Full-temporal-spatial-scale global surface radiation income and expenditure total-factor collaborative inversion method and system

Abstract

The invention belongs to the technical field of remote sensing, and relates to a full-temporal-spatial-scale global surface radiation income and expenditure total-factor collaborative inversion method and system. The method comprises the following steps: constructing a machine learning model, taking remote sensing data, meteorological reanalysis data, topographic data and sky vision field factors as inputs, taking global surface radiation income and expenditure total elements as outputs, and setting a cost function to obtain a full-temporal-spatial-scale surface radiation income and expenditure total element collaborative inversion model; and inputting the remote sensing data, the meteorological reanalysis data, the topographic data and the sky vision field factors into the full-temporal-spatial-scale earth surface radiation income and expenditure total element collaborative inversion model to obtain target full-temporal-spatial-scale earth surface radiation income and expenditure total elements. According to the method, all-spatio-temporal scale earth surface radiation income and expenditure total-factor collaborative inversion can be realized, so that radiation income and expenditure balance among different components is realized.

Description

Technical Field

[0001] The present invention belongs to the field of remote sensing technology, and specifically relates to a method and system for collaborative inversion of all elements of global surface radiation budget at all time and space scales. Background Art

[0002] The energy incident, reflected, absorbed, and emitted by the Earth system is part of the Earth's radiation budget. The surface radiation budget represents the balance between the incoming and outgoing radiation on the surface. The incoming solar radiation includes both direct and diffuse radiation. The total elements of surface radiation balance include direct solar radiation, diffuse solar radiation, total solar radiation incident on the surface, direct photosynthetically active radiation, diffuse photosynthetically active radiation, total photosynthetically active radiation, outgoing solar radiation, incoming long-wave radiation, outgoing long-wave radiation, net short-wave radiation, net long-wave radiation and net radiation. The sum of direct solar radiation and diffuse solar radiation is called total solar radiation incident on the surface, the sum of direct photosynthetically active radiation and diffuse photosynthetically active radiation is called total photosynthetically active radiation, the total photosynthetically active radiation is 0.45 times of total solar radiation incident on the surface, the difference between total solar radiation incident on the surface and outgoing solar radiation is called net short-wave radiation, the difference between outgoing long-wave radiation and incoming long-wave radiation is called net long-wave radiation, and the sum of net short-wave radiation and net long-wave radiation is called net radiation. The surface radiation balance is the key to the earth's energy balance, and it dominates the surface temperature distribution, surface latent heat and sensible heat distribution, global climate patterns, and global hydrological processes. The accurate inversion of all elements of the surface radiation balance is of great significance to the global radiation balance, energy balance, water cycle, and carbon balance. However, the current remote sensing inversion of each component of the global surface radiation balance is mostly limited to the independent modeling and estimation of a single component variable. This situation makes it difficult to achieve radiation balance between different components of different remote sensing products.

[0003] The existing remote sensing inversion methods for each component of the global radiation budget all independently construct inversion models for each component of the radiation budget. For example, GLASS (Global LAnd Surface Satellite) products, Hi-GLASS (High - Resolution Global Land - Atmosphere System Simulator) products and FLUXCOM (Flux Comparison) products all independently construct remote sensing inversion models of net radiation; for example, BESS (Biophysical and Environmental Science Satellite) products independently construct remote sensing inversion models of incident shortwave radiation; some studies independently construct remote sensing inversion models of incident shortwave radiation, incident longwave radiation or incident shortwave net radiation. The products produced by existing technologies have great uncertainty, resulting in inconsistencies among the existing components of the global surface radiation budget, making it difficult to achieve a balance in the radiation budget. This not only introduces great uncertainty into the global radiation budget and its relationship with climate change, turbulent heat flux, and other changes and attribution studies, but also interferes with research on extreme weather events and regional heat island effects, and increases the difficulty of formulating measures to address regional climate change. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a method and system for collaborative inversion of all elements of global surface radiation budget at all time and space scales.

[0005] In a first aspect, the present invention provides a method for collaborative inversion of all elements of global surface radiation budget at all time and space scales, including: Construct a machine learning model, take remote sensing data, meteorological reanalysis data, terrain data and sky view factors as input, take the global surface radiation budget as output, and set the cost function to obtain the coordinated inversion model of the surface radiation budget at all time and space scales; The remote sensing data, meteorological reanalysis data, terrain data and sky view factors are input into the collaborative inversion model of all elements of surface radiation budget at all temporal and spatial scales to obtain the target surface radiation budget at all temporal and spatial scales.

[0006] In a second aspect, the present invention provides a global surface radiation budget full-element collaborative inversion system at all time and space scales, including a model building unit and a data processing unit; A model building unit is used to build a machine learning model, which takes remote sensing data, meteorological reanalysis data, terrain data and sky view factors as input, takes the global surface radiation budget as output, and sets a cost function to obtain a coordinated inversion model of the surface radiation budget at all time and space scales; The output unit is used to input remote sensing data, meteorological reanalysis data, terrain data and sky view factors into the collaborative inversion model of all elements of surface radiation budget at all time and space scales to obtain the target all time and space scale surface radiation budget full elements.

[0007] Based on the above technical solution, the present invention can also be improved as follows.

[0008] Furthermore, the full elements of the global surface radiation balance include direct solar radiation, diffuse solar radiation, surface incident total solar radiation, direct photosynthetically active radiation, diffuse photosynthetically active radiation, total photosynthetically active radiation, surface outgoing solar radiation, surface incident longwave radiation, surface outgoing longwave radiation, surface net shortwave radiation, surface net longwave radiation and surface net radiation.

[0009] Furthermore, based on the site observation data, remote sensing data, meteorological reanalysis data, terrain data and sky view factors, a full-time and space-scale surface radiation budget full-factor collaborative inversion model-driven dataset is constructed; the constructed full-time and space-scale surface radiation budget full-factor collaborative inversion model is trained using the full-time and space-scale surface radiation budget full-factor collaborative inversion model-driven dataset to determine the optimal model parameters and obtain the optimal full-time and space-scale surface radiation budget full-factor collaborative inversion model.

[0010] Furthermore, the data required for the coordinated inversion of all elements of the global surface radiation budget are collected, including site observation data, remote sensing data, meteorological reanalysis data, terrain data and sky view factors; terrain data include digital elevation values, slope and slope aspect; Preprocess and screen the site observation data to obtain effective observation data of radiation sites; effective observation data of radiation sites include direct solar radiation observation data, scattered solar radiation observation data, direct photosynthetically active radiation observation data, scattered photosynthetically active radiation observation data, surface outgoing solar radiation observation data, surface incident long-wave radiation observation data and surface outgoing long-wave radiation observation data; remote sensing data include top of atmosphere solar radiation data, soil moisture, normalized vegetation index, leaf area index and vegetation coverage; meteorological reanalysis data include air temperature, surface temperature and relative humidity; Calculate the surface incident total solar radiation observation, total photosynthetically active radiation observation, surface net shortwave radiation observation, surface net longwave radiation observation and surface net radiation observation data based on the effective observation data of the radiation station, and obtain the surface radiation observation data by combining the effective observation data of the radiation station; Performing data preprocessing on the collected remote sensing data and meteorological reanalysis data to obtain remote sensing data after data preprocessing and meteorological reanalysis data after data preprocessing; Calculate digital elevation values, slope, aspect and sky view factor based on digital elevation model data of several resolutions.

[0011] Furthermore, the surface incident total solar radiation observation, total photosynthetically active radiation observation, surface net shortwave radiation observation, surface net longwave radiation observation and surface net radiation observation data are calculated based on the effective observation data of the radiation stations. The surface total solar radiation observation is the sum of the direct solar radiation observation and the diffuse solar radiation observation, the total photosynthetically active radiation is the sum of the direct photosynthetically active radiation and the diffuse photosynthetically active radiation, the surface net shortwave radiation is the difference between the surface incident total solar radiation and the surface outgoing solar radiation, the surface net longwave radiation is the difference between the surface incident longwave radiation and the surface outgoing longwave radiation, and the surface net radiation is the sum of the surface net shortwave radiation and the surface net longwave radiation.

[0012] Furthermore, the machine learning model is any one of a random forest model, a gradient boosting tree model, a deep neural network model, and a convolutional neural network model.

[0013] Furthermore, the driving dataset of the full-time and space-time scale surface radiation budget full-factor collaborative inversion model is divided into a model training dataset, a test dataset and a verification dataset; the model training dataset is used to train the constructed full-time and space-time scale surface radiation budget full-factor collaborative inversion model; the test dataset is used to evaluate the trained full-time and space-time scale surface radiation budget full-factor collaborative inversion model, and the verification dataset is used to adjust the hyperparameters of the constructed full-time and space-time scale surface radiation budget full-factor collaborative inversion model to determine the optimal full-time and space-time scale surface radiation budget full-factor collaborative inversion model.

[0014] Further, suppose represents the slope, Represents the slope direction, represents the sky view factor, Represents the digital elevation value, represents the solar radiation at the top of the atmosphere, represents the air temperature, represents the surface temperature, represents relative humidity, Represents soil moisture, represents the Normalized Difference Vegetation Index, represents the leaf area index, represents the vegetation coverage, Represents the surface direct solar radiation observation data, represents the scattered solar radiation observation data, represents the total incident solar radiation observation data on the surface, represents direct photosynthetically active radiation, represents diffuse photosynthetically active radiation, represents the total photosynthetically active radiation, Represents the surface outgoing solar radiation observation data, represents the observation data of incident longwave radiation on the surface, represents the observation data of the outgoing long-wave radiation from the surface, represents the surface net shortwave radiation observation data, represents the surface net longwave radiation observation data, represents the surface net radiation observation data, Represents the coordinated inversion model of all elements of surface radiation budget at all time and space scales, represents a set, then the coordinated inversion model of all elements of surface radiation budget at all time and space scales is expressed as:

[0015] .

[0016] Further, suppose Represents the surface direct solar radiation observation data, represents the scattered solar radiation observation data, represents the total incident solar radiation observation data on the surface, represents the direct photosynthetically active radiation observation data, represents the diffuse photosynthetically active radiation observation data, represents the total photosynthetically active radiation observation data, Represents the surface outgoing solar radiation observation data, represents the observation data of incident longwave radiation on the surface, represents the observation data of the outgoing long-wave radiation from the surface, represents the surface net shortwave radiation observation data, represents the surface net longwave radiation observation data, represents the surface net radiation observation data, Represents the estimated value of direct solar radiation on the surface, represents the estimated diffuse solar radiation, Represents the estimated total solar radiation incident on the surface, represents the estimated direct photosynthetically active radiation, represents the estimated diffuse photosynthetically active radiation, represents the estimated total photosynthetically active radiation, Represents the estimated value of outgoing solar radiation from the surface, represents the estimated incident longwave radiation on the surface, represents the estimated value of the outgoing longwave radiation from the surface, represents the estimated net shortwave radiation from the surface, represents the estimated net longwave radiation at the surface, represents the estimated net radiation at the surface, for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of represents a set, then the cost function is: .

[0017] The beneficial effects of the present invention are as follows: the present invention constructs several types of machine learning models and uses the constructed full-time and space-scale surface radiation budget full-factor collaborative inversion model to drive the data set for model training, thereby being able to select the optimal full-time and space-scale surface radiation budget full-factor collaborative inversion model, which is beneficial to improving the accuracy of the full-time and space-scale surface radiation budget full-factor collaborative inversion; in addition, the present invention realizes the full-time and space-scale surface radiation budget full-factor collaborative inversion by collecting observation data, remote sensing data, meteorological reanalysis data and digital elevation model data, so that radiation budget balance is achieved between different components. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of a method for collaborative inversion of all elements of global surface radiation budget at all time and space scales provided in Example 1 of the present invention; Figure 2 Schematic diagram of the full-element collaborative inversion system for global surface radiation budget at all time and space scales provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0020] Example 1 As an example, Figure 1 As shown, in order to solve the above technical problems, this embodiment provides a method for collaborative inversion of all elements of global surface radiation budget at all time and space scales, including: Construct a machine learning model, take remote sensing data, meteorological reanalysis data, terrain data and sky view factors as input, take the global surface radiation budget as output, and set the cost function to obtain the coordinated inversion model of the surface radiation budget at all time and space scales; The remote sensing data, meteorological reanalysis data, terrain data and sky view factors are input into the collaborative inversion model of all elements of surface radiation budget at all temporal and spatial scales to obtain the target surface radiation budget at all temporal and spatial scales.

[0021] In actual application, the observation data comes from the global eddy covariance flux observation network, the terrestrial ecosystem flux observation research network, the ecohydrological remote sensing experiment hydrometeorological observation network, the multi-scale surface flux and meteorological element observation data set, the surface process comprehensive observation network and the cold and arid area scientific observation network.

[0022] Optionally, the full elements of the global surface radiation balance include direct solar radiation, diffuse solar radiation, surface incident total solar radiation, direct photosynthetically active radiation, diffuse photosynthetically active radiation, total photosynthetically active radiation, surface outgoing solar radiation, surface incident longwave radiation, surface outgoing longwave radiation, surface net shortwave radiation, surface net longwave radiation and surface net radiation.

[0023] Optionally, a full-time and space-scale surface radiation budget full-factor collaborative inversion model-driven dataset is constructed based on site observation data, remote sensing data, meteorological reanalysis data, terrain data and sky view factors; the constructed full-time and space-scale surface radiation budget full-factor collaborative inversion model is trained using the full-time and space-scale surface radiation budget full-factor collaborative inversion model-driven dataset to determine the optimal model parameters and obtain the optimal full-time and space-scale surface radiation budget full-factor collaborative inversion model.

[0024] Optionally, collect the data required for the coordinated inversion of all elements of the global surface radiation budget, including site observation data, remote sensing data, meteorological reanalysis data, terrain data and sky view factors; terrain data includes digital elevation values, slope and aspect; Preprocess and screen the site observation data to obtain effective observation data of radiation sites; effective observation data of radiation sites include direct solar radiation observation data, scattered solar radiation observation data, direct photosynthetically active radiation observation data, scattered photosynthetically active radiation observation data, surface outgoing solar radiation observation data, surface incident long-wave radiation observation data and surface outgoing long-wave radiation observation data; remote sensing data include top of atmosphere solar radiation data, soil moisture, normalized vegetation index, leaf area index and vegetation coverage; meteorological reanalysis data include air temperature, surface temperature and relative humidity; Calculate the surface incident total solar radiation observation, total photosynthetically active radiation observation, surface net shortwave radiation observation, surface net longwave radiation observation and surface net radiation observation data based on the effective observation data of the radiation station, and obtain the surface radiation observation data by combining the effective observation data of the radiation station; Performing data preprocessing on the collected remote sensing data and meteorological reanalysis data to obtain remote sensing data after data preprocessing and meteorological reanalysis data after data preprocessing; Calculate digital elevation values, slope, aspect and sky view factor based on digital elevation model data of several resolutions.

[0025] Remote sensing data include sentinel series remote sensing data, Landsat series remote sensing data, MODIS series remote sensing products and GLASS series remote sensing products; the temporal and spatial scales of meteorological reanalysis data include hourly scale, daily scale and monthly scale.

[0026] Typical time scales include hourly scale, daily scale, weekly scale, ten-day scale, monthly scale and annual scale; typical spatial scales include 10m, 30m, 500m and 1km; typical angles include 0.05°, 0.1° and 0.25°.

[0027] Optionally, the surface incident total solar radiation observation, total photosynthetically active radiation observation, surface net shortwave radiation observation, surface net longwave radiation observation and surface net radiation observation data are calculated based on the effective observation data of the radiation station. The surface total solar radiation observation is the sum of direct solar radiation observation and diffuse solar radiation observation, the total photosynthetically active radiation is the sum of direct photosynthetically active radiation and diffuse photosynthetically active radiation, the surface net shortwave radiation is the difference between the surface incident total solar radiation and the surface outgoing solar radiation, the surface net longwave radiation is the difference between the surface incident longwave radiation and the surface outgoing longwave radiation, and the surface net radiation is the sum of the surface net shortwave radiation and the surface net longwave radiation.

[0028] Optionally, the machine learning model is any one of a random forest model, a gradient boosting tree model, a deep neural network model, and a convolutional neural network model.

[0029] By adopting the machine learning model, training any one of the machine learning models including the random forest model, gradient boosting tree model, deep neural network model and convolutional neural network model respectively is beneficial to obtain the optimal coordinated inversion model of all elements of surface radiation budget at all time and space scales, and improve the accuracy of coordinated inversion of all elements of surface radiation budget at all time and space scales.

[0030] Optionally, the driving dataset of the full-time and space-time scale surface radiation budget full-factor collaborative inversion model is divided into a model training dataset, a test dataset and a verification dataset; the model training dataset is used to train the constructed full-time and space-time scale surface radiation budget full-factor collaborative inversion model; the test dataset is used to evaluate the trained full-time and space-time scale surface radiation budget full-factor collaborative inversion model; the verification dataset is used to adjust the hyperparameters of the constructed full-time and space-time scale surface radiation budget full-factor collaborative inversion model to determine the optimal full-time and space-time scale surface radiation budget full-factor collaborative inversion model.

[0031] Optional, set represents the slope, Represents the slope direction, represents the sky view factor, Represents the digital elevation value, represents the solar radiation at the top of the atmosphere, represents the air temperature, represents the surface temperature, represents relative humidity, Represents soil moisture, represents the Normalized Difference Vegetation Index, represents the leaf area index, represents the vegetation coverage, Represents the surface direct solar radiation observation data, represents the scattered solar radiation observation data, represents the total incident solar radiation observation data on the surface, represents direct photosynthetically active radiation, represents diffuse photosynthetically active radiation, represents the total photosynthetically active radiation, Represents the surface outgoing solar radiation observation data, represents the observation data of incident longwave radiation on the surface, represents the observation data of the outgoing long-wave radiation from the surface, represents the surface net shortwave radiation observation data, represents the surface net longwave radiation observation data, represents the surface net radiation observation data, Represents the coordinated inversion model of all elements of surface radiation budget at all time and space scales, represents a set, then the coordinated inversion model of all elements of surface radiation budget at all time and space scales is expressed as:

[0032] .

[0033] Optional, set Represents the surface direct solar radiation observation data, represents the scattered solar radiation observation data, represents the total incident solar radiation observation data on the surface, represents the direct photosynthetically active radiation observation data, represents the diffuse photosynthetically active radiation observation data, represents the total photosynthetically active radiation observation data, Represents the surface outgoing solar radiation observation data, represents the observation data of incident longwave radiation on the surface, represents the observation data of the outgoing long-wave radiation from the surface, represents the surface net shortwave radiation observation data, represents the surface net longwave radiation observation data, represents the surface net radiation observation data, Represents the estimated value of direct solar radiation on the surface, represents the estimated diffuse solar radiation, Represents the estimated total solar radiation incident on the surface, represents the estimated direct photosynthetically active radiation, represents the estimated diffuse photosynthetically active radiation, represents the estimated total photosynthetically active radiation, Represents the estimated value of outgoing solar radiation from the surface, represents the estimated incident longwave radiation on the surface, represents the estimated value of the outgoing longwave radiation from the surface, represents the estimated net shortwave radiation from the surface, represents the estimated net longwave radiation at the surface, represents the estimated net radiation at the surface, for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of represents a set, then the cost function is: .

[0034] The present invention constructs several types of machine learning models and uses the constructed full-time and space-scale surface radiation budget full-factor collaborative inversion model to drive the data set for model training, which can select the optimal full-time and space-scale surface radiation budget full-factor collaborative inversion model, which is beneficial to improving the accuracy of full-time and space-scale surface radiation budget full-factor collaborative inversion; in addition, the present invention realizes full-time and space-scale surface radiation budget full-factor collaborative inversion by collecting observation data, remote sensing data, meteorological reanalysis data and digital elevation model data, so that radiation budget balance is achieved between different components.

[0035] Example 2 Based on the same principle as the method shown in Example 1 of the present invention, as shown in the attached Figure 2 As shown, the embodiment of the present invention also provides a global surface radiation budget full-element collaborative inversion system at all time and space scales, including a model building unit and an input data processing unit; A model building unit is used to build a machine learning model, which takes remote sensing data, meteorological reanalysis data, terrain data and sky view factors as input, takes the global surface radiation budget as output, and sets a cost function to obtain a coordinated inversion model of the surface radiation budget at all time and space scales; The output unit is used to input remote sensing data, meteorological reanalysis data, terrain data and sky view factors into the collaborative inversion model of all elements of surface radiation budget at all time and space scales to obtain the target all time and space scale surface radiation budget full elements.

[0036] Optionally, the full elements of the global surface radiation balance include direct solar radiation, diffuse solar radiation, surface incident total solar radiation, direct photosynthetically active radiation, diffuse photosynthetically active radiation, total photosynthetically active radiation, surface outgoing solar radiation, surface incident longwave radiation, surface outgoing longwave radiation, surface net shortwave radiation, surface net longwave radiation and surface net radiation.

[0037] Optionally, a full-time and space-scale surface radiation budget full-factor collaborative inversion model-driven dataset is constructed based on site observation data, remote sensing data, meteorological reanalysis data, terrain data and sky view factors; the constructed full-time and space-scale surface radiation budget full-factor collaborative inversion model is trained using the full-time and space-scale surface radiation budget full-factor collaborative inversion model-driven dataset to determine the optimal model parameters and obtain the optimal full-time and space-scale surface radiation budget full-factor collaborative inversion model.

[0038] Optionally, collect the data required for the coordinated inversion of all elements of the global surface radiation budget, including site observation data, remote sensing data, meteorological reanalysis data, terrain data and sky view factors; terrain data includes digital elevation values, slope and aspect; Preprocess and screen the site observation data to obtain effective observation data of radiation sites; effective observation data of radiation sites include direct solar radiation observation data, scattered solar radiation observation data, direct photosynthetically active radiation observation data, scattered photosynthetically active radiation observation data, surface outgoing solar radiation observation data, surface incident long-wave radiation observation data and surface outgoing long-wave radiation observation data; remote sensing data include top of atmosphere solar radiation data, soil moisture, normalized vegetation index, leaf area index and vegetation coverage; meteorological reanalysis data include air temperature, surface temperature and relative humidity; Calculate the surface incident total solar radiation observation, total photosynthetically active radiation observation, surface net shortwave radiation observation, surface net longwave radiation observation and surface net radiation observation data based on the effective observation data of the radiation station, and obtain the surface radiation observation data by combining the effective observation data of the radiation station; Performing data preprocessing on the collected remote sensing data and meteorological reanalysis data to obtain remote sensing data after data preprocessing and meteorological reanalysis data after data preprocessing; Calculate digital elevation values, slope, aspect and sky view factor based on digital elevation model data of several resolutions.

[0039] Optionally, the surface incident total solar radiation observation, total photosynthetically active radiation observation, surface net shortwave radiation observation, surface net longwave radiation observation and surface net radiation observation data are calculated based on the effective observation data of the radiation station. The surface total solar radiation observation is the sum of direct solar radiation observation and diffuse solar radiation observation, the total photosynthetically active radiation is the sum of direct photosynthetically active radiation and diffuse photosynthetically active radiation, the surface net shortwave radiation is the difference between the surface incident total solar radiation and the surface outgoing solar radiation, the surface net longwave radiation is the difference between the surface incident longwave radiation and the surface outgoing longwave radiation, and the surface net radiation is the sum of the surface net shortwave radiation and the surface net longwave radiation.

[0040] Optionally, the machine learning model is any one of a random forest model, a gradient boosting tree model, a deep neural network model, and a convolutional neural network model.

[0041] Optionally, the driving dataset of the full-time and space-time scale surface radiation budget full-factor collaborative inversion model is divided into a model training dataset, a test dataset and a verification dataset; the model training dataset is used to train the constructed full-time and space-time scale surface radiation budget full-factor collaborative inversion model; the test dataset is used to evaluate the trained full-time and space-time scale surface radiation budget full-factor collaborative inversion model; the verification dataset is used to adjust the hyperparameters of the constructed full-time and space-time scale surface radiation budget full-factor collaborative inversion model to determine the optimal full-time and space-time scale surface radiation budget full-factor collaborative inversion model.

[0042] Optional, set represents the slope, Represents the slope direction, represents the sky view factor, Represents the digital elevation value, represents the solar radiation at the top of the atmosphere, represents the air temperature, represents the surface temperature, represents relative humidity, Represents soil moisture, represents the Normalized Difference Vegetation Index, represents the leaf area index, represents the vegetation coverage, Represents the surface direct solar radiation observation data, represents the scattered solar radiation observation data, represents the total incident solar radiation observation data on the surface, represents direct photosynthetically active radiation, represents diffuse photosynthetically active radiation, represents the total photosynthetically active radiation, Represents the surface outgoing solar radiation observation data, represents the observation data of incident longwave radiation on the surface, represents the observation data of the outgoing long-wave radiation from the surface, represents the surface net shortwave radiation observation data, represents the surface net longwave radiation observation data, represents the surface net radiation observation data, Represents the coordinated inversion model of all elements of surface radiation budget at all time and space scales, represents a set, then the coordinated inversion model of all elements of surface radiation budget at all time and space scales is expressed as:

[0043] .

[0044] Optional, set Represents the surface direct solar radiation observation data, represents the scattered solar radiation observation data, represents the total incident solar radiation observation data on the surface, represents the direct photosynthetically active radiation observation data, represents the diffuse photosynthetically active radiation observation data, represents the total photosynthetically active radiation observation data, Represents the surface outgoing solar radiation observation data, represents the observation data of incident longwave radiation on the surface, represents the observation data of the outgoing long-wave radiation from the surface, represents the surface net shortwave radiation observation data, represents the surface net longwave radiation observation data, represents the surface net radiation observation data, Represents the estimated value of direct solar radiation on the surface, represents the estimated diffuse solar radiation, Represents the estimated total solar radiation incident on the surface, represents the estimated direct photosynthetically active radiation, represents the estimated diffuse photosynthetically active radiation, represents the estimated total photosynthetically active radiation, Represents the estimated value of outgoing solar radiation from the surface, represents the estimated incident longwave radiation on the surface, represents the estimated value of the outgoing longwave radiation from the surface, represents the estimated net shortwave radiation from the surface, represents the estimated net longwave radiation at the surface, represents the estimated net radiation at the surface, for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of represents a set, then the cost function is: .

[0045] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for the coordinated inversion of all elements of global surface radiation budget at all time and space scales, characterized by: include: Construct a machine learning model, take remote sensing data, meteorological reanalysis data, terrain data and sky view factors as input, take the global surface radiation budget as output, and set the cost function to obtain the coordinated inversion model of the surface radiation budget at all time and space scales; The remote sensing data, meteorological reanalysis data, terrain data and sky view factors are input into the collaborative inversion model of all elements of surface radiation budget at all temporal and spatial scales to obtain the target surface radiation budget at all temporal and spatial scales.

2. The method for collaborative inversion of global surface radiation budget at all time and space scales according to claim 1 is characterized in that: The full elements of the global surface radiation balance include direct solar radiation, diffuse solar radiation, total solar radiation incident on the surface, direct photosynthetically active radiation, diffuse photosynthetically active radiation, total photosynthetically active radiation, outgoing solar radiation, incoming longwave radiation, outgoing longwave radiation, net shortwave radiation, net longwave radiation and net radiation.

3. The method for collaborative inversion of global surface radiation budget at all time and space scales according to claim 1 is characterized in that: Based on site observation data, remote sensing data, meteorological reanalysis data, terrain data and sky view factors, a full-scale spatial and temporal surface radiation budget full-element collaborative inversion model driving data set is constructed; The constructed surface radiation budget full-factor collaborative inversion model of all time and space scales is trained using the full-time and space scale surface radiation budget full-factor collaborative inversion model driving data set to determine the optimal model parameters and obtain the optimal full-time and space scale surface radiation budget full-factor collaborative inversion model.

4. The method for collaborative inversion of global surface radiation budget at all time and space scales according to claim 2 is characterized in that: Collect the data required for the coordinated inversion of all elements of the global surface radiation budget, including site observation data, remote sensing data, meteorological reanalysis data, terrain data and sky view factors; The site observation data include direct solar radiation observation data, scattered solar radiation observation data, direct photosynthetically active radiation observation data, scattered photosynthetically active radiation observation data, surface outgoing solar radiation observation data, surface incident longwave radiation observation data and surface outgoing longwave radiation observation data; remote sensing data include top of atmosphere solar radiation data, soil moisture, normalized difference vegetation index, leaf area index and vegetation coverage; meteorological reanalysis data include air temperature, surface temperature and relative humidity; terrain data include digital elevation value, slope and slope aspect; Preprocess and screen the site observation data to obtain effective observation data of the radiation site; Performing data preprocessing on the collected remote sensing data and meteorological reanalysis data to obtain remote sensing data after data preprocessing and meteorological reanalysis data after data preprocessing; Calculate digital elevation values, slope, aspect and sky view factor based on digital elevation model data of several resolutions.

5. The method for collaborative inversion of global surface radiation budget at all time and space scales according to claim 4 is characterized in that: According to the effective observation data of radiation stations, the surface total incident solar radiation observation, total photosynthetically active radiation observation, surface net shortwave radiation observation, surface net longwave radiation observation and surface net radiation observation data are calculated. The surface total solar radiation observation is the sum of direct solar radiation observation and diffuse solar radiation observation, the total photosynthetically active radiation is the sum of direct photosynthetically active radiation and diffuse photosynthetically active radiation, the surface net shortwave radiation is the difference between the surface incident total solar radiation and the surface outgoing solar radiation, the surface net longwave radiation is the difference between the surface incident longwave radiation and the surface outgoing longwave radiation, and the surface net radiation is the sum of the surface net shortwave radiation and the surface net longwave radiation.

6. The method for collaborative inversion of global surface radiation budget at all time and space scales according to claim 1 is characterized in that: The machine learning model is any one of a random forest model, a gradient boosting tree model, a deep neural network model, and a convolutional neural network model.

7. The method for collaborative inversion of global surface radiation budget at all time and space scales according to claim 3 is characterized in that: The dataset driven by the collaborative inversion model of all elements of the surface radiation budget at all time and space scales is divided into model training dataset, test dataset and validation dataset; The model training data set is used to train the constructed surface radiation budget full-factor collaborative inversion model at all time and space scales; The test dataset is used to evaluate the trained collaborative inversion model of all-factors of surface radiation budget at all time and space scales, and the validation dataset is used to adjust the hyperparameters of the constructed collaborative inversion model of all-factors of surface radiation budget at all time and space scales to determine the optimal collaborative inversion model of all-factors of surface radiation budget at all time and space scales.

8. The method for collaborative inversion of global surface radiation budget at all time and space scales according to claim 1 is characterized in that: set up represents the slope, Represents the slope direction, represents the sky view factor, Represents the digital elevation value, represents the solar radiation at the top of the atmosphere, represents the air temperature, represents the surface temperature, represents relative humidity, Represents soil moisture, represents the Normalized Difference Vegetation Index, represents the leaf area index, represents the vegetation coverage, Represents the surface direct solar radiation observation data, represents the scattered solar radiation observation data, represents the total incident solar radiation observation data on the surface, represents direct photosynthetically active radiation, represents diffuse photosynthetically active radiation, represents the total photosynthetically active radiation, Represents the surface outgoing solar radiation observation data, represents the surface incident longwave radiation observation data, represents the observation data of the outgoing long-wave radiation from the surface, represents the surface net shortwave radiation observation data, represents the surface net longwave radiation observation data, represents the surface net radiation observation data, Represents the coordinated inversion model of all elements of surface radiation budget at all time and space scales, represents a set, then the coordinated inversion model of all elements of surface radiation budget at all time and space scales is expressed as: 。 9. The method for collaborative inversion of global surface radiation budget at all time and space scales according to claim 1 is characterized in that: set up Represents the surface direct solar radiation observation data, represents the scattered solar radiation observation data, represents the total incident solar radiation observation data on the surface, represents the direct photosynthetically active radiation observation data, represents the diffuse photosynthetically active radiation observation data, represents the total photosynthetically active radiation observation data, Represents the surface outgoing solar radiation observation data, represents the observation data of incident longwave radiation on the surface, represents the observation data of the outgoing long-wave radiation from the surface, represents the surface net shortwave radiation observation data, represents the surface net longwave radiation observation data, represents the surface net radiation observation data, Represents the estimated value of direct solar radiation on the surface, represents the estimated diffuse solar radiation, Represents the estimated total solar radiation incident on the surface, represents the estimated direct photosynthetically active radiation, represents the estimated diffuse photosynthetically active radiation, represents the estimated total photosynthetically active radiation, Represents the estimated value of outgoing solar radiation from the surface, represents the estimated incident longwave radiation on the surface, represents the estimated value of the outgoing longwave radiation from the surface, represents the estimated net shortwave radiation from the surface, represents the estimated net longwave radiation at the surface, represents the estimated net radiation at the surface, for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of for The weight coefficient of represents a set, then the cost function is: 。 10. The global surface radiation budget full-element collaborative inversion system at all time and space scales is characterized by: It includes a model building unit and a data processing unit; A model building unit is used to build a machine learning model, which takes remote sensing data, meteorological reanalysis data, terrain data and sky view factors as input, takes the global surface radiation budget as output, and sets a cost function to obtain a coordinated inversion model of the surface radiation budget at all time and space scales; The output unit is used to input remote sensing data, meteorological reanalysis data, terrain data and sky view factors into the collaborative inversion model of all elements of surface radiation budget at all time and space scales to obtain the target all time and space scale surface radiation budget full elements.

Images