Surface temperature reconstruction method, system, equipment and medium
By combining the data assimilation method of MODIS satellite and micrometeorological site data, the Noah-MP model is used to update the surface temperature, which solves the problems of spatial and spatial continuity and spatial resolution in remote sensing observations, and achieves efficient and accurate spatial and temporal continuity of surface temperature in mountainous areas Observation.
Patent Information
- Application Number
- CN202410880092.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-02
- Publication Date
- 2024-11-08
AI Technical Summary
The spatial and spatial continuity and spatial resolution of surface temperatures of remote sensing observations in the prior art are not possible, especially in the case of cloudy and precipitation, and the spatial resolution of surface temperatures in mountainous areas is poor, and there is a lack of comprehensive land surface process parameters support.
Using high-space resolution MODIS satellite sensor and micrometeorological site data combined with Noah-MP model, the surface temperature simulation value is updated through data assimilation methods (such as ensemble Kalman filtering), and the high-space resolution spatiotemporal continuous surface temperature data are obtained, and the High resolution features are continued by weighted average when there is no remote sensing observation data.
While achieving the spatial and temporal continuity and high spatial resolution of surface temperature in mountainous areas, the accuracy and application scenarios of surface temperature observation are improved, and the problem of overestimation of surface temperature synthesis is reduced. It is suitable for different terrain and vegetation types.
Smart Images

Figure CN118917044A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the fields of remote sensing technology and time-space reconstruction, and in particular to a method, system, equipment and medium for reconstructing surface temperature. Background Art
[0002] Land surface temperature LST is an important variable that characterizes the energy balance state between the surface and the atmospheric boundary layer. It has an important impact on the energy balance state of the earth's surface, changes in extreme weather events, and the health and stability of the ecosystem. The land surface temperature LST based on remote sensing can be divided into bare soil, dense vegetation, and sparse vegetation surface temperatures according to different underlying surface types. The advantages of obtaining land surface temperature LST by remote sensing mainly include large spatial coverage, fast data acquisition speed, short cycle, and sensitivity to surface temperature. It can truly reflect the spatial heterogeneity of surface temperature and can clearly characterize and reflect the differences in surface temperature between pixels, that is, the spatial heterogeneity of surface temperature LST.
[0003] In the prior art, the LST methods for obtaining land surface temperature through thermal infrared remote sensing mainly include dual-channel split window calculation, single-channel algorithm and temperature emissivity separation algorithm. The land surface temperature inverted by polar orbit satellites using the above methods can achieve a very high spatial resolution; but its disadvantage is that it is instantaneous observation and it is often difficult to achieve continuous observation in time. Although geostationary meteorological satellites can achieve continuous observation every ten minutes, the influence of clouds and rain will still cause a large number of missing pixels in the land surface temperature inverted by thermal infrared remote sensing, making its continuity in space worse, and thus not continuous in time and space, especially in mountainous areas. The most important application processing method is to obtain the land surface temperature of 8 days, 16 days and months based on the synthesis method of maximum value, mean value, etc. However, such a method can easily lead to random overestimation of the land surface temperature value, which brings great uncertainty to the further application of thermal infrared remote sensing inversion of land surface temperature. In order to make up for the above shortcomings and obtain the land surface temperature that is continuous in time and space, a series of data fusion algorithms have been proposed, mainly including reconstruction methods that rely on space, time or spatiotemporal information, algorithms based on surface energy balance and algorithms based on multivariate remote sensing data. However, the fusion model of simple multivariate remote sensing data will limit the temporal continuity of the output results. The land surface process model can enrich the temporal information of remote sensing data, but it is often less applicable to high spatial resolution numerical simulations. The fusion framework that combines remote sensing products with land surface model output variables is used to obtain hourly surface temperature data. It mainly focuses on matching similar pixel values of remote sensing surface temperature with model output results, which is limited by the normalized matching of model output variables with remote sensing pixels at corresponding positions. Based on the land surface process model data assimilation method, remote sensing observation data is dynamically fused, and spatial heterogeneity information of different underlying surfaces is added to the land surface process model suitable for coarse resolution. It can also construct the temporal continuity of remote sensing observation surface temperature. For mountain systems such as the Qilian Mountains-Altun Mountains-Kunlun Mountains, there are currently no comprehensive land surface process parameters to support the accurate simulation of surface process models, and real measurements of aerodynamic roughness length, emissivity, etc. are urgently needed.
[0004] However, the above methods cannot continuously remotely sense the surface temperature under cloudy and precipitation conditions, and the observed surface temperature has poor spatial resolution. Summary of the invention
[0005] The embodiments of the present invention provide a method, system, device and medium for reconstructing land surface temperature, which can solve the problems of poor spatiotemporal continuity and spatial resolution of land surface temperature observed by remote sensing in the prior art.
[0006] The embodiment of the present invention provides a method for reconstructing land surface temperature, comprising the following steps: using a high spatial resolution MODIS satellite sensor to observe the land surface to obtain instantaneous land surface temperature observation data; simultaneously obtaining land surface temperature observation data observed by a micro-meteorological station; inputting a land surface variable data set ERA5-Land at a plurality of consecutive moments into a Noah-MP model for simulating land surface temperature according to land surface conditions, and obtaining land surface temperature simulation values at a plurality of consecutive moments; evaluating the simulation value error between the land surface temperature simulation value and the land surface temperature observation data, and evaluating the observation value error between the instantaneous land surface temperature observation data and the land surface temperature observation data; when a MODIS When the satellite sensor can observe the surface, the simulated value and the observed data are assimilated according to the surface temperature simulation value and the simulation value error at the current moment, as well as the instantaneous surface temperature observation data and the observation value error at the current moment; and the surface temperature simulation value at the current moment is updated according to the assimilated data, so that the updated surface temperature simulation value has the high spatial resolution characteristics of the instantaneous surface temperature observation data; when the MODIS satellite sensor cannot observe the surface at a certain moment, the surface temperature simulation value updated at the previous moment and the surface temperature simulation value at the current moment are weighted averaged, and the surface temperature simulation value at the current moment is updated, so that the high spatial resolution characteristics can be continued.
[0007] Furthermore, the data assimilation of the simulation value and the observation data specifically includes: using an ensemble Kalman filter data assimilation method to perform data assimilation of the simulation value and the observation data.
[0008] Furthermore, the method of updating the surface temperature simulation value at the current moment according to the assimilated data specifically comprises the following steps: obtaining instantaneous surface temperature observation data, obtaining surface temperature simulation values according to the Noah-MP model, obtaining simulation value errors and observation value errors using an error evaluation method; constructing a prediction error covariance matrix using surface temperature simulation values and simulation value errors, and constructing an observation error covariance matrix using instantaneous surface temperature observation data and observation value errors; obtaining a Kalman gain matrix according to ensemble Kalman filter data assimilation; and updating the surface temperature simulation value according to the Kalman gain matrix.
[0009] Furthermore, the Noah-MP model also includes updating the aerodynamic roughness length and soil emissivity in the input parameters before the Noah-MP model is used. The specific steps are: based on the observation data of the micrometeorological station, the aerodynamic roughness length is estimated by using the logarithmic wind speed method on the underlying surface types inlaid with open shrubs, sparse grasslands, bare soil or sparse vegetation, grasslands, irrigated farmland and cultivated land with weeds; based on the remote sensing observation data of the MODIS satellite sensor, the soil emissivity is estimated by using the averaging method or the soil emissivity is measured using a Fourier transform infrared spectrometer.
[0010] Furthermore, the acquisition of instantaneous surface temperature observation data specifically includes: observing two surface temperature and soil emissivity data products MYD11A1 and MOD11A1 of the high spatial resolution MODIS satellite sensor to obtain daily surface temperature data and soil emissivity data with a high spatial resolution of 1 km.
[0011] An embodiment of the present invention provides a system for reconstructing land surface temperature, including:
[0012] The data acquisition module is used to observe the surface using high spatial resolution MODIS satellite sensors to obtain instantaneous surface temperature observation data; at the same time, it obtains surface temperature observation data observed by micro-meteorological stations; the error evaluation module is used to input the land surface variable dataset ERA5-Land at multiple consecutive moments into the Noah-MP model used to simulate the surface temperature according to the land surface conditions, and obtain the surface temperature simulation values at multiple consecutive moments; evaluate the simulation value error of the surface temperature simulation value and the surface temperature observation data, and evaluate the observation value error of the instantaneous surface temperature observation data and the surface temperature observation data; the surface temperature reconstruction module is used to When the S satellite sensor can observe the surface, the simulated value and the observed data are assimilated according to the surface temperature simulation value and the simulation value error at the current moment, as well as the instantaneous surface temperature observation data and the observation value error at the current moment; and the surface temperature simulation value at the current moment is updated according to the assimilated data, so that the updated surface temperature simulation value has the high spatial resolution characteristics of the instantaneous surface temperature observation data; when the MODIS satellite sensor cannot observe the surface at a certain moment, the surface temperature simulation value updated at the previous moment and the surface temperature simulation value at the current moment are weighted averaged, and the surface temperature simulation value at the current moment is updated, so that the high spatial resolution characteristics can be continued.
[0013] An embodiment of the present invention provides a computer device, comprising: a memory and a processor; the memory stores a computer program, and the processor implements the above-mentioned method for reconstructing land surface temperature when executing the computer program.
[0014] An embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned method for reconstructing the ground surface temperature is implemented.
[0015] The embodiments of the present invention provide a method, system, device and medium for reconstructing land surface temperature. Compared with the prior art, the embodiments of the present invention have the following beneficial effects:
[0016] The land surface variable dataset ERA5-Land at multiple consecutive moments is input into the Noah-MP model used to simulate the surface temperature according to the land surface conditions, and the surface temperature simulation values at multiple consecutive moments are obtained; the simulation value error between the surface temperature simulation value and the surface temperature observation data is evaluated, and the observation value error between the instantaneous surface temperature observation data and the surface temperature observation data is evaluated; when the MODIS satellite sensor can observe the surface at a certain moment, the simulation value and the observation data are assimilated according to the surface temperature simulation value at the current moment and the simulation value error, and the instantaneous surface temperature observation data at the current moment and the observation value error; and the surface temperature simulation value at the current moment is updated according to the assimilated data, so that the updated surface temperature simulation value has the high spatial resolution characteristics of the instantaneous surface temperature observation data; when the MODIS satellite sensor cannot observe the surface at a certain moment, the surface temperature simulation value updated at the previous moment and the surface temperature simulation value at the current moment are weighted averaged, and the surface temperature simulation value at the current moment is updated, so that the high spatial resolution characteristics can be continued.
[0017] Among them, when the MODIS satellite sensor is able to observe the surface at a certain moment, the surface temperature simulation value at the current moment is updated according to the assimilated data, so that the updated surface temperature simulation value has the high spatial resolution characteristics of the instantaneous surface temperature observation data; when the MODIS satellite sensor cannot observe the surface conditions at a certain moment, the surface temperature simulation value at the current moment is updated according to the surface temperature simulation value updated at the previous moment, so that the high spatial resolution characteristics can be continued, ensuring that regardless of whether the surface can be observed or not, the updated surface temperature simulation value with high spatial resolution can be obtained, and finally the obtained surface temperature simulation value has high spatial resolution while maintaining temporal and spatial continuity. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] Figure 1 The spatial distribution map of sites in the Kunlun Mountains-Altun Mountains-Qilian Mountains region provided by the embodiment of the present invention ( Figure 1 a is the undulating topographic map of the tower; Figure 1 b is the distance between flux stations in undulating terrain; Figure 1 c is the site distribution map; Figure 1 d is a real picture of Huazhaizi Station, Daman Station, Kalasai Station and Arou Station);
[0019] Figure 2 The frequency distribution diagram of aerodynamic roughness length of different underlying surfaces of a method for reconstructing surface temperature provided by an embodiment of the present invention ( Figure 2 a is the Kalasai station; Figure 2 b is the growing season of Daman Station; Figure 2 c is the non-growing season of Daman Station; Figure 2 d is Tazhong West Station; Figure 2e is the growing season of the shrubbery in the central Tarim shelterbelt; Figure 2 f is the non-growing season of the protective forest shrubs in the central part of the tower);
[0020] Figure 3 A spatiotemporal reconstruction framework diagram of the land surface temperature at 1 km in a mountainous area based on data assimilation according to a land surface temperature reconstruction method provided by an embodiment of the present invention;
[0021] Figure 4 A rectangular range map of the Qilian Mountains for a method for reconstructing land surface temperature provided by an embodiment of the present invention;
[0022] Figure 5 A comparison diagram of the surface temperature before and after reconstruction of the rectangular range of Qilian Mountain according to a surface temperature reconstruction method provided by an embodiment of the present invention ( Figure 5 a is the assimilation result from 21:00 on July 11 to 03:00 on July 12, 2021. The red rectangle is the surface temperature observed by MODIS during assimilation. The names starting with "MOD11A1" or "MYD11A1" are remote sensing LST observations; Figure 5 b is the coarse-resolution LST of the Noah-MP simulation without assimilation). DETAILED DESCRIPTION
[0023] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific embodiments disclosed below.
[0024] This study takes advantage of the dynamic fusion and multi-parameterized simulation of DA and constructs a data assimilation downscaling framework (DADF). Compared with the fusion reanalysis data method, the DADF method fully utilizes the dynamic simulation capability of the land surface process model and can reasonably retain the fine spatial heterogeneity information of high spatial resolution observation data. This DADF that assimilates high spatial resolution remote sensing data can fully express the differences between sub-pixels in large pixels. DADF can dynamically integrate remote sensing observations, add spatial heterogeneity information of different underlying surfaces to LSMs suitable for coarse resolution analysis, and reconstruct the temporal continuity of LST remote sensing observations. However, there are four issues that need to be paid attention to when assimilating high spatial resolution land surface temperature with DADF. First, the accuracy of the reconstruction is mainly determined by the simulation accuracy of the land surface process model. Studies have shown that emissivity and aerodynamic roughness length have an important impact on surface temperature simulation and energy balance. Before assimilating remote sensing observation data, parameterization correction of emissivity and aerodynamic roughness length of different underlying surface types is the guarantee of assimilation accuracy. Second, when there is no remote sensing observation data, how can the improved spatial resolution be sustained? The factors that determine the spatial resolution of the surface temperature of the land surface process model are mainly the resolution of the forcing data, the spatial resolution of the underlying surface type (determines the emissivity), and the initial spatial resolution of the surface temperature at the previous simulation moment. The surface temperature is a variable that responds quickly to the forcing data, so the spatial resolution of the forcing data has the greatest impact on the resolution of the model output. At present, the spatial resolution of ERA5-land, which has the highest spatial resolution on a global scale, is only about 10km, which cannot meet the 1km simulation of polar-orbiting satellites. Based on the assumption that the surface temperature changes as a continuous curve, we constructed a strategy to weight the previous simulation moment to make the improved spatial resolution of the assimilation moment sustainable in time. In addition, it is necessary to select high-spatial-resolution forcing data and emissivity schemes to complete the high-spatial-resolution simulation of the surface temperature. Third, the choice of assimilation algorithm. The most commonly used data assimilation algorithms are continuous assimilation methods and sequential assimilation methods (Kalman filter and its variants, particle filter and Bayes filter, etc.); continuous assimilation methods can be mainly classified into variational methods (three-dimensional variational and four-dimensional variational, etc.) and Monte Carlo global optimization methods (simulated annealing and genetic algorithm, etc.). Among them, the ensemble Kalman filter is a variant of the Kalman filter. It can solve the nonlinear problem of the model through the Monte Carlo method, generate a set of estimated parameter values and predicted state variables for continuous assimilation, and the uncertainty of the assimilation results will decrease as the assimilation proceeds. Fourth, the selection of process models. Since the Noah-MP model has the characteristics of relatively high computational efficiency, multi-parameterization scheme and high accuracy, the simulation of surface temperature fully considers factors such as radiation and terrain, and the calculated surface temperature under the cloud is a true reflection of the pixels under the cloud.Therefore, Noah-MP was finally selected to participate in the reconstruction of high spatiotemporal surface temperature.
[0025] See also Figure 1 to Figure 5 , an embodiment of the present invention provides a method for reconstructing land surface temperature, comprising the following steps:
[0026] Step 1: Obtain instantaneous surface temperature observation data observed by high spatial resolution MODIS satellite sensors and surface temperature observation data observed by micrometeorological stations. Update the input parameters of the Noah-MP model used to simulate the relationship between the surface and the atmosphere. The input parameters are the aerodynamic roughness length estimated by the logarithmic wind speed method based on the observation data of micrometeorological stations, and the soil emissivity estimated by the average method based on satellite remote sensing observation data or measured by Fourier transform infrared spectrometer FTIR, which solves the current problem of extreme lack of land surface parameters in the uninhabited area of Qilian Mountains-Altun Mountains-Kunlun Mountains. Specifically, based on the gradient wind speed data observed by micrometeorological stations, the logarithmic wind speed method is used to systematically and comprehensively estimate the aerodynamic roughness length on the underlying surface types mosaic of open shrubs, sparse grasslands, bare soil or sparse vegetation, grasslands, irrigated farmland and cultivated weeds.
[0027] Step 2: Input the ERA5-Land micro-meteorological station observation data into the Noah-MP model to obtain the surface temperature simulation values at multiple consecutive moments. The error evaluation method is used to evaluate the simulation value error of the surface temperature simulation value and the surface temperature observation data, and the error evaluation method is used to evaluate the observation value error of the instantaneous surface temperature observation data and the surface temperature observation data.
[0028] Step 3: When the MODIS satellite sensor is able to observe the surface conditions at a certain moment, specifically when there is remotely sensed leaf area index LAI or surface temperature LST data at a certain moment, the ensemble Kalman filter data assimilation method is used to assimilate the simulated value and the observed data according to the surface temperature simulation value and the simulation value error at the current moment, as well as the instantaneous surface temperature observation data and the observation value error at the current moment, and the surface temperature simulation value of Noah-MP at the current moment is updated according to the assimilated data, so that the updated surface temperature simulation value has the high spatial resolution characteristics of the instantaneous surface temperature observation data; when the MODIS satellite sensor cannot observe the surface conditions at a certain moment, specifically when there is no remotely sensed leaf area index LAI or surface temperature LST data at a certain moment, the surface temperature simulation value updated at the previous moment and the surface temperature simulation value at the current moment are weighted averaged, and the surface temperature simulation value at the current moment is updated, so that the high spatial resolution characteristics of the previous moment can continue at the current moment; the high spatial resolution reconstruction of the surface temperature at multiple consecutive moments is completed.
[0029] The specific technical solutions are as follows:
[0030] (1) Quality control of remote sensing surface temperature (MYD11A1 and MOD11A1 products): We have conducted the most stringent quality control on LST through the OCday and OC Night fields of MOD11A1. We selected surface temperature pixels that simultaneously met the data quality requirements of "Dataquality flag = good data quality", average emissivity error <= 0.01, and average LST error <= 1K to evaluate the accuracy of remote sensing products, and finally included them in the data assimilation of Noah-MP. Although this most stringent quality control ensures the accuracy of the fused product, it also leads to a reduction in fused sample data. In this patent, the fused samples of MYD11A1 and MOD11A1 during the day and night only reached 38% and 42% respectively, and an average of 1 assimilation can be achieved every 19.2 hours. Due to cloud contamination and observation quality, the ideal assimilation of 1 time every 12 hours cannot be achieved.
[0031] (2) Calibration of land surface process parameters based on ground observation data: First, the aerodynamic roughness length and soil emissivity parameters of the six underlying surface types in the land surface process model were calibrated based on the flux observation of the mountain underlying surface. For the soil emissivity parameter, we measured a value of 0.91 using the Fourier transform infrared spectrometer FTIR in the Tazhong area, an estimated value of 0.912-0.916 using the narrow band radiation of MODIS, and an estimated value of 0.912-0.928 using Landsat-8. Finally, 0.913 was selected as the soil emissivity in this study. For the aerodynamic roughness length Z 0m Based on the site observation data, we used the modified logarithmic wind speed method to calculate on 6 underlying surfaces. The specific calculation formula of the logarithmic wind speed method is as follows:
[0032] lnz0m=ln(zd)-ku / u*-ψm(z / L)(1)
[0033] Among them, u(ms -1 ) is the wind speed at height z (m), k (0.4) is the von Kalman constant, u* (ms -1 ) is the friction wind speed, z / L is the atmospheric stability parameter (negative value indicates instability, positive value indicates stability), L(m) is the Obukhov length, d(m) is the displacement height, Ψm(z / L) is the comprehensive stability correction term of the wind profile, if z / L>0 it is stable, z / L<0 it is unstable.
[0034]
[0035] Where x = (1-19z / L) 1 / 4According to formula (1), ln(zd) is a value that changes with the seasonal vegetation cover. Because u* is closely related to u, ln(Z 0m ) is u and Ψ m In addition, changes in wind speed will change the atmospheric turbulence, thereby changing the atmospheric stability z / L, and thus Ψ m (z / L) changes.
[0036] (3) Accuracy assessment of the Noah-MP land surface model and MODIS land surface temperature (MYD11A1 and MOD11A1) to obtain the model simulation error and remote sensing observation error. The main evaluation methods include root mean square error RMSE, mean absolute error MAE, bias, and determination coefficient R2. The main formulas are as follows:
[0037]
[0038]
[0039] (4) Data assimilation method: The Ensemble Kalman Filter (EnKF) data assimilation method is used to combine remote sensing observation errors and LSMs simulation errors to achieve dynamic fusion of remote sensing observations and LSMs simulation outputs for different terrains (such as irrigated farmland, alpine meadows, desert undulating terrain, desert shrubs and Gobi underlying surfaces). The following is the conceptual formula of the Ensemble Kalman Filter used in this patent. First, we must initialize the model and define is the initial value:
[0040]
[0041] In formula (7), represents the i-th analysis variable at the initial time, where the superscript a represents the analysis variable; u i is the background error vector of Gaussian distribution, with mean 0 and variance P; P represents the background error. In the second step, we need to predict the model state variables, as shown in the prediction equation (8):
[0042]
[0043] In formula (8), represents the i-th (i is the number of sets) prediction variable at the k+1 time node; the superscript f represents the prediction variable; represents the i-th (i is the number of sets) analysis variable at time node k; w i is the Gaussian distribution model error vector with mean 0 and variance Q; Q represents the model error. In the third step, we define the observation operator, which is expressed as follows:
[0044]
[0045] In formula (9), Y i,k+1 represents the ith simulated observation at time k+1 (in this study, it mainly refers to the surface temperature products from MYD11A1 and MOD11A1); H represents the observation operator corresponding to the model state variable; v i is the observation error vector from a Gaussian distribution on a certain land cover type, with a mean of 0 and a variance of R; R represents the observation error, which is determined in this study based on the comparison of MODIS and ground observations on different land cover types. When MODIS observation data does not exist, the model will simulate forward; when MODIS observation data is available at a certain time node, the model variables will be updated according to the assimilation model:
[0046]
[0047] In the above formula, K represents the state variable analysis value of the i-th set at the k+1 time node; k+1 is the coefficient matrix of the k+1 time observation increment; represents the MODIS surface temperature increment coefficient matrix at time k+1; is the background error covariance matrix of the k+1 time node; represents the average value of the forecast state variable at time k+1; [] T Represents the transpose of a matrix; is the simulation state variable of the i-th set member at time k+1; is the average value of the state variable in the k+1 time simulation.
[0048] (5) Generation of spatiotemporal continuous surface temperature data: Based on the calibration and verification of a large number of ground stations, spatiotemporal continuous surface temperature data with a spatial resolution of 1 km × 1 km and a temporal resolution of one hour are generated for the Kunlun Mountains, the Altun Mountains, and the Qilian Mountains. We take the surface temperature reconstruction results of the rectangular range of the Qilian Mountains as an example (see Appendix). Figure 4 and attached Figure 5 shown).
[0049] The values of aerodynamic roughness of six types of underlying surfaces in mountainous areas, soil emissivity values, remote sensing observation errors of six types of underlying surfaces, and corresponding Noah-MP simulation errors obtained in this patent, such as the observation errors, simulation errors, and aerodynamic roughness lengths of six types of underlying surfaces in the model constructed in Table 1, are all original and provide a basis for the innovation and accuracy of the entire assimilation model. The Noah-MP model and the ensemble Kalman filter data assimilation method are not the innovations of this patent. The innovation of this patent lies in the correction of the Noah-MP model and the combination of the two to construct a downscaling data fusion framework, dynamically integrating the spatial heterogeneity information provided by remote sensing data into the land surface process model, and solving the problem of downscaling of surface temperature in mountainous areas and the problem of spatiotemporal reconstruction.
[0050] Table 1 - Observation error, simulation error and aerodynamic roughness length of the six underlying surfaces in the model
[0051]
[0052] The present invention also has the following beneficial effects:
[0053] 1. Whether it is remote sensing observation of surface temperature or numerical simulation of surface temperature by land surface process model, it is a true reflection of the surface temperature on the underlying surface at a certain moment, rather than a virtual reconstruction through data formulas and statistical relationships. This is particularly important for the reconstruction of surface temperature on complex underlying surfaces in mountains.
[0054] 2. The surface temperature obtained by this patent significantly improves the temporal and spatial continuity and accuracy of surface temperature observation in mountainous areas, especially in areas with frequent clouds and rain.
[0055] 3. It effectively reduces the overestimation problem of surface temperature synthesis in traditional optical remote sensing observations and enhances the application scenarios of surface temperature data.
[0056] 4. The method is easy to implement, suitable for different terrains and vegetation types, and has good universality and scalability.
[0057] 5. For the first time, parameter revisions were made on six underlying surfaces in the Qilian Mountains, Altun Mountains, and Kunlun Mountains, which improved the applicability of Noah-MP and its assimilation algorithm and provided parameter basis for future related work.
[0058] Conclusion: Through advanced data assimilation technology and multi-source data integration, this paper provides an effective technical approach for the spatiotemporal continuous observation of land surface temperature in mountainous areas, which is of great significance for in-depth understanding and response to climate change.
[0059] An embodiment of the present invention provides a system for reconstructing land surface temperature, including:
[0060] The data acquisition module is used to observe the surface using high spatial resolution MODIS satellite sensors to obtain instantaneous surface temperature observation data; at the same time, it obtains surface temperature observation data observed by micro-meteorological stations. The error evaluation module is used to input the land surface variable dataset ERA5-Land at multiple consecutive moments into the Noah-MP model used to simulate the surface temperature according to the land surface conditions, and obtain the surface temperature simulation values at multiple consecutive moments; evaluate the simulation value error of the surface temperature simulation value and the surface temperature observation data, and evaluate the observation value error of the instantaneous surface temperature observation data and the surface temperature observation data. The surface temperature reconstruction module is used to assimilate the simulated value and the observed data according to the surface temperature simulation value and the simulation value error at the current moment, and the instantaneous surface temperature observation data and the observation value error at the current moment when the MODIS satellite sensor can observe the surface at a certain moment; and update the surface temperature simulation value at the current moment according to the assimilated data, so that the updated surface temperature simulation value has the high spatial resolution characteristics of the instantaneous surface temperature observation data; when the MODIS satellite sensor cannot observe the surface at a certain moment, the surface temperature simulation value updated at the previous moment and the surface temperature simulation value at the current moment are weighted averaged, and the surface temperature simulation value at the current moment is updated, so that the high spatial resolution characteristics can be continued.
[0061] An embodiment of the present invention provides a computer device, comprising: a memory and a processor; the memory stores a computer program, and the processor implements the steps of a method for reconstructing ground surface temperature when executing the computer program.
[0062] An embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of a method for reconstructing land surface temperature are implemented.
[0063] A specific embodiment is as follows:
[0064] Step 1: Collect and preprocess ground station observation data and remote sensing observation data.
[0065] Step 2: Locally calibrate the aerodynamic roughness length and soil emissivity parameters of LSMs based on ground station data.
[0066] Step 3: Calculate the simulation error of the corrected Noah-MP model and the observation error of MODIS surface temperature based on ground station data.
[0067] Step 4: Initialize and run Noah-MP. When there is LAI or LST data at a certain moment in the simulation, the ensemble Kalman filter data assimilation method EnKF and the simulation error and observation error are used for data assimilation.
[0068] Step 5: If there is no LAI or LST observation data at a certain moment of operation, the output surface temperature at this moment is determined by the weighted co-prosperity of the simulation results at the previous moment and the simulation results at this moment, and then the next moment is initialized.
[0069] Step 6: Finally generate and verify the spatiotemporal continuous surface temperature data for the Kunlun Mountains-Altun Mountains-Qilian Mountains region.
[0070] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.
Claims
1. A method for reconstructing land surface temperature, It is characterized in that The following steps are involved: Use high spatial resolution MODIS satellite sensors to observe the surface to obtain instantaneous surface temperature observation data; at the same time, obtain surface temperature observation data observed by micro-meteorological stations; The land surface variable dataset ERA5-Land at multiple consecutive moments is input into the Noah-MP model used to simulate the surface temperature according to the land surface conditions, and the simulated surface temperature values at multiple consecutive moments are obtained; Evaluate the simulated value error between the simulated land surface temperature and the observed land surface temperature data, and evaluate the observed value error between the instantaneous land surface temperature observation data and the observed land surface temperature observation data; When the MODIS satellite sensor is able to observe the surface at a certain moment, the simulated value and the observed data are assimilated according to the surface temperature simulation value and the simulation value error at the current moment, as well as the instantaneous surface temperature observation data and the observation value error at the current moment; and the surface temperature simulation value at the current moment is updated according to the assimilated data, so that the updated surface temperature simulation value has the high spatial resolution characteristics of the instantaneous surface temperature observation data; When the MODIS satellite sensor cannot observe the surface at a certain moment, the surface temperature simulation value updated at the previous moment is weighted averaged with the surface temperature simulation value at the current moment, and the surface temperature simulation value at the current moment is updated, so that the high spatial resolution feature can be continued.
2. A method for reconstructing land surface temperature according to claim 1, It is characterized in that The data assimilation of the simulation value and the observation data specifically includes: using the ensemble Kalman filter data assimilation method to perform the data assimilation of the simulation value and the observation data.
3. A method for reconstructing land surface temperature according to claim 1, It is characterized in that The specific steps of updating the surface temperature simulation value at the current moment according to the assimilated data include: Obtain instantaneous surface temperature observation data, obtain surface temperature simulation values according to the Noah-MP model, and use the error evaluation method to obtain simulation value errors and observation value errors; The predicted error covariance matrix is constructed using the simulated surface temperature values and the simulated value errors, and the observed error covariance matrix is constructed using the instantaneous surface temperature observation data and the observed value errors; A Kalman gain matrix is obtained according to ensemble Kalman filter data assimilation; and the surface temperature simulation value is updated according to the Kalman gain matrix.
4. A method for reconstructing land surface temperature according to claim 1, It is characterized in that The Noah-MP model also includes updating the aerodynamic roughness length and soil emissivity in the input parameters before the Noah-MP model is used, and the specific steps are as follows: Based on the observation data of micrometeorological stations, the logarithmic wind speed method was used to estimate the aerodynamic roughness length on the underlying surface types of open shrubs, sparse grasslands, bare soil or sparse vegetation, grasslands, irrigated farmland, and cultivated wild grass mosaics; Based on the remote sensing observation data of MODIS satellite sensors, the soil emissivity is estimated using the averaging method or measured using a Fourier transform infrared spectrometer.
5. A method for reconstructing land surface temperature according to claim 1, It is characterized in that The obtaining of instantaneous surface temperature observation data specifically includes: Daily surface temperature data and soil emissivity data with a high spatial resolution of 1 km are obtained through observations of two surface temperature and soil emissivity data products MYD11A1 and MOD11A1 from the high spatial resolution MODIS satellite sensor.
6. A system for reconstructing land surface temperature, It is characterized in that include: A data acquisition module is used to observe the surface using a high spatial resolution MODIS satellite sensor to obtain instantaneous surface temperature observation data; At the same time, the surface temperature observation data observed by the micro-meteorological station are obtained; The error evaluation module is used to input the land surface variable data set ERA5-Land at multiple consecutive moments into the Noah-MP model used to simulate the surface temperature according to the land surface conditions, and obtain the surface temperature simulation values at multiple consecutive moments; evaluate the simulation value error of the surface temperature simulation value and the surface temperature observation data, and evaluate the observation value error of the instantaneous surface temperature observation data and the surface temperature observation data; The surface temperature reconstruction module is used to assimilate the simulated value and the observed data according to the surface temperature simulation value and the simulation value error at the current moment, and the instantaneous surface temperature observation data and the observation value error at the current moment when the MODIS satellite sensor can observe the surface at a certain moment; and update the surface temperature simulation value at the current moment according to the assimilated data, so that the updated surface temperature simulation value has the high spatial resolution characteristics of the instantaneous surface temperature observation data; When the MODIS satellite sensor cannot observe the surface at a certain moment, the surface temperature simulation value updated at the previous moment is weighted averaged with the surface temperature simulation value at the current moment, and the surface temperature simulation value at the current moment is updated, so that the high spatial resolution feature can be continued.
7. A computer device, include: Memory and processor; The memory stores a computer program, wherein the processor implements a method for reconstructing ground surface temperature according to any one of claims 1 to 5 when executing the computer program.
8. A computer-readable storage medium storing a computer program. It is characterized in that When the computer program is executed by a processor, a method for reconstructing ground surface temperature according to any one of claims 1 to 5 is implemented.
Citation Information
Cited By
Lake surface temperature reconstruction and freeze thawing phenology extraction method based on filtering-automatic identification
CN122064952A
