A method and system for downscaling climate dynamics in the Loess Plateau region
By constructing a multi-source dataset, a multi-layered nested dynamic downscaling framework, and dual error correction, the problems of insufficient static processing and error correction of underlying surface data in climate simulation of the Loess Plateau region are solved, achieving high-precision and high-reliability climate simulation and supporting climate change monitoring and ecological environment assessment.
Patent Information
- Application Number
- CN202611412503.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-11
- Publication Date
- 2026-11-10
AI Technical Summary
Existing technologies for downscaling climate dynamics in the Loess Plateau region suffer from problems such as insufficient simulation accuracy due to static processing of underlying surface data, insufficient resolution of multi-layer nested grids, circulation deviation after long-term integration, lack of physical consistency in error correction, and lack of adaptive optimization mechanisms, which affect the accuracy and reliability of simulation results.
A multi-source dataset was constructed for preprocessing and spatiotemporal matching. A multi-layered nested dynamic downscaling framework was established, and a spectral approximation coefficient was introduced to constrain large-scale circulation. The parameter configuration was dynamically adjusted using dual error correction and a comprehensive simulation performance index to form a high-resolution climate element map.
It improves the accuracy and reliability of climate simulation in the Loess Plateau region, takes into account both large-scale circulation and local details, ensures the statistical accuracy and physical rationality of the data, and achieves adaptive optimization of the system.
Smart Images

Figure CN123068901A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of dynamic downscaling technology, and more specifically, to a method and system for downscaling climate dynamics in the Loess Plateau region. Background Technology
[0002] The Loess Plateau, located in central China, is a transitional zone between semi-arid and semi-humid climates, with a fragile ecosystem and high sensitivity to climate change. Against the backdrop of global warming, accurately simulating and predicting the spatiotemporal characteristics of climate change in this region is of significant scientific importance and practical value for local water resource management, agricultural planning, ecological protection, and disaster prevention and mitigation.
[0003] Regional climate models (RCMs) are core tools for dynamic downscaling. Using coarse-resolution global climate models or reanalysis data as lateral boundary conditions, they drive high-resolution regional models to obtain high-resolution predictions of regional climate. This can compensate for insufficient observational data in sparsely populated areas and provide data support for local climate change research. Current studies have conducted relevant dynamic downscaling simulations on the Loess Plateau based on models such as PRECIS, WRF, and RegCM.
[0004] However, existing technologies still have several shortcomings: First, the underlying surface data is processed statically, making it difficult to match the dramatic land use / cover changes and spatiotemporal dynamics of underlying surface parameters in the Loess Plateau, affecting the accuracy of land surface process simulations and limiting the accuracy of downscaling results. Second, most models use single-layer or double-layer nested grids with horizontal resolutions mostly between 30 and 50 km, making it difficult to balance large-scale circulation with local fine-grained simulations, and failing to effectively capture local climate variations, especially small-scale climate variations, under complex terrain. Third, after long-term integration, regional climate models are affected by their own systematic biases, and the simulated large-scale circulation gradually deviates from the driving field, reducing the reliability of the results. The relaxation coefficients of existing spectral approximation constraint schemes are fixed values and cannot be dynamically adjusted according to regional and seasonal bias characteristics, making it difficult to achieve optimal constraint effects. Fourth, error correction relies heavily on single statistical methods, and the correction results lack physical consistency constraints, which may violate basic physical laws such as energy and water balance, affecting the rationality and reliability of the data. Fifth, there is a lack of a multi-dimensional comprehensive quantitative evaluation system and an adaptive mechanism for dynamically adjusting parameters based on simulation results, making it impossible to achieve continuous system optimization and automatic selection of optimal configurations. Therefore, there is an urgent need for a method to downscale the climate dynamics of the Loess Plateau region. Summary of the Invention
[0005] In view of this, the present invention proposes a method and system for downscaling the climate dynamics of the Loess Plateau region, aiming to solve one or more of the above-mentioned problems.
[0006] On the one hand, this invention proposes a method for downscaling the climate dynamics of the Loess Plateau region, characterized by comprising: Multi-source data of the Loess Plateau region were acquired, and the multi-source data were preprocessed and spatiotemporally matched to construct a static geographic dataset and a dynamic underlying surface dataset of the Loess Plateau. The multi-source data included: coarse-resolution output data of global climate models, regional ground observation data, satellite remote sensing underlying surface data, and ecological characteristic data of the Loess Plateau. Based on regional climate models, a multi-layered nested dynamic downscaling framework is constructed using static geographic datasets and dynamic underlying surface datasets. Spectral approximation coefficients are introduced to constrain large-scale circulation, resulting in dynamic downscaling outputs. The dynamic downscaling output results are subjected to dual error correction based on multi-source observation data to obtain a high-resolution regional climate element dataset; the multi-source observation data includes regional ground observation data, satellite remote sensing underlying surface data, and Loess Plateau ecological characteristic data. A climate simulation evaluation index system for the Loess Plateau region is constructed. Based on the high-resolution regional climate element dataset and regional ground observation data, a comprehensive simulation performance index is determined. The parameter configuration and update strategy of the downscaling system are dynamically adjusted according to the comprehensive simulation performance index, and a multi-level climate element map is output.
[0007] Preferably, the regional ground observation data includes hourly / daily temperature, precipitation, wind speed, humidity, and air pressure observation data from meteorological stations throughout the Loess Plateau. The satellite remote sensing underlying surface data includes land use / cover change, vegetation index, and surface albedo of the Loess Plateau. The ecological characteristics data of the Loess Plateau include soil moisture, vegetation coverage, and degree of soil erosion.
[0008] Preferably, the multi-source data is preprocessed and spatiotemporally matched to construct a static geographic dataset and a dynamic underlying surface dataset of the Loess Plateau, including: The preprocessing includes data format standardization, spatiotemporal resolution unification, missing value imputation, outlier detection and removal, and spatiotemporal alignment of multi-source data. Based on preprocessed satellite remote sensing underlying surface data, annual land use / cover change data of the Loess Plateau are extracted using classification and regression methods to construct a time-series dynamic underlying surface dataset. Based on a high-resolution digital elevation model, parameters of slope, aspect, and topographic relief of the Loess Plateau are extracted to construct a static geographic dataset.
[0009] Preferably, after preprocessing and spatiotemporal matching of the multi-source data, the method further includes: The quality of the preprocessed multi-source data is assessed to determine the preprocessing quality score for each type of multi-source data. The quality assessment includes data integrity assessment, spatiotemporal continuity assessment, and outlier removal rate assessment. Pre-set a data quality qualification threshold; The preprocessing quality score is compared with the data quality pass threshold; If the preprocessing quality score is greater than or equal to the data quality qualification threshold, then the corresponding type of multi-source data is determined to be qualified. If the preprocessing quality score is less than the data quality pass threshold, the corresponding type of multi-source data is determined to be unqualified, and a re-preprocessing process is triggered until the preprocessing quality scores of all types of multi-source data are greater than or equal to the data quality pass threshold.
[0010] Preferably, based on regional climate models, a multi-layered nested dynamic downscaling framework is constructed using static geographic datasets and dynamic underlying surface datasets, including: The parent region covers the entire country with a horizontal resolution of 30km; the sub-region covers the Loess Plateau region with a horizontal resolution of 10km; and the grandchild region covers typical ecologically fragile areas of the Loess Plateau with a horizontal resolution of 3km. Based on the complex topographic features of the Loess Plateau, the subgrid topographic parameterization scheme in the regional climate model is activated to correct the deviation between the model grid topographic height and the actual topographic height. Based on static geographic datasets and dynamic underlying surface datasets, we optimized the combination of cumulus convection parameterization schemes, microphysical schemes, boundary layer schemes, and land surface process schemes in regional climate models. Using regional surface observation data as a benchmark, we selected and determined the optimal combination of parameterization schemes for each season / region.
[0011] Preferably, large-scale circulation is constrained by spectral approximation coefficients, including: During the integration process of the regional climate model, a Newtonian relaxation term is applied to the spectral components above a certain wavenumber so that the large-scale circulation of the regional climate model approximates the concurrent large-scale circulation of the driving field. The spectral approximation coefficients are dynamically adjusted based on the simulation deviation characteristics of different regions and seasons of the Loess Plateau; The spectral approximation coefficients are determined according to the following formula: α(t,x)=αbase×[1+β×B(t,x)]; Where α(t,x) represents the spectral approximation coefficient at position x at time t, αbase represents the basic spectral approximation coefficient, B(t,x) represents the deviation index based on historical simulation deviation statistics, and β represents the weighting coefficient of the deviation index.
[0012] Preferably, the dual error correction includes a first statistical error correction and a second physical constraint error correction; The first statistical error correction specifically involves: based on the statistical characteristics of the deviation between historical multi-source observation data and dynamic downscaling output results, using the quantile mapping method to correct the probability distribution deviation of climate elements; The secondary physical constraint error correction specifically involves: verifying the physical rationality of the results after the first correction based on the physical consistency constraints of the land-atmosphere coupling process in the Loess Plateau region; and making physical coordination adjustments when the corrected results do not conform to the physical constraints. Among them, physical constraints include the surface energy balance equation and the regional water balance equation.
[0013] Preferably, a climate simulation evaluation index system for the Loess Plateau region is constructed, and a comprehensive simulation performance index is determined based on the high-resolution regional climate element dataset and regional ground observation data, including: The evaluation index system includes climate mean state simulation evaluation index, interannual variability simulation evaluation index, extreme event simulation evaluation index, and spatial distribution simulation evaluation index. The climate mean state simulation evaluation indicators include multi-year average temperature deviation, multi-year average precipitation deviation, and seasonal cycle correlation coefficient. The interannual variability simulation and evaluation indicators include the interannual variability ratio of temperature and the interannual variability ratio of precipitation; The extreme event simulation and evaluation indicators include the extreme high temperature simulation index and the extreme precipitation simulation index. The spatial distribution simulation evaluation indicators include the spatial correlation coefficient of temperature and the spatial correlation coefficient of precipitation; The subjective weights of each evaluation indicator are determined by combining the analytic hierarchy process with expert experience. The objective weights are then determined by statistical analysis of the historical simulation deviations of each evaluation indicator. The comprehensive weight of each evaluation indicator is obtained by weighted summation. The comprehensive simulation performance index is determined based on the comprehensive weights and the values of each evaluation index.
[0014] Preferably, the formula for calculating the comprehensive simulation performance index is: ; Where PI represents the comprehensive simulation performance index, wi represents the comprehensive weight of the i-th evaluation index, and si represents the value of the i-th evaluation index.
[0015] Preferably, the strategy for dynamically adjusting the parameter configuration and update of the downscaling system based on the comprehensive simulation performance index includes: A first performance threshold and a second performance threshold are preset, wherein the first performance threshold is less than the second performance threshold; If the overall simulation performance index is less than or equal to the first performance threshold, the parameterized scheme re-optimization process is triggered. If the overall simulation performance index is greater than the first performance threshold and less than or equal to the second performance threshold, then adjust the current configuration parameter update frequency. If the overall simulation performance index is greater than the second performance threshold, the current configuration is output as the recommended configuration, and a multi-level climate element map is output based on the recommended configuration.
[0016] Compared with existing technologies, the advantages of this invention are as follows: This invention constructs a dynamic underlying surface dataset to drive a regional climate model, utilizing multi-source, annually updated land use / cover change datasets to reflect the spatiotemporal dynamic characteristics of the underlying surface, eliminating static underlying surface simulation biases, and improving the accuracy and reliability of climate simulation in the Loess Plateau region. Combining a three-layer nested dynamic downscaling framework (30km-10km-3km) with a subgrid topographic parameterization scheme, it ensures reasonable simulation of large-scale circulation while achieving high-resolution downscaling across the entire Loess Plateau and vulnerable areas, correcting topographic height biases, and enhancing the local climate simulation capabilities in complex topographic regions. Introducing a spectral approximation coefficient dynamically adjusted based on regional and seasonal simulation biases, it adaptively optimizes the constraint strength of large-scale circulation, preserving local detail simulation capabilities while avoiding circulation drift, and balancing the restoration of large-scale circulation with the simulation of small-scale characteristics. Employing a dual error correction scheme of quantile mapping statistical correction plus energy and water balance physical constraint correction, it eliminates systematic biases in dynamic downscaling while avoiding physical inconsistencies, ensuring that the final high-resolution climate dataset possesses both statistical accuracy and physical rationality. A comprehensive evaluation system is constructed from four dimensions: climate mean state, interannual variability, extreme events, and spatial distribution. A comprehensive simulation performance index is obtained by assigning subjective and objective weights. The downscaling system configuration is automatically adjusted based on the index results, realizing the transition from static configuration to dynamic adaptive optimization, thereby improving the system's applicability and simulation stability.
[0017] On the other hand, this application also provides a climate-dynamic downscaling system for the Loess Plateau region, used to apply the aforementioned climate-dynamic downscaling method for the Loess Plateau region, including: The data preprocessing module is configured to acquire multi-source data of the Loess Plateau region, and to preprocess and perform spatiotemporal matching on the multi-source data to construct a static geographic dataset and a dynamic underlying surface dataset of the Loess Plateau. The multi-source data includes: coarse-resolution output data of global climate models, regional ground observation data, satellite remote sensing underlying surface data, and ecological characteristic data of the Loess Plateau. The dynamic downscaling module is configured to construct a multi-layered nested dynamic downscaling framework based on regional climate models, static geographic datasets and dynamic underlying surface datasets, and introduce spectral approximation coefficients to constrain large-scale circulation, thereby obtaining dynamic downscaling output results. The error correction module is configured to perform dual error correction on the dynamic downscaling output based on multi-source observation data to obtain a high-resolution regional climate element dataset; the multi-source observation data includes regional ground observation data, satellite remote sensing underlying surface data, and Loess Plateau ecological characteristic data. The climate element output module is configured to construct a climate simulation evaluation index system for the Loess Plateau region. Based on the high-resolution regional climate element dataset and regional ground observation data, it determines the comprehensive simulation performance index and dynamically adjusts the parameter configuration and update strategy of the downscaling system according to the comprehensive simulation performance index, and outputs a multi-level climate element map.
[0018] It is understandable that the climate dynamics downscaling method and system provided in this application for the Loess Plateau region have the same beneficial effects, and will not be elaborated here. Attached Figure Description
[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly described below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 A flowchart of a method for downscaling the climate dynamics of the Loess Plateau region provided in an embodiment of the present invention; Figure 2 A functional block diagram of a climate dynamics downscaling system for the Loess Plateau region provided in an embodiment of the present invention. Detailed Implementation
[0020] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.
[0021] In some embodiments of this application, see Figure 1 This embodiment provides a method for downscaling the climate dynamics of the Loess Plateau region, including: Multi-source data of the Loess Plateau region were acquired, and the multi-source data were preprocessed and spatiotemporally matched to construct a static geographic dataset and a dynamic underlying surface dataset of the Loess Plateau. The multi-source data included: coarse-resolution output data of global climate models, regional ground observation data, satellite remote sensing underlying surface data, and ecological characteristic data of the Loess Plateau. Based on regional climate models, a multi-layered nested dynamic downscaling framework is constructed using static geographic datasets and dynamic underlying surface datasets. Spectral approximation coefficients are introduced to constrain large-scale circulation, resulting in dynamic downscaling outputs. The dynamic downscaling output results are subjected to dual error correction based on multi-source observation data to obtain a high-resolution regional climate element dataset; the multi-source observation data includes regional ground observation data, satellite remote sensing underlying surface data, and Loess Plateau ecological characteristic data. A climate simulation evaluation index system for the Loess Plateau region is constructed. Based on the high-resolution regional climate element dataset and regional ground observation data, a comprehensive simulation performance index is determined. The parameter configuration and update strategy of the downscaling system are dynamically adjusted according to the comprehensive simulation performance index, and a multi-level climate element map is output.
[0022] This application effectively addresses the problems of lagging underlying surface information updates and insufficient spatial heterogeneity representation in traditional downscaling methods by constructing a multi-source dataset containing both static geographic information and dynamic underlying surface features, significantly improving the model's ability to characterize the complex topography and surface processes of the Loess Plateau. Secondly, it introduces spectral approximation coefficients to constrain large-scale circulation, ensuring consistency between the physical processes within the regional climate model and the coordination between the downscaling results and the large-scale background field, reducing simulation biases caused by boundary effects. Thirdly, it employs a dual error correction strategy based on multi-source observation data, fully utilizing the spatial continuity advantages of high-precision ground observation data and satellite remote sensing and ecological characteristic data to achieve systematic correction of the dynamic downscaling output results, significantly improving the simulation accuracy and spatial distribution rationality of key climate elements such as temperature and precipitation. Finally, it establishes a dynamic parameter adjustment mechanism based on a comprehensive simulation performance index, achieving adaptive optimization of the downscaling system. This mechanism can automatically update the configuration according to the simulation performance under different climate backgrounds, thereby generating more reliable and detailed multi-level climate element maps, providing strong data support and technical assurance for climate change monitoring, ecological environment assessment, and water resource management in the Loess Plateau region.
[0023] In some embodiments of this application, the regional ground observation data includes hourly / daily temperature, precipitation, wind speed, humidity, and air pressure observation data from meteorological stations across the Loess Plateau; the satellite remote sensing underlying surface data includes land use / cover change, vegetation index, and surface albedo of the Loess Plateau; and the ecological characteristic data of the Loess Plateau includes soil moisture, vegetation cover, and soil erosion.
[0024] Understandably, systematically acquiring representative multi-type observation and remote sensing data covering the entire Loess Plateau provides a comprehensive and reliable data foundation for driving regional climate models, dynamically updating underlying surface parameters, correcting simulation errors, and ultimately evaluating simulation performance. Specifically, ground observation data serves as a true benchmark for model validation and a target reference for error correction; satellite remote sensing data captures the spatiotemporal variations of land cover and albedo to support the construction of dynamic underlying surface datasets; and ecological characteristic data characterizes the unique land surface process parameters of the Loess Plateau, enhancing the model's ability to depict regional eco-hydrological processes.
[0025] In some embodiments of this application, the multi-source data is preprocessed and spatiotemporally matched to construct a static geographic dataset and a dynamic underlying surface dataset of the Loess Plateau, including: The preprocessing includes data format standardization, spatiotemporal resolution unification, missing value imputation, outlier detection and removal, and spatiotemporal alignment of multi-source data. Based on preprocessed satellite remote sensing underlying surface data, annual land use / cover change data of the Loess Plateau are extracted using classification and regression methods to construct a time-series dynamic underlying surface dataset. Based on a high-resolution digital elevation model, parameters of slope, aspect, and topographic relief of the Loess Plateau are extracted to construct a static geographic dataset.
[0026] Understandably, by standardizing multi-source raw data with diverse sources, formats, and spatiotemporal resolutions, heterogeneity and inconsistency at the data level can be eliminated, providing high-quality static geographic parameters and dynamic underlying surface parameters that can be directly input into regional climate models. The purpose of preprocessing is to ensure the spatiotemporal comparability and consistency of multi-source data, avoiding additional simulation biases introduced by differences in data quality or format. The construction of the static geographic dataset aims to provide regional climate models with a fixed field of topographic-related parameters that do not change over time during the simulation period, used to correct for topographic effects at the model's grid scale. The construction of the dynamic underlying surface dataset aims to provide regional climate models with annually updated land use and surface parameters, enabling the models to reflect the interannual variation characteristics of the Loess Plateau's underlying surface, thereby improving the timeliness and accuracy of land surface process simulations.
[0027] Specifically, regarding the preprocessing steps, the first step is data format standardization. Global climate model output data typically uses NetCDF or GRIB formats, ground observation data is mostly in text tables or specialized meteorological data formats, satellite remote sensing data is mostly in HDF or GeoTIFF formats, and ecological characteristic data may originate from field surveys or Excel spreadsheets, resulting in significant format differences. Data format standardization involves converting data from these different sources into a common data format that regional climate models can recognize. This standardization also unifies the data's coordinate system (e.g., to WGS84 geographic coordinates or Albers iso-area projection coordinates), as well as the data storage structure and variable naming rules.
[0028] Secondly, spatiotemporal resolution is unified. In terms of temporal resolution, hourly observation data is aggregated or interpolated to the time step required for model integration, and daily data is organized into a continuous time series. In terms of spatial resolution, raster data with different spatial resolutions (e.g., remote sensing data may be 1km, 500m, or even finer) are uniformly resampled to the model grid resolution using nearest neighbor interpolation or bilinear interpolation methods. During this process, care must be taken to maintain the appropriate interpolation method for discontinuous variables such as precipitation to avoid introducing unreasonable values.
[0029] Missing value imputation is performed again. Ground observation stations may have missing data due to instrument malfunction, communication interruption, or maintenance, while satellite remote sensing may also have data gaps due to cloud cover or orbital period. Missing value imputation uses a combination of methods: for short-term missing data in a single station's time series (such as several consecutive hours), linear interpolation or spline interpolation methods are used to complete the data; for missing data over a longer period, the correlation regression relationship of neighboring stations is used for estimation; for missing pixels in remote sensing raster, spatiotemporal kriging interpolation or substitution methods based on similar pixels are used to ensure the integrity and continuity of the dataset.
[0030] Then, outlier detection and removal are performed. Outliers may originate from instrument malfunctions, abnormal records of extreme weather events, or errors in data transmission. Statistical methods are used for outlier identification. For example, for temperature series, the multi-year monthly average and standard deviation for each station are calculated. Records that exceed the average by several times plus or minus the standard deviation and fail the physical plausibility test (such as temperature abrupt changes exceeding the climatologically reasonable threshold) are marked as outliers. The locations of removed outliers are also filled using the missing value imputation method described above to ensure the continuity of the time series.
[0031] Finally, spatiotemporal alignment of multi-source data is performed. Spatiotemporal alignment matches the processed data within a unified temporal and spatial reference framework, enabling data from different sources at the same time, grid point, or site location to correspond and be compared. Specifically, this includes temporal alignment (unifying the timestamps of different data to the same time period) and spatial alignment (mapping site observation data to the model grid through spatial interpolation, while simultaneously extracting remote sensing raster data to the site location), ensuring that subsequent model validation, error correction, and evaluation index calculations can be performed based on consistent data pairs.
[0032] In some embodiments of this application, after preprocessing and spatiotemporal matching of the multi-source data, the method further includes: performing quality assessment on the preprocessed multi-source data to determine the preprocessing quality score for each type of multi-source data. The quality assessment includes data integrity assessment, spatiotemporal continuity assessment, and outlier removal rate assessment. A pre-set data quality qualification threshold is then set. The preprocessing quality score is compared with the data quality qualification threshold. If the preprocessing quality score is greater than or equal to the data quality qualification threshold, the corresponding type of multi-source data is determined to be qualified. If the preprocessing quality score is less than the data quality qualification threshold, the corresponding type of multi-source data is determined to be unqualified, and a re-preprocessing process is triggered until the preprocessing quality scores of each type of multi-source data are all greater than or equal to the data quality qualification threshold.
[0033] Understandably, by conducting a systematic quality assessment of the preprocessed multi-source data and establishing quantitative standards for data access, it is possible to ensure that all types of data entering the subsequent dynamic downscaling framework meet the preset quality requirements. Specifically, this step aims to address the technical deficiency of lacking closed-loop quality monitoring in the preprocessing stage, avoiding the cascading deterioration effect of "low-quality data input—inferior simulation results output" caused by problems such as incorrect data format conversion, unreasonable interpolation, outlier residues, or spatiotemporal matching deviations. By setting a data quality qualification threshold and executing a mandatory reprocessing loop, a quality control barrier is established at the source of data input, ensuring the basic quality of static geographic datasets, dynamic underlying surface datasets, and driving data, thereby laying a data foundation for the reliability of subsequent dynamic downscaling simulations, error correction, and evaluation index calculations.
[0034] In some embodiments of this application, a multi-layered nested dynamic downscaling framework is constructed based on regional climate models and static geographic datasets and dynamic underlying surface datasets, including: The parent region covers the entire country with a horizontal resolution of 30km; the sub-region covers the Loess Plateau region with a horizontal resolution of 10km; the grandchild region covers typical ecologically fragile areas of the Loess Plateau with a horizontal resolution of 3km; each nested region adopts a two-way nesting scheme, with the sub-region obtaining boundary conditions from the parent region, and the simulation results of the sub-region feeding back to affect the parent region. Based on the complex topographic features of the Loess Plateau, the subgrid topographic parameterization scheme in the regional climate model is activated to correct the deviation between the model grid topographic height and the actual topographic height. Based on static geographic datasets and dynamic underlying surface datasets, we optimized the combination of cumulus convection parameterization schemes, microphysical schemes, boundary layer schemes, and land surface process schemes in regional climate models. Using regional surface observation data as a benchmark, we selected and determined the optimal combination of parameterization schemes for each season / region.
[0035] Understandably, by rationally setting up a multi-layered nested grid structure, the dual simulation requirements of large-scale circulation fields and local refined climate characteristics can be met under limited computing resources. By activating the sub-grid topographic parameterization scheme, the systematic deviation between the model grid topographic height and the actual topographic height under the complex terrain of the Loess Plateau can be eliminated. Through the combination and optimization of multiple physical parameterization schemes, the physical configuration with the best simulation performance can be selected for different seasons and different sub-regions. Thus, at the physical mechanism level, the applicability and accuracy of the dynamic downscaling framework for climate simulation in the Loess Plateau region can be maximized.
[0036] Specifically, regarding the grid setting of the multi-layered nested dynamic downscaling framework, this invention adopts a three-layer unidirectional nested structure. The parent region has a horizontal resolution of 30 km, covering the entire country. Its function is to provide a large-scale circulation background field for the inner regions, enabling the regional climate model to fully capture large-scale forcing information such as the East Asian monsoon system, westerly disturbances, and the dynamic uplift of the Tibetan Plateau. The sub-region has a horizontal resolution of 10 km, covering the entire Loess Plateau region. Its function is to preliminarily characterize the topographic relief and climate zoning characteristics within the Loess Plateau while maintaining large-scale circulation features. The grandchild region has a horizontal resolution of 3 km, covering typical ecologically fragile areas of the Loess Plateau. This region is determined according to the ecological functional zoning of the Loess Plateau, for example, selecting areas with severe soil erosion or significant vegetation degradation in the heart of the Loess Plateau. The high resolution of 3 km can more finely distinguish gully topography, valley wind circulation, and local convection processes, significantly improving the simulation capabilities for extreme precipitation and local temperature changes. The three grid layers employ a bidirectional nesting approach, meaning each nested region uses a two-way nesting scheme, with the sub-region obtaining boundary conditions from the parent region, and the simulation results of the sub-region influencing the parent region. Furthermore, each grid layer uses the same projection method and geographic coordinate benchmark to ensure clear and unambiguous spatial correspondences between grid points.
[0037] Regarding the activation and implementation of the subgrid topographic parameterization scheme, in the Loess Plateau region, due to the high degree of topographic fragmentation and crisscrossing gullies, there is a significant deviation between the default grid average topographic height of the regional climate model and the actual topographic height. This deviation leads to systematic errors in the model's calculations of surface radiation, sensible heat flux, latent heat flux, and near-surface air temperature and wind field. This invention activates the subgrid topographic parameterization scheme built into the regional climate model. The basic principle of this scheme is as follows: within each model grid cell, the subgrid distribution characteristics of topographic height are statistically analyzed using a high-resolution digital elevation model, including the mean, standard deviation, and skewness of the topographic height within the grid. Then, based on these statistics, the grid cell is divided into multiple topographic categories (such as valley floor, slope, and slope top). For each topographic category, the received solar shortwave and longwave radiation, surface energy budget, and near-surface turbulent exchange processes are calculated separately. Finally, the calculation results of all topographic categories are summarized by area weight to obtain the net flux of the grid cell. By employing the above method, the deviation between the model grid terrain height and the actual terrain height is corrected, and the flux calculation at the grid scale more closely approximates the comprehensive effect of real terrain, thereby improving the simulation results of temperature, precipitation, and boundary layer structure in complex terrain regions. This parameterization scheme remains active on both the sub-regional and sub-regional grids of the entire Loess Plateau.
[0038] Regarding the optimal combination of physical parameterization schemes, this invention systematically optimizes the combination of four key physical process parameterization schemes in regional climate models: cumulus convection parameterization scheme, microphysics scheme, boundary layer scheme, and land surface process scheme. The cumulus convection parameterization scheme is used to handle convective precipitation processes that cannot be directly resolved by the model grid; different schemes differ in their descriptions of convection triggering mechanisms and precipitation timescales. The microphysics scheme describes the formation, growth, and transformation processes of condensates such as cloud droplets, ice crystals, raindrops, and snow crystals; different schemes vary in the level of detail in their characterization of condensate types and phase transition processes. The boundary layer scheme handles turbulent mixing and vertical transport processes within the planetary boundary layer; different schemes differ in their descriptions of turbulent closure order and convective mixing. The land surface process scheme describes the exchange of momentum, energy, and moisture between the Earth's surface and the atmosphere; different schemes differ in their characterization of soil stratification, vegetation parameters, and runoff processes.
[0039] The specific implementation method of the optimal combination is as follows: First, based on the climate characteristics of the Loess Plateau region and existing literature surveys, multiple alternative schemes are selected from each scheme category (for example, 3 schemes for cumulus convection, 3 schemes for microphysics, 2 schemes for boundary layer, and 2 schemes for land surface processes, forming a total of 36 combinations); Second, under the same multi-layered nested framework and the same external forcing driving field conditions, simulation integration is performed for each scheme combination for historical periods; Third, the results of key climate elements such as temperature and precipitation obtained from the simulation of each scheme combination are compared and verified with regional ground observation data station by station and time by time, and the deviation statistics of each scheme combination (such as mean deviation, root mean square error, correlation coefficient, etc.) are calculated; Finally, based on the deviation statistics of each season (spring, summer, autumn, and winter) and each sub-region (such as different ecological zones of the Loess Plateau), the scheme combination that performs best in the season and region is selected as the recommended parameterized configuration for the season and region for subsequent downscaling simulations.
[0040] In some embodiments of this application, spectral approximation coefficients are introduced to constrain large-scale circulation, including: During the integration process of the regional climate model, a Newtonian relaxation term is applied to the spectral components above a certain wavenumber so that the large-scale circulation of the regional climate model approximates the concurrent large-scale circulation of the driving field. The spectral approximation coefficients are dynamically adjusted based on the simulation deviation characteristics of different regions and seasons of the Loess Plateau; The spectral approximation coefficients are determined according to the following formula: α(t,x)=αbase×[1+β×B(t,x)]; Where α(t,x) represents the spectral approximation coefficient at position x at time t, αbase represents the basic spectral approximation coefficient, B(t,x) represents the deviation index based on historical simulation deviation statistics, and β represents the weighting coefficient of the deviation index.
[0041] Understandably, by applying Newtonian relaxation constraints in the spectral space to the large-scale circulation field during the integration process of regional climate models, the large-scale circulation evolution within the model does not deviate from the large-scale characteristics of the driving field, while preserving the model's responsiveness to small-scale local forcing. On this basis, by dynamically adjusting the spectral approximation coefficients according to the simulation bias characteristics of different regions and seasons of the Loess Plateau, the shortcomings of fixed coefficients in taking into account the constraint strength under different simulation conditions are overcome, achieving an adaptive balance between large-scale constraints and local degrees of freedom, thereby improving the overall simulation performance of dynamic downscaling results under various spatiotemporal conditions.
[0042] Specifically, regarding the basic principles and implementation methods of spectral approximation, during the integration process of a regional climate model, model state variables (such as wind field, temperature field, and geopotential height field) are transformed to wavenumber space at each integration step. This involves decomposing the gridded field in the spatial domain into spectral components with different wavenumbers using a Fourier transform. Low wavenumber components correspond to large-scale circulation characteristics, while high wavenumber components correspond to mesoscale and local-scale circulation characteristics. Based on this, a Newtonian relaxation term is applied to spectral components above a specific wavenumber threshold (i.e., the cutoff wavenumber). Specifically, this is achieved by adding a relaxation forcing term proportional to the difference between the driving field's corresponding spectral component and the right-hand side of the model's governing equations. This term is expressed as follows: for each spectral component to be constrained, its rate of change equals the contribution of the original physical process plus the relaxation coefficient multiplied by the difference between the driving field's spectral component and the model's current spectral component. Through this method, spectral components above the cutoff wavenumber are continuously pulled towards the direction of the driving field's concurrent large-scale circulation during the integration process, thus keeping the model's large-scale circulation consistent with the driving field and avoiding circulation drift caused by systematic model biases. The specific value of the cutoff wavenumber is determined comprehensively based on the grid resolution of the regional climate model, the size of the simulated region, and the climate characteristics of the Loess Plateau region. The preferred wavelength range corresponding to the cutoff wavenumber is 1500 to 2500 kilometers. The high wavenumber components below the cutoff wavenumber (i.e., small-scale fluctuations) are not subject to relaxation constraints and are allowed to develop freely within the model's own dynamic framework and physical parameterization scheme, in order to preserve the regional model's ability to add value to the simulation of local topography, land surface processes, and small- and medium-scale weather systems.
[0043] Regarding the shortcomings of fixed spectral approximation coefficients and the necessity of dynamic adjustment in this invention, existing technologies typically set spectral approximation coefficients to a fixed constant, which remains consistent throughout the simulation period and across all spatial grid points. However, the Loess Plateau has complex topography, with significant differences in topographic features and underlying surface conditions across different regions (such as the northern aeolian sandy area, the central hilly and gully area, and the southern plateau and terrace area). Furthermore, the large-scale circulation background varies drastically across different seasons (in summer influenced by the East Asian summer monsoon, and in winter by the westerly winds and the Mongolian cold high pressure), resulting in significant differences in model simulation bias characteristics across regions and seasons. Using fixed-coefficient spectral approximation may impose excessive constraints in regions and seasons with smaller simulation biases, suppressing beneficial adjustments to local features by the regional model; while in regions and seasons with larger simulation biases, the constraints may be insufficient, failing to effectively suppress circulation drift. Therefore, this invention employs dynamically adjusted spectral approximation coefficients, enabling them to adaptively change according to specific spatiotemporal locations and simulation bias conditions.
[0044] In some embodiments of this application, the dual error correction includes primary statistical error correction and secondary physical constraint error correction. Specifically, the primary statistical error correction involves using a quantile mapping method to correct the probability distribution deviation of climate elements based on the statistical characteristics of the deviation between historical multi-source observation data and dynamic downscaling output. The secondary physical constraint error correction involves verifying the physical rationality of the primary correction result based on the physical consistency constraints of the land-atmosphere coupling process in the Loess Plateau region. When the corrected result does not conform to the physical constraints, physical coordination adjustments are made. The physical constraints include the surface energy balance equation and the regional water balance equation.
[0045] Understandably, by introducing a dual error correction mechanism combining statistical correction and physical constraints, the shortcomings of a single statistical correction method—which only focuses on probability distribution matching and ignores physical consistency—can be overcome. The primary statistical error correction aims to eliminate probability distribution distortions caused by systematic biases in the model in the dynamic downscaling output, making the corrected climate elements more consistent with observations in terms of statistical distribution characteristics. The secondary physical constraint error correction aims to verify the physical laws governing the statistically corrected results, ensuring that the corrected climate elements satisfy basic physical constraints such as surface energy balance and regional water balance, avoiding physical inconsistencies that may be introduced by statistical correction. This simultaneously improves the quality of the final data product in terms of both statistical accuracy and physical rationality.
[0046] In some embodiments of this application, a climate simulation evaluation index system for the Loess Plateau region is constructed. Based on the high-resolution regional climate element dataset and regional ground observation data, a comprehensive simulation performance index is determined. This includes: the evaluation index system comprising climate mean-state simulation evaluation indicators, interannual variability simulation evaluation indicators, extreme event simulation evaluation indicators, and spatial distribution simulation evaluation indicators; the climate mean-state simulation evaluation indicators include multi-year average temperature deviation, multi-year average precipitation deviation, and seasonal cycle correlation coefficient; the interannual variability simulation evaluation indicators include the interannual temperature variability ratio and the interannual precipitation variability ratio; the extreme event simulation evaluation indicators include the extreme high temperature simulation index and the extreme precipitation simulation index; the spatial distribution simulation evaluation indicators include the spatial correlation coefficient of temperature and the spatial correlation coefficient of precipitation; the subjective weights of each evaluation indicator are determined based on the analytic hierarchy process (AHP) combined with expert experience, and the objective weights are determined based on the historical simulation deviation statistics of each evaluation indicator. A weighted summation is then used to obtain the comprehensive weight of each evaluation indicator; and the comprehensive simulation performance index is determined based on the comprehensive weights and the values of each evaluation indicator.
[0047] In some embodiments of this application, the formula for calculating the comprehensive simulation performance index is: ; Where PI represents the comprehensive simulation performance index, wi represents the comprehensive weight of the i-th evaluation index, and si represents the value of the i-th evaluation index.
[0048] Understandably, by constructing a systematic evaluation index system covering four dimensions—climate mean, interannual variability, extreme events, and spatial distribution—the limitations of single indicators in comprehensively reflecting regional climate simulation performance can be overcome, enabling multi-angle and multi-scale quantitative evaluation of the simulation capabilities of dynamic downscaling systems. Based on this, a comprehensive weighting method combining subjective weights determined by expert experience using the analytic hierarchy process (AHP) with objective weights determined by historical simulation bias statistics is employed. This approach balances prior knowledge of the importance of indicators with data-driven objective information, forming a scientifically sound comprehensive simulation performance index. This provides a unified and comparable quantitative decision-making basis for the subsequent dynamic adjustment and optimization of downscaling system parameter configuration.
[0049] Specifically, regarding the overall architecture of the evaluation index system, this invention constructs an evaluation index system from four dimensions. Each dimension evaluates different aspects of simulation performance, complementing each other to form a complete evaluation framework.
[0050] The first dimension is the climate mean state simulation evaluation index, used to assess the ability of the dynamic downscaling system to simulate the long-term mean climate state of the Loess Plateau region. This dimension includes three specific indicators: multi-year mean temperature bias, calculated as the multi-year average difference between simulated and observed temperatures at all grid points or stations within the simulation period, reflecting the model's systematic cold or warm bias in relation to the regional mean thermal state; multi-year mean precipitation bias, calculated as the multi-year average difference between simulated and observed precipitation at all grid points or stations within the simulation period, reflecting the model's systematic dry or wet bias in relation to the regional mean moisture state; and seasonal cycle correlation coefficient, calculated as the correlation coefficient between the simulated and observed monthly climate mean sequences, reflecting the model's ability to reproduce the phase and amplitude of intra-annual seasonal variations in climate elements. This indicator is superior to the simple monthly bias and can comprehensively evaluate the waveform similarity of the seasonal cycle.
[0051] The second dimension is the interannual variability simulation evaluation index, used to assess the ability of the dynamic downscaling system to simulate the amplitude of interannual climate fluctuations in the Loess Plateau region. This dimension includes two specific indicators: the temperature interannual variability ratio, calculated as the ratio of the standard deviation of annual temperature anomalies during the simulation period to the standard deviation of observed annual temperature anomalies. A ratio close to 1 indicates that the model's simulation of the amplitude of interannual temperature fluctuations is relatively accurate; a ratio greater than 1 indicates that the simulated interannual fluctuations are too drastic; and the precipitation interannual variability ratio, calculated as the ratio of the standard deviation of annual precipitation anomalies during the simulation period to the standard deviation of observed annual precipitation anomalies. Its meaning is similar to the temperature interannual variability ratio, used to evaluate the model's ability to simulate interannual precipitation variability.
[0052] The third dimension is the extreme event simulation evaluation index, used to assess the model's ability to simulate extreme climate events in the Loess Plateau region using a dynamic downscaling system. This dimension includes two specific indicators: the extreme high temperature simulation index, calculated as the ratio or deviation between the simulated and observed values of the extreme high temperature threshold (e.g., the 95th percentile of daily maximum temperature) or the number of extreme high temperature days (the number of days exceeding the threshold) during the simulation period, reflecting the model's ability to capture heat waves and extreme warm events; and the extreme precipitation simulation index, calculated as the ratio or deviation between the simulated and observed values of the extreme precipitation threshold (e.g., the 95th or 99th percentile of daily precipitation) or the extreme precipitation contribution rate (the proportion of extreme precipitation to total precipitation) during the simulation period, reflecting the model's ability to simulate heavy rain and extreme wet events. In the calculation of these extreme indicators, the thresholds are independently determined based on the historical observation sequences of each grid point or station to ensure the comparability of the indicators across different geographical locations.
[0053] The fourth dimension is the spatial distribution simulation evaluation index, used to assess the ability of the dynamic downscaling system to simulate the spatial patterns of climate elements in the Loess Plateau region. This dimension includes two specific indicators: the spatial correlation coefficient of temperature, calculated as the spatial correlation between the simulated multi-year average temperature field and the observed temperature field within a specific time period, reflecting the model's ability to reproduce spatial temperature distribution patterns (such as north-south temperature differences, temperature differences between mountains and valleys); and the spatial correlation coefficient of precipitation, calculated as the spatial correlation between the simulated multi-year average precipitation field and the observed precipitation field, reflecting the model's ability to reproduce spatial precipitation distribution patterns (such as rainy areas on windward slopes, dry areas on leeward slopes, etc.). The closer the spatial correlation coefficient is to 1, the more accurate the model's simulation of the spatial patterns of climate elements. The calculation of each indicator is based on high-resolution regional climate element datasets and regional surface observation data under the same spatiotemporal reference, and the quantitative value of each indicator is calculated for each indicator, denoted as si.
[0054] Regarding the method for determining the comprehensive weights, this invention adopts a comprehensive weighting strategy that combines subjective and objective weighting. Subjective weights are determined using the analytic hierarchy process (AHP) combined with expert experience. The specific implementation is as follows: First, a hierarchical model is established. The target layer is the comprehensive simulation performance index, the criterion layer consists of the four evaluation dimensions mentioned above, and the indicator layer consists of the specific evaluation indicators under each dimension. Second, a judgment matrix is constructed. Several experts in the field of climate research in the Loess Plateau region are invited to conduct pairwise comparisons of the relative importance of each evaluation indicator using the Saaty scaling method (1 to 9 proportional scale). For example, experts use their professional experience to judge the relative importance of the interannual temperature variability ratio and the interannual precipitation variability ratio to the comprehensive performance evaluation and provide scaling values. Then, a consistency test is performed on the judgment matrix provided by each expert, and the consistency ratio is calculated. If the consistency ratio is less than 0.1, the judgment matrix is considered to have passed the consistency test; otherwise, it is returned to the experts for re-judgment. Finally, for each expert's judgment matrix that has passed the consistency test, an eigenvector is calculated as the weight vector for each indicator. Then, the arithmetic mean of all experts' weight vectors is taken to obtain the subjective weight of each evaluation indicator.
[0055] Objective weights are determined based on historical simulation deviation statistics for each evaluation indicator. The basic idea is that for a given evaluation indicator, if the deviation fluctuates significantly across different historical simulations, it indicates that the indicator is more sensitive to changes in model configuration, contains more information about simulation performance differences, and should be assigned a higher objective weight. Conversely, if the deviation of the indicator is relatively stable across different experiments, its discriminative ability is limited, and it should be assigned a lower objective weight. The specific calculation method is as follows: collect the deviation values of each evaluation indicator from multiple sets of historical simulations (e.g., simulation results from different years or under different parameter configurations), calculate the deviation statistics (such as absolute deviation or root mean square error) for each indicator in each set of experiments, then calculate the standard deviation of each indicator's deviation value across different experiments, and normalize this standard deviation (e.g., divide by the sum of the standard deviations of all indicators). The normalized result is the objective weight of that indicator.
[0056] After obtaining the subjective and objective weights, a weighted summation method is used to calculate the comprehensive weight of each evaluation indicator. The formula is: the comprehensive weight equals the weighted average of the subjective and objective weights. An adjustment coefficient is used to balance the contribution ratio of subjective experience and objective data in the final weight. The preferred range for the adjustment coefficient is 0.3 to 0.7, and its specific value is determined based on expert consensus and practical application needs. This comprehensive weighting method incorporates experts' prior knowledge of key physical processes in climate simulation of the Loess Plateau region while avoiding the arbitrariness that may arise from subjective judgment, making the weight allocation more scientific and reasonable.
[0057] In some embodiments of this application, the parameter configuration and update strategy of the downscaling system is dynamically adjusted according to the comprehensive simulation performance index, including: pre-setting a first performance threshold and a second performance threshold, wherein the first performance threshold is less than the second performance threshold; if the comprehensive simulation performance index is less than or equal to the first performance threshold, a parameterization scheme re-optimization process is triggered; if the comprehensive simulation performance index is greater than the first performance threshold and less than or equal to the second performance threshold, the parameter update frequency of the current configuration is adjusted; if the comprehensive simulation performance index is greater than the second performance threshold, the current configuration is output as the recommended configuration, and a multi-level climate element map is output according to the recommended configuration.
[0058] Understandably, by pre-setting tiered performance thresholds, the quantitative evaluation results of the comprehensive simulation performance index are transformed into executable adaptive control commands, establishing a closed-loop optimization mechanism of "performance evaluation - decision-making - configuration adjustment". Specifically, this strategy aims to take differentiated control measures for different levels of simulation performance: when performance is severely insufficient, a fundamental re-optimization of the parameterized scheme is triggered, reconstructing the mode configuration from the physical mechanism level; when performance is close to but has not yet reached the target, the simulation effect is gradually improved by adjusting the parameter update frequency; when performance reaches an excellent level, the current configuration is locked and the final product is output. Through the above-mentioned tiered response strategy, computational resource consumption is minimized while ensuring simulation quality, achieving adaptive optimization of downscaled system configuration and efficient generation of the final product.
[0059] On the other hand, see Figure 2 This application also provides a climate-dynamic downscaling system for the Loess Plateau region, used to apply the aforementioned climate-dynamic downscaling method for the Loess Plateau region, including: The data preprocessing module is configured to acquire multi-source data of the Loess Plateau region, and to preprocess and perform spatiotemporal matching on the multi-source data to construct a static geographic dataset and a dynamic underlying surface dataset of the Loess Plateau. The multi-source data includes: coarse-resolution output data of global climate models, regional ground observation data, satellite remote sensing underlying surface data, and ecological characteristic data of the Loess Plateau. The dynamic downscaling module is configured to construct a multi-layered nested dynamic downscaling framework based on regional climate models, static geographic datasets and dynamic underlying surface datasets, and introduce spectral approximation coefficients to constrain large-scale circulation, thereby obtaining dynamic downscaling output results. The error correction module is configured to perform dual error correction on the dynamic downscaling output based on multi-source observation data to obtain a high-resolution regional climate element dataset; the multi-source observation data includes regional ground observation data, satellite remote sensing underlying surface data, and Loess Plateau ecological characteristic data. The climate element output module is configured to construct a climate simulation evaluation index system for the Loess Plateau region. Based on the high-resolution regional climate element dataset and regional ground observation data, it determines the comprehensive simulation performance index and dynamically adjusts the parameter configuration and update strategy of the downscaling system according to the comprehensive simulation performance index, and outputs a multi-level climate element map.
[0060] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0061] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0062] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0063] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0064] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.
Claims
1. A method for downscaling the climate dynamics of the Loess Plateau region, characterized in that, include: Multi-source data of the Loess Plateau region were acquired, and the multi-source data were preprocessed and spatiotemporally matched to construct a static geographic dataset and a dynamic underlying surface dataset of the Loess Plateau. The multi-source data included: coarse-resolution output data of global climate models, regional ground observation data, satellite remote sensing underlying surface data, and ecological characteristic data of the Loess Plateau. Based on regional climate models, a multi-layered nested dynamic downscaling framework is constructed using static geographic datasets and dynamic underlying surface datasets. Spectral approximation coefficients are introduced to constrain large-scale circulation, resulting in dynamic downscaling outputs. The dynamic downscaling output results are subjected to dual error correction based on multi-source observation data to obtain a high-resolution regional climate element dataset; the multi-source observation data includes regional ground observation data, satellite remote sensing underlying surface data, and Loess Plateau ecological characteristic data. A climate simulation evaluation index system for the Loess Plateau region is constructed. Based on the high-resolution regional climate element dataset and regional ground observation data, a comprehensive simulation performance index is determined. The parameter configuration and update strategy of the downscaling system are dynamically adjusted according to the comprehensive simulation performance index, and a multi-level climate element map is output.
2. The method for downscaling the climate dynamics of the Loess Plateau region according to claim 1, characterized in that, The regional ground observation data includes hourly / daily temperature, precipitation, wind speed, humidity, and air pressure observation data from meteorological stations throughout the Loess Plateau. The satellite remote sensing underlying surface data includes land use / cover change, vegetation index, and surface albedo of the Loess Plateau. The ecological characteristics data of the Loess Plateau include soil moisture, vegetation coverage, and degree of soil erosion.
3. The method for downscaling the climate dynamics of the Loess Plateau region according to claim 1, characterized in that, The multi-source data is preprocessed and spatiotemporally matched to construct a static geographic dataset and a dynamic underlying surface dataset for the Loess Plateau, including: The preprocessing includes data format standardization, spatiotemporal resolution unification, missing value imputation, outlier detection and removal, and spatiotemporal alignment of multi-source data. Based on preprocessed satellite remote sensing underlying surface data, annual land use / cover change data of the Loess Plateau are extracted by classification and regression methods to construct a time series dynamic underlying surface dataset. Based on the high-resolution digital elevation model, parameters of slope, aspect, and topographic relief of the Loess Plateau are extracted to construct a static geographic dataset.
4. The method for downscaling the climate and dynamics of the Loess Plateau region according to claim 3, characterized in that, After preprocessing and spatiotemporal matching of the multi-source data, the process further includes: The quality of the preprocessed multi-source data is assessed to determine the preprocessing quality score for each type of multi-source data. The quality assessment includes data integrity assessment, spatiotemporal continuity assessment, and outlier removal rate assessment. Pre-set a data quality qualification threshold; The preprocessing quality score is compared with the data quality pass threshold; If the preprocessing quality score is greater than or equal to the data quality qualification threshold, then the corresponding type of multi-source data is determined to be qualified. If the preprocessing quality score is less than the data quality pass threshold, the corresponding type of multi-source data is determined to be unqualified, and a re-preprocessing process is triggered until the preprocessing quality scores of all types of multi-source data are greater than or equal to the data quality pass threshold.
5. The method for downscaling the climate dynamics of the Loess Plateau region according to claim 1, characterized in that, Based on regional climate models, a multi-layered nested dynamic downscaling framework is constructed using static geographic datasets and dynamic underlying surface datasets, including: The parent region covers the entire country with a horizontal resolution of 30km; the sub-region covers the Loess Plateau region with a horizontal resolution of 10km; and the grandchild region covers typical ecologically fragile areas of the Loess Plateau with a horizontal resolution of 3km. Based on the complex topographic features of the Loess Plateau, the subgrid topographic parameterization scheme in the regional climate model is activated to correct the deviation between the model grid topographic height and the actual topographic height. Based on static geographic datasets and dynamic underlying surface datasets, we optimized the combination of cumulus convection parameterization schemes, microphysical schemes, boundary layer schemes, and land surface process schemes in regional climate models. Using regional surface observation data as a benchmark, we selected and determined the optimal combination of parameterization schemes for each season / region.
6. The method for downscaling the climate dynamics of the Loess Plateau region according to claim 1, characterized in that, Introducing spectral approximation coefficients to constrain large-scale circulation, including: During the integration process of the regional climate model, a Newtonian relaxation term is applied to the spectral components above a certain wavenumber so that the large-scale circulation of the regional climate model approximates the concurrent large-scale circulation of the driving field. The spectral approximation coefficients are dynamically adjusted based on the simulation deviation characteristics of different regions and seasons of the Loess Plateau; The spectral approximation coefficients are determined according to the following formula: α(t,x)=αbase×[1+β×B(t,x)]; Where α(t,x) represents the spectral approximation coefficient at position x at time t, αbase represents the basic spectral approximation coefficient, B(t,x) represents the deviation index based on historical simulation deviation statistics, and β represents the weighting coefficient of the deviation index.
7. The method for downscaling the climate dynamics of the Loess Plateau region according to claim 1, characterized in that, The dual error correction includes a first statistical error correction and a second physical constraint error correction; The first statistical error correction specifically involves: based on the statistical characteristics of the deviation between historical multi-source observation data and dynamic downscaling output results, using the quantile mapping method to correct the probability distribution deviation of climate elements; The secondary physical constraint error correction specifically involves: verifying the physical rationality of the results after the first correction based on the physical consistency constraints of the land-atmosphere coupling process in the Loess Plateau region; and making physical coordination adjustments when the corrected results do not conform to the physical constraints. Among them, physical constraints include the surface energy balance equation and the regional water balance equation.
8. The method for downscaling the climate dynamics of the Loess Plateau region according to claim 1, characterized in that, A climate simulation evaluation index system for the Loess Plateau region was constructed, and a comprehensive simulation performance index was determined based on the aforementioned high-resolution regional climate element dataset and regional ground observation data, including: The evaluation index system includes climate mean state simulation evaluation index, interannual variability simulation evaluation index, extreme event simulation evaluation index, and spatial distribution simulation evaluation index. The climate mean state simulation evaluation indicators include multi-year average temperature deviation, multi-year average precipitation deviation, and seasonal cycle correlation coefficient. The interannual variability simulation and evaluation indicators include the interannual variability ratio of temperature and the interannual variability ratio of precipitation; The extreme event simulation and evaluation indicators include the extreme high temperature simulation index and the extreme precipitation simulation index. The spatial distribution simulation evaluation indicators include the spatial correlation coefficient of temperature and the spatial correlation coefficient of precipitation; The subjective weights of each evaluation indicator are determined by combining the analytic hierarchy process with expert experience. The objective weights are then determined by statistical analysis of the historical simulation deviations of each evaluation indicator. The comprehensive weight of each evaluation indicator is obtained by weighted summation. The comprehensive simulation performance index is determined based on the comprehensive weights and the values of each evaluation index. The formula for calculating the comprehensive simulation performance index is: ; Where PI represents the comprehensive simulation performance index, wi represents the comprehensive weight of the i-th evaluation index, and si represents the value of the i-th evaluation index.
9. The method for downscaling the climate and dynamics of the Loess Plateau region according to claim 8, characterized in that, The parameter configuration and update strategy of the downscaling system is dynamically adjusted based on the comprehensive simulation performance index, including: A first performance threshold and a second performance threshold are preset, wherein the first performance threshold is less than the second performance threshold; If the overall simulation performance index is less than or equal to the first performance threshold, the parameterized scheme re-optimization process is triggered. If the overall simulation performance index is greater than the first performance threshold and less than or equal to the second performance threshold, then adjust the current configuration parameter update frequency. If the overall simulation performance index is greater than the second performance threshold, the current configuration is output as the recommended configuration, and a multi-level climate element map is output based on the recommended configuration.
10. A climate-dynamic downscaling system for the Loess Plateau region, used to apply the climate-dynamic downscaling method for the Loess Plateau region as described in any one of claims 1-9, characterized in that, include: The data preprocessing module is configured to acquire multi-source data of the Loess Plateau region, and to preprocess and perform spatiotemporal matching on the multi-source data to construct a static geographic dataset and a dynamic underlying surface dataset of the Loess Plateau. The multi-source data includes: coarse-resolution output data of global climate models, regional ground observation data, satellite remote sensing underlying surface data, and ecological characteristic data of the Loess Plateau. The dynamic downscaling module is configured to construct a multi-layered nested dynamic downscaling framework based on regional climate models, static geographic datasets and dynamic underlying surface datasets, and introduce spectral approximation coefficients to constrain large-scale circulation, thereby obtaining dynamic downscaling output results. The error correction module is configured to perform dual error correction on the dynamic downscaling output based on multi-source observation data to obtain a high-resolution regional climate element dataset; the multi-source observation data includes regional ground observation data, satellite remote sensing underlying surface data, and Loess Plateau ecological characteristic data. The climate element output module is configured to construct a climate simulation evaluation index system for the Loess Plateau region. Based on the high-resolution regional climate element dataset and regional ground observation data, it determines the comprehensive simulation performance index and dynamically adjusts the parameter configuration and update strategy of the downscaling system according to the comprehensive simulation performance index, and outputs a multi-level climate element map.
