimal temperature parameters at the station scale are obtained by inversion from the station observation data.

[0022] Specifically, the implementation process of this embodiment includes: Obtain the PAR and FAPAR datasets from the GLASS data platform. Based on the geographic coordinates of the flux tower locations, spatially match these raster data with flux stations in the FLUXNET2015 dataset; calculate the LUE for each station-year. Then, select dates in the annual LUE distribution that reach or exceed the 90th percentile. Define Topt as the average temperature of the days among these candidate dates whose daily average temperature is within the 90th percentile of the annual temperature distribution for that station.

[0023] Furthermore, based on the response relationship between light energy utilization efficiency and temperature, the process of retrieving the optimal temperature parameters at the site scale from site observation data includes: Calculate the solar energy utilization efficiency of the site; Dates that achieve a preset high percentile in light energy utilization efficiency are selected as candidate dates; From the candidate dates, select dates whose daily average temperature falls within the preset high percentile of the annual temperature distribution of that station. The average temperature of the selected dates is determined as the optimal temperature parameter at the station scale.

[0024] Specifically, the implementation process of this embodiment includes: Topt values ​​are obtained by inverting the LUE-Temperature response relationship. Most LUE models are represented in the following form: (1); The PAR (Photosynthetically Active Radiation) and FPAR (Photosynthetically Active Radiation Absorption Ratio) data were obtained from the GLASS data platform, with a spatial resolution of 0.5° and a time range of 2000–2020. PAR data was daily data, and FAPAR was an 8-day composite product. Air temperature data were obtained from the fifth-generation ECMWF reanalysis product (ERA5), with an hourly temporal resolution and a spatial resolution of 0.25°. First, the LUE (Light Energy Efficiency) for each site-year was calculated. Then, dates reaching or exceeding the 90th percentile of the annual LUE distribution were selected. For these high-LUE days, the corresponding daily average temperature was extracted from FLUXNET2015. Considering that the annual temperature range in some regions (especially cold regions) may not fully cover the optimal temperature for photosynthesis, this embodiment further defines Topt as the average temperature of the days among these candidate dates whose daily average temperature is within the 90th percentile of the annual temperature distribution for that site. This two-step selection criterion ensures that Topt reflects both high light energy utilization efficiency and suitable temperature conditions.

[0025] like Figure 1 The diagram illustrates the process of estimating the optimal temperature (Topt) using light energy utilization efficiency (LUE) and air temperature (Tair). The horizontal axis represents the daily average air temperature, and the vertical axis represents the daily LUE value. Gray dots represent daily LUE data from 2009 to 2013; dots represent the 90th percentile of LUE calculated for each 3% interval of temperature, with red dots representing the two points with the highest LUE among these quantiles.

[0026] Furthermore, the method also includes: Build a machine learning regression model; The model is trained by using multiple annual average characteristic variables related to vegetation physiology and climate as inputs to the machine learning regression model and the optimal temperature parameter at the station scale as the output target. Using the trained machine learning regression model, the optimal temperature parameters at the site scale are extrapolated to a continuous spatial scale to obtain the optimal temperature parameters for vegetation photosynthesis in the spatiotemporal dynamics.

[0027] Specifically, the implementation process of this embodiment includes: The saturated vapor pressure difference (VPD), CO2 concentration (CO2), photosynthetically active radiation absorption ratio (FAPAR), evapotranspiration (ET), and downward shortwave radiation (DSR) were obtained; a random forest model was constructed based on decision tree analysis to estimate the global Topt. After calculating the Topt-LUE at the site scale, a random forest model was used to extrapolate the Topt-LUE to a global scale. The random forest (RF) algorithm is a decision tree-based ensemble learning method used for classification and regression tasks. This method improves the accuracy and robustness of the model by constructing multiple decision trees and integrating their results. This invention selected five annual average feature variables as input independent variables, including average saturated vapor pressure difference (Mean_VPD), average CO2 concentration (Mean_CO2), average vegetation absorbance of photosynthetically active radiation (Mean_FAPAR), average downward shortwave radiation (Mean_DSR), and average evapotranspiration (Mean_ET) (summarized from the GLASS11G01.V50 dataset). The site-scale Topt-LUE was used as the dependent variable input to the random forest model. The same regression method as for Topt-LUE was further applied to Topt-GPP and Topt-EVI. The results showed that Topt-GPP had low prediction accuracy (R²=0.54), significantly lower than Topt-LUE (R²=0.82). This indicates that the optimal temperature determined based on GPP peak values ​​lacks universality. In contrast, Topt-EVI and Topt-LUE both showed better regression performance (R²=0.82), suggesting their potential for estimating the global optimum temperature for photosynthesis.

[0028] Figure 2 This is a scatter plot of the actual and estimated Topt values, where the x-axis represents the actual Topt and the y-axis represents the estimated Topt. (A) represents the model fit accuracy, and (B) represents the prediction accuracy. The actual Topt was calculated using the LUE model, while the estimated Topt was obtained by regressing Topt-LUE using a random forest model. The red dashed line represents the y=x (1:1) reference line. 80% of the data was used for model training, and the remaining 20% ​​was used to evaluate the model's fit and prediction accuracy.

[0029] Furthermore, the vegetation photosynthesis model is a light energy utilization efficiency model; the replacement refers to replacing the fixed optimal temperature parameter in the temperature response function of the light energy utilization efficiency model with the spatiotemporally dynamic optimal temperature parameter for vegetation photosynthesis.

[0030] Specifically, the implementation process of this embodiment includes: Fix the original biological community in the VPM model to T opt Parameters replaced with LUE-based global Topt Inversion results; the performance of the VPM model optimization was evaluated using observation data from 82 flux stations.

[0031] The VPM model is based on photosynthetically active radiation (PAR) and the proportion of vegetation photosynthetically active radiation absorbed by chlorophyll (FPAR). chl ) and light energy utilization efficiency (ε g To estimate GPP, use FPAR as shown in Formula 2. chl The enhanced vegetation index (EVI) is calculated using Equation 3. εg is derived from the temperature limiting factor (T). scalar ) and water stress factor (W scalar The maximum value (ε0) is adjusted downwards, as shown in Formula 4. The values ​​of ε0 are 0.42 gC / mol for C3 plants and 0.63 gC / mol for C4 plants. Wscalar is based on the Surface Water Index (LSWI) and its annual maximum value (LSWI). max The calculation is performed using Formula 5. In the original VPM model, the minimum temperature (T) min ), highest temperature (T) max ) and optimal temperature (T) opt All values ​​are set to fixed values.

[0032] (2); (3); (4); (5); (6).

[0033] In the improved VPM model, the T value in the original VPM model, which was based on the biological community setting, is now lower. opt Replaced with the global optimal temperature (T) based on random forest prediction. opt Due to the minimum temperature (T) min ) and maximum temperature (T) max The spatial distribution of [the organism] is still unclear, and the improved model still uses the biome settings from the original VPM model. Therefore, the difference between the original model and the improved model in the GPP estimation results is only due to T [the difference in values]. opt The replacement caused this. In this embodiment, the original and improved VPM models were run to estimate the GPP on an 8-day scale, and the results were compared with the GPP observed by the flux tower.

[0034] Figure 3This diagram illustrates the comparison between observed GPP and model-estimated GPP in flux towers. (A) shows the estimation result from the original VPM model; (B) shows the estimation result from the VPM model adjusted using Topt-LUE. The black dashed line represents the 1:1 reference line, and the red solid line represents the regression relationship between the observed and estimated values.

[0035] Furthermore, the method also includes: generating grid data products covering a specified time period and geographical range based on the optimal temperature parameters for vegetation photosynthesis in the spatiotemporal dynamics.

[0036] Specifically, the implementation process of this embodiment includes: Construct the Topt-LUE dataset from 2001 to 2020; analyze the spatiotemporal characteristics of optimal temperature for different vegetation types, regions, and latitudes; and compare the performance of different algorithms in estimating optimal temperature. Analyze the spatiotemporal pattern of GPP from 2001 to 2020; compare the performance improvements of models that estimate GPP based on LUE.

[0037] Based on a random forest model, using five variables—VPD, CO2, FAPAR, DSR, and ET—Topt-LUE at the site scale is extrapolated to global vegetation regions, generating global Topt-LUE data with a 0.5° spatial resolution for the period 2001–2020. The average results from 2001–2020 are presented as an example. Figure 4 The mean Topt-LUE estimates for global vegetation zones range from 4°C to 30°C, exhibiting significant spatial heterogeneity. Figure 4 (A)). The highest Topt-LUE (approximately 30°C) is mainly distributed in central Africa; while the lowest values ​​are mainly distributed in high-latitude regions. The latitudinal distribution of Topt also shows a clear geographical dependence. Figure 4 (B)). Among the 12 vegetation types analyzed, EBF had the highest mean Topt-LUE (24.9 ± 2.2°C), while the mean Topt-LUE for other vegetation types ranged from 10 to 25°C. Figure 4 (C)). These results indicate that, compared to fixed Topt based on biological communities, spatiotemporally dynamic Topt-LUE can more accurately characterize the global vegetation response to temperature.

[0038] Figure 4The image shows the global Topt-LUE spatial distribution map based on the random forest regression algorithm, with a spatial resolution of 0.5°. The white areas in the land region represent arid and semi-arid regions with low vegetation cover. (A) shows the spatial differentiation characteristics of global Topt-LUE; (B) shows the latitudinal distribution characteristics of Topt-LUE; and (C) shows the Topt-LUE box plots for different vegetation types, where red dots represent fixed Topts defined based on biomes in the original VPM model.

[0039] The method based on light energy utilization efficiency (LUE) proposed in this embodiment effectively solves the problem of the traditional optimal photosynthetic temperature (T0). opt The fundamental limitation of estimation methods. Traditional methods typically take T... opt Defined as the temperature at which air temperature and GPP peak occur simultaneously, but this method is inherently subject to the "co-peak phenomenon" ( Figure 5 The interference of vegetation indices (such as EVI) can lead to peak carbon uptake. Under these conditions, the peak carbon uptake is often the result of simultaneous optimization of favorable conditions such as radiation, soil moisture, and phenology, rather than a physiologically optimal temperature driven solely by temperature. Although surrogate methods based on vegetation indices (such as EVI) have improved spatial applicability and avoided some of the problems associated with directly determining Topt from GPP peak values, these methods are still indirect estimations. Vegetation indices mainly reflect canopy greenness and structure, but due to phenological lag, these indicators may become decoupled from transient photosynthetic physiological processes.

[0040] Figure 5 This diagram illustrates the seasonal variation of photosynthetic variables at an evergreen coniferous forest (ENF) site. PAR, EVI, FAPAR, and canopy water status are used as environmental inputs, while GPP and LUE are derived through model relationships between these variables. The vertical lines represent the time points of apparent optima (peak GPP) and physiological optima (peak LUE). All variables are normalized to their seasonal maximum values; for ease of observation, the Water, EVI, PAR, and FAPAR curves are offset vertically.

[0041] The optimal temperature based on LUE is defined as the temperature at which LUE reaches its maximum value, thus providing a direct physiological indicator of temperature optimality. Unlike vegetation indices or chlorophyll fluorescence (ΦF), LUE explicitly reflects the efficiency of converting absorbed light energy into chemical energy for carbon fixation, while integrating the photochemical and biochemical stages of photosynthesis, including electron transport, Rubisco carboxylation, and CO2 assimilation. These coupled processes collectively determine the net carbon fixation rate and are highly sensitive to temperature. As temperature increases, enzymatic reactions and electron transport accelerate until a thermal threshold is reached; above this temperature, protein denaturation, increased photorespiration, and decreased Rubisco activity lead to a decline in photosynthetic efficiency, resulting in a typical single-peak response curve of LUE to air temperature. The peak value reflects the balance between the activation and inhibition processes of photosynthesis. Therefore, the temperature at which LUE reaches its peak corresponds to the moment when the overall conversion efficiency of the photosynthetic system is at its highest, representing the true physiological optimum temperature, rather than the apparent optimum.

[0042] In colder regions, the adjusted VPM yielded a higher GPP estimate ( Figure 6 This phenomenon further confirms that ecosystem productivity in cold regions is primarily limited by low temperatures. With climate warming, this temperature limitation is alleviated, the growing season is extended, and the "greening" effect in the Arctic region further promotes increased vegetation productivity. In contrast, in tropical regions, especially the Amazon rainforest, the adjusted VPM generates a lower GPP estimate because Topt-LUE is lower than the Topt based on the biome setting. This suggests that changes in Topt in these regions may be moderated by non-temperature factors, such as water stress or high-temperature inhibition, ultimately leading to a lower GPP estimate for the adjusted VPM compared to the original VPM. These results demonstrate that incorporating spatiotemporally dynamic Topt-LUE is more effective in improving VPM performance compared to the traditional biome-fixed Topt method.

[0043] Figure 6 The global distribution map of the daily average GPP difference between the adjusted VPM and the original VPM is shown, with the average value from 2001 to 2020 and a spatial resolution of 0.5°. (A) shows the global spatial differentiation characteristics of the GPP difference before and after the VPM model adjustment; (B) shows the latitudinal variation characteristics of the GPP difference before and after the VPM model adjustment.

[0044] An electronic device, the electronic device comprising: a processor and a memory storing computer program instructions; The processor implements the method for optimizing a vegetation photosynthesis model when executing the computer program instructions.

[0045] A computer storage medium storing computer program instructions, which, when executed by a processor, implement the method for optimizing a vegetation photosynthesis model.

[0046] This invention discloses a method for optimizing vegetation photosynthesis models, improving the accuracy of remote sensing estimation of total primary productivity (TPP) of vegetation. By replacing the fixed parameters in the original model with dynamic optimal temperature parameters derived from light energy utilization efficiency inversion, the model can more accurately characterize the spatiotemporal dynamic response of vegetation photosynthesis to temperature. This invention generates a globally continuous optimal temperature dataset that can be directly transferred and used, possessing both universality and adaptability, providing key parameter support for ecosystem productivity research, vegetation adaptability analysis, and global carbon cycle simulation. This method does not rely on intensive field observations, but mainly utilizes multi-source remote sensing and reanalysis data, achieving simple and efficient large-scale parameter estimation. Furthermore, the paradigm of parameter optimization and extrapolation based on a process mechanism framework combined with machine learning has good scalability, providing a feasible technical approach for improving the simulation of land-climate processes in other Earth system models.

[0047] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for optimizing a vegetation photosynthesis model, characterized in that, include: Obtain the optimal temperature parameters for vegetation photosynthesis in a spatiotemporal dynamic manner; The spatiotemporal dynamic optimal temperature parameters for vegetation photosynthesis are input into the vegetation photosynthesis model, replacing the original fixed optimal temperature parameters set based on the biological community in the vegetation photosynthesis model. Based on the vegetation photosynthesis model with replaced parameters, the total primary productivity is estimated.

2. The method for optimizing a vegetation photosynthesis model as described in claim 1, characterized in that, Obtaining the optimal temperature parameters for vegetation photosynthesis in a spatiotemporal dynamic manner includes: Based on the response relationship between light energy utilization efficiency and temperature, the optimal temperature parameters at the station scale are obtained by inversion from the station observation data.

3. The method for optimizing a vegetation photosynthesis model as described in claim 2, characterized in that, Based on the response relationship between light energy utilization efficiency and temperature, the process of retrieving the optimal temperature parameters at the site scale from site observation data includes: Calculate the solar energy utilization efficiency of the site; Dates that achieve a preset high percentile in light energy utilization efficiency are selected as candidate dates; From the candidate dates, select dates whose daily average temperature falls within the preset high percentile of the annual temperature distribution of that station. The average temperature of the selected dates is determined as the optimal temperature parameter at the station scale.

4. The method for optimizing a vegetation photosynthesis model as described in claim 2, characterized in that, The method further includes: Build a machine learning regression model; The model is trained by using multiple annual average characteristic variables related to vegetation physiology and climate as inputs to the machine learning regression model and the optimal temperature parameter at the station scale as the output target. Using the trained machine learning regression model, the optimal temperature parameters at the site scale are extrapolated to a continuous spatial scale to obtain the optimal temperature parameters for vegetation photosynthesis in the spatiotemporal dynamics.

5. The method for optimizing a vegetation photosynthesis model as described in claim 4, characterized in that, Several annual average characteristic variables related to vegetation physiology and climate include: saturated vapor pressure difference, carbon dioxide concentration, photosynthetically active radiation absorption ratio, evapotranspiration, and downward shortwave radiation.

6. The method for optimizing a vegetation photosynthesis model as described in claim 4, characterized in that, The machine learning regression model is a random forest model.

7. The method for optimizing a vegetation photosynthesis model as described in claim 1, characterized in that, The vegetation photosynthesis model is a light energy utilization efficiency model; the replacement refers to replacing the fixed optimal temperature parameter in the temperature response function of the light energy utilization efficiency model with the spatiotemporally dynamic optimal temperature parameter for vegetation photosynthesis.

8. The method for optimizing a vegetation photosynthesis model as described in claim 1, characterized in that, The method further includes generating grid data products covering a specified time period and geographical range based on the optimal temperature parameters for vegetation photosynthesis in the spatiotemporal dynamics.

9. An electronic device, characterized in that, The electronic device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements a method for optimizing a vegetation photosynthesis model as described in any one of claims 1-8.

10. A computer storage medium, characterized in that, The computer storage medium stores computer program instructions, which, when executed by a processor, implement a method for optimizing a vegetation photosynthesis model as described in any one of claims 1-8.

Images