Accumulated snow land surface assimilation method based on EnSRF + DI

Abstract

The invention discloses an'EnSRF + DI '-based accumulated snow land surface assimilation method, which comprises the following steps of: (1) land surface mode simulation: performing land surface mode accumulated snow simulation by taking atmosphere driving data and earth surface parameters as input data to obtain accumulated snow simulation data; step (2), construction of an accumulated snow coverage super observation field: using satellite remote sensing observation data and accumulated snow site observation data to construct the accumulated snow coverage super observation field; and (3) snow assimilation: performing snow land assimilation by combining snow simulation data obtained by land mode simulation with the snow coverage super observation field and adopting an EnSRF + DI assimilation method to obtain a snow land assimilation result. The method can provide a basis for obtaining accumulated snow assimilation analysis products with better product quality.

Description

Technical Field

[0001] The present invention relates to the technical field of snow cover land surface assimilation, and more specifically to a snow cover land surface assimilation method based on "EnSRF+DI". Background Art

[0002] Snow plays a crucial role in numerical weather forecasting, severe weather, climate change, and water resources. Snow's high reflectivity and low thermal conductivity affect the Earth's surface energy balance, thus influencing regional and even global weather and climate. Snow accumulation and melting influence the water cycle. Excessive snowfall can cause widespread snowstorms, impacting transportation, agriculture, animal husbandry, and the safety of people's lives and property. Rapid spring temperature rises can cause extreme hydrological events such as snowmelt floods. Therefore, accurate and timely understanding of the spatiotemporal variations of snow cover is crucial for numerical weather forecasting, climate change, and severe weather.

[0003] Stationary observations can provide information on snow cover at meteorological stations, but they cannot provide information on its spatial distribution and spatiotemporal variations. Satellite remote sensing technology, with its advantages of macroscopicity and short-term periodicity, can effectively compensate for the spatial distribution limitations of stationary observations. For example, visible light remote sensing can directly observe snow cover during cloudless daytime conditions, but it relies on sunlight and is severely affected by cloud cover, making it inoperable in cloudy skies. Passive microwave remote sensing, while penetrating and all-weather, is unaffected by weather conditions, but it has low spatial resolution and limited snow penetration. Simulating land surface processes with robust physical processes and dynamics, such as the new generation of land surface models CLM and Noah-MP, is one means of obtaining continuous snow depth over time and space. However, factors affecting land surface model simulation performance primarily include model parameterization schemes, surface parameters, and atmospheric driving data. Therefore, integrating the advantages of stationary observations, satellite data, and numerical simulation data, and using data assimilation methods to integrate stationary observations, satellite data, and model snow cover, is an effective means of obtaining high-quality, high-spatiotemporal resolution snow cover analysis products. Summary of the Invention

[0004] Therefore, the technical problem to be solved by the present invention is to provide a snow cover land surface assimilation method based on "EnSRF+DI" to improve the quality of snow cover land surface real-time products.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] A snow cover land surface assimilation method based on "EnSRF+DI" includes the following steps:

[0007] Step (1), land surface model simulation: using atmospheric driving data and surface parameters as input data to simulate snow accumulation in the land surface model to obtain snow accumulation simulation data;

[0008] Step (2), construction of a snow cover super observation field: constructing a snow cover super observation field using snow cover data observed by Fengyun geostationary satellites, snow depth data observed by Fengyun polar-orbiting satellites, and snow station observation data;

[0009] Step (3), snow cover assimilation: Using the snow cover simulation data obtained by land surface model simulation and the constructed snow cover super observation field, the "EnSRF+DI" assimilation method is used to perform snow cover land surface assimilation to obtain the snow cover land surface assimilation results.

[0010] In the above-mentioned snow cover land surface assimilation method based on "EnSRF+DI", in step (1), the atmospheric driving data are provided by the China Meteorological Administration's land surface data assimilation system CLDAS, including temperature, air pressure, wind speed, precipitation, humidity and incident ground solar shortwave radiation data; the surface parameters include soil texture, surface cover type, leaf area index, topography, etc.; the obtained snow cover simulation data include snow cover simulation data, snow depth simulation data and snow water equivalent simulation data.

[0011] In the above-mentioned snow cover land surface assimilation method based on "EnSRF+DI", in step (1), at least one of the CLM land surface model, the Noah land surface model and the NoahMP land surface model is used to perform land surface model simulation.

[0012] In the above-mentioned snow cover land surface assimilation method based on “EnSRF+DI”, in step (2), the snow cover station observation data comes from the quality-controlled national snow depth station observation data provided by TianQing, the meteorological big data platform of the National Meteorological Information Center.

[0013] In the above-mentioned snow cover land surface assimilation method based on "EnSRF+DI", in step (2), the satellite remote sensing observation data include the observation data provided by the Fengyun-4 geostationary meteorological satellite FY4A and the satellite observation data provided by the Fengyun-3 polar-orbiting meteorological satellite FY3D.

[0014] In the above-mentioned EnSRF+DI-based snow land surface assimilation method, in step (2), the method for constructing the snow cover super observation field includes the following steps:

[0015] Step (2-1), pre-processing the projection of the snow cover data retrieved from the Fengyun-4 geostationary meteorological satellite FY4A;

[0016] Step (2-2), preprocessing the snow depth data provided by the Fengyun-3 polar-orbiting meteorological satellite FY3D;

[0017] Step (2-3), performing spatiotemporal matching of the pre-processed FY4A snow cover data and FY3D snow depth data to obtain satellite snow cover data;

[0018] Steps (2-4) use monthly vegetation cover data, real-time precipitation data, and ground analysis fields of snow cover station observations to conduct collaborative quality control of satellite snow cover data;

[0019] Steps (2-5) use the quality-controlled satellite snow cover data and snow station observation data to construct a snow cover super observation field.

[0020] In the above-mentioned snow cover land surface assimilation method based on "EnSRF+DI", in step (3), the "EnSRF+DI" assimilation method includes the following steps:

[0021] Step (3-1), calculate the difference between the snow cover super observation field O and the snow cover simulated by the land surface model P, denoted as Y-HX b ;

[0022] Step (3-2): Use the random perturbation method based on normal distribution to perturb the snow cover simulated by the land surface model to obtain 50 sets of snow cover data and construct the background error covariance matrix P in the EnSRF assimilation method;

[0023] Step (3-3), Y-HX obtained by step (3-1) and step (3-2) b and P, using the EnSRF assimilation method to achieve snow-covered land surface assimilation;

[0024] Steps (3-4) perform DI judgment based on the relationship between whether the land surface model simulates snow accumulation and whether the snow-covered super observation field has snow accumulation. If the land surface model simulates snow accumulation and the snow-covered super observation field has no snow accumulation, set the melting rate to remove excess snow. If the land surface model simulates no snow accumulation and the snow-covered super observation field has snow accumulation, add some snow accumulation.

[0025] Steps (3-5) use the assimilated snow cover to update the snow depth and snow water equivalent based on the assimilated snow cover, combined with the relationship between snow cover, snow water equivalent, and snow depth.

[0026] In the above-mentioned snow cover land surface assimilation method based on “EnSRF+DI”, in step (3-1), the following formula is used to calculate Y-HX b :

[0027] X a =X b +K(Y-HX b );

[0028] Where, X ais the analytical field, X b is the background field, K is the Kalman gain, Y is the observation vector, and H is the observation operator;

[0029] Where K = PH T (HPH T +R) -1 ; R is the observation error matrix, P is the background error covariance matrix;

[0030] The observation operator H is

[0031] in,

[0032] BDSNO, SWE, SD, FMELT, and FSNO represent snow density, snow water equivalent, snow depth, intermediate snow density, and snow cover, respectively;

[0033] In step (3-2), the background error covariance matrix P is calculated as follows:

[0034]

[0035] In the above formula, X′ b is the set sample, and N is the number of set samples, that is, 50.

[0036] In the above-mentioned snow cover land surface assimilation method based on "EnSRF+DI", in steps (3-4), when the snow cover super observation site observes snow but the land surface model simulates no snow, snow is added to the snow water equivalent simulated by the land surface model; when the snow cover super observation site observes no snow but the land surface model simulates snow, the snow melting rate is set to reduce the snow. The calculation formula is:

[0037]

[0038] In the above formula, ΔSWE, SWE in 、SCF obs 、SCF model and β0 represent the change in snow water equivalent, input snow water equivalent, observed snow cover, model snow cover, and snowmelt rate, respectively.

[0039] In the above-mentioned snow land surface assimilation method based on "EnSRF+DI", the snow land surface assimilation results obtained by assimilation in step (3) include the snow depth assimilation analysis results, snow cover assimilation analysis results and snow water equivalent assimilation analysis results; when "EnSRF+DI" assimilation is run, configurable block operation is implemented.

[0040] The technical solution of the present invention achieves the following beneficial technical effects:

[0041] The present invention is based on the "EnSRF+DI" snow cover land surface assimilation method. By adopting the EnSRF+DI assimilation method and selecting appropriate set samples and observation operators, it can obtain better assimilation results for the situations where "snow is observed but not in the model, and snow is observed but not in the model", providing a basis for obtaining snow cover assimilation analysis products with better product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 The technical roadmap for snow cover land surface assimilation based on "EnSRF+DI" in the embodiment of the present invention;

[0043] Figure 2 A flowchart of the construction of a snow super observation site in an embodiment of the present invention;

[0044] Figure 3 Schematic diagram of the three-layer snow model of the NoahMP land surface model in an embodiment of the present invention;

[0045] Figure 4 Schematic diagram of automatic block assimilation operation in an embodiment of the present invention;

[0046] Figure 5 The efficiency of automatic assimilation and segmentation under different sets of sample data in the embodiment of the present invention;

[0047] Figure 6 Deviation (a), root mean square error (b), correlation coefficient (c), and root mean square error (d) under different sets of sample data in the embodiment of the present invention;

[0048] Figure 7 TS scores under different sets of sample data and different snow levels in the embodiment of the present invention;

[0049] Figure 8 Snow assimilation bias, root mean square error, correlation coefficient, hit rate, false alarm rate and TS score under different assimilation methods of DI, EnSRF and "EnSRF+DI" in the embodiment of the present invention;

[0050] Figure 9 Case analysis and comparison of snow assimilation under different assimilation methods in the embodiments of the present invention (the left picture takes the snow in the northwest region on May 6, 2021 as an example, and the right picture takes the snow in the northeast region on May 6, 2021 as an example). DETAILED DESCRIPTION

[0051] This embodiment takes the Fengyun satellite snow-covered land surface assimilation method as an example to further illustrate the snow-covered land surface assimilation method based on "EnSRF+DI" of the present invention.

[0052] 1. Overall technical route

[0053] Figure 1The technical route of the snow land surface assimilation method in this embodiment is given, which mainly includes land surface model simulation, construction of snow cover super observation field, and snow assimilation. In the land surface model simulation part, the CLM with biogeophysical-hydrological-biogeochemical processes is selected. The 3.5 land surface model, the Noah model that depicts the snow accumulation-sublimation-ablation-heat exchange process, and the Noah-MP land surface model with a multi-parameterization scheme for snow cover, runoff, and soil water use CLDAS-V2.0 surface parameters such as temperature, pressure, humidity, wind, precipitation, solar radiation, soil texture, and vegetation cover type as input data. After decades of spin-up, the simulation of snow cover in the CLM, Noah, and NoahMP land surface models was achieved. The construction of the snow cover super-observation field uses snow depth station observations, geostationary satellite inversion of snow cover, and polar-orbiting satellite inversion of snow depth to conduct collaborative quality control and construct a snow cover super-observation field. The snow assimilation part uses the "EnSRF+DI" assimilation method to determine the selection of ensemble samples and observation operators in the assimilation, and realizes the assimilation of snow cover to improve the model's snow depth and snow water equivalent, and obtains the assimilation analysis results of snow depth, snow water equivalent, and snow cover ratio after assimilation.

[0054] 2. China Meteorological Administration Land Surface Data Assimilation System

[0055] The China Meteorological Administration Land Data Assimilation System (CLDAS) was developed by Shi Chunxiang's research team at the National Meteorological Information Center of the China Meteorological Administration. The CLDAS-V2.0 atmospheric driving data mainly includes temperature, pressure, wind speed, humidity, precipitation, and incident ground solar shortwave radiation.

[0056] Among them, the temperature, pressure, humidity and wind driving data are mainly obtained by using the multi-grid variational analysis method, integrating the temperature, pressure, humidity and wind observation data of more than 2,400 national automatic stations and more than 40,000 regional automatic stations of the China Meteorological Administration with the ECMWF numerical forecast / ERA-Interim reanalysis products; the precipitation driving data are obtained by using the multi-grid variational analysis method, downscaling method and rain and snow identification method, integrating the precipitation observation data of more than 2,400 national automatic stations and nearly 70,000 regional automatic stations of the China Meteorological Administration with CMOPRH precipitation / EMSIP precipitation / MERRA2 precipitation; the radiation driving data are based on the DISORT radiation transfer model, with ozone, atmospheric precipitable water and surface air pressure in the GFS numerical analysis product as the dynamic input parameters of the radiation transfer model, and are formed by inverting the full-disk nominal map data of the FY2 geostationary satellite VIS channel. When the FY2 satellite is not available, the hybrid estimated station radiation data is fused with the ERA5 radiation data.

[0057] 3. Observational data

[0058] Currently, the China Meteorological Administration observes snow depth as the vertical depth from the snow surface to the ground. Manual snow depth measurements are made using a snow ruler or a standard meter stick, while automatic measurements are made using an ultrasonic snow depth detector. The snow depth observation data used in this paper's fusion and assimilation comes from over 2,400 national snow depth station observations on the Meteorological Big Data Cloud Platform, which have undergone quality control by Jiang Hui, an engineer in the Data Research Laboratory.

[0059] 4. Quality Control of Wind, Cloud and Snow Retrieval Products and Construction of Super Observation Sites

[0060] Fengyun snow inversion products are derived from the FY4A snow cover data and FY3D snow depth data from the National Satellite Meteorological Center. In view of the sensitivity of geostationary and polar-orbiting satellite snow inversion products to areas with high vegetation coverage and underlying surfaces in precipitation areas, and the relationship between snow cover and snow depth, combined with monthly vegetation coverage data, real-time precipitation data, and ground analysis fields of snow cover station observations, quality control of FY4A snow cover and FY3D snow depth is carried out, and a snow cover assimilation super observation field is constructed. The technical roadmap is as follows: Figure 2 As shown. First, the FY4A snow cover and FY3D snow cover are re-projected, and the spatiotemporal matching of geostationary satellites and polar-orbiting satellites is performed. Then, the ground analysis field of vegetation coverage, precipitation, and snow cover observations is used for quality control. The snow cover and snow depth are then coordinated for quality control. Finally, a snow cover super observation field is constructed. Specifically, the construction method of the snow cover super observation field includes the following steps:

[0061] Step (1), pre-processing the projection of the snow cover data retrieved from the Fengyun-4 geostationary meteorological satellite FY4A;

[0062] Step (2), preprocessing the snow depth data provided by the Fengyun-3 polar-orbiting meteorological satellite FY3D;

[0063] Step (3), performing spatiotemporal matching of the pre-processed FY4A snow cover data and FY3D snow depth data to obtain satellite snow cover data;

[0064] Step (4) uses monthly vegetation coverage data, real-time precipitation data, and ground analysis fields of snow cover station observations to conduct collaborative quality control of satellite snow cover data;

[0065] Step (5): Use the quality-controlled satellite snow cover data and snow station observation data to construct a snow cover super observation field.

[0066] The super observation field constructed in this embodiment integrates the results of FY4, FY3, and station observations, and can more comprehensively reflect the current spatial distribution of snow cover.

[0067] 5. CLM3.5, Noah, and Noah-MP land surface models

[0068] CLM3.5 incorporates new parameterization improvements to CLM3.0, enabling coupling with the LPJ model to simulate vegetation dynamics, enabling dynamic representation of vegetation in the land surface model. CLM includes up to five snow layers and ten unequally spaced soil layers. To represent the complexity of the land surface within the climate model grid, CLM employs a subgridding technique to account for underlying surface heterogeneity. Each CLM grid contains four possible land cover types: glaciers, wetlands, lakes, and vegetation. Within the vegetation layer, vegetation is categorized into different functional types based on their biophysical and chemical properties, for a total of 16 vegetation types. Each PFT solves its own hydrological and energy balance relationships and integrates them within each grid. Biophysical processes simulated in CLM3.5 include the interaction between shortwave and longwave radiation and the vegetation canopy and soil; momentum and turbulent fluxes in the soil and canopy; heat transfer between the soil and snow layer; hydrological processes in the canopy, soil, and snow layers; physiological changes in plant leaf stomata and photosynthesis.

[0069] The Noah land surface model was developed from the OSU-LSM (Oregon State University / Land Surface Model) and was officially named Noah in 2000. The soil moisture is calculated using the Richard equation and takes into account four soil layers, namely 0-10cm, 10cm-40cm, 40-100cm, and 100cm-200cm. After years of continuous improvement, the Noah land surface model has been widely used in the comprehensive simulation of land surface processes and in many foreign land surface data assimilation systems, such as the US Global Land Data Assimilation System GLDAS and the North American Land Data Assimilation System NLDAS. Figure 3 Noah characterizes the accumulation and ablation of snow, simulating the exchange of energy, momentum, and water between land and air based on a physical parameterization scheme. It also simulates snow accumulation, sublimation, melting, and heat exchange. The snow parameterization scheme in Noah was developed by Koren V et al. based on the NCEP climate model's snow cover and permafrost parameterization scheme. Snow depth is affected by snow water equivalent, snow density, snow surface temperature, and soil surface temperature, and is calculated from the snow water equivalent and snow density.

[0070] The Community Noah Land Surface Model with Multi-Parameterization Options (NOAH-MP) is a further development of the Noah land surface model. Unlike the Noah model, NOAH-MP separates vegetation from the land surface, revising the overall model framework and improving the energy balance of vegetation-covered areas, snow cover, frozen ground and infiltration, and soil moisture-groundwater interactions. Furthermore, the model offers thousands of parameterization scheme combinations for various physical process options, such as dynamic vegetation, runoff, and groundwater, allowing users to configure parameterization schemes according to their needs. The number of soil layers in NOAH-MP remains the same as in the Noah model, remaining four. This model is currently used in the NCAR High-Resolution Land Data Assimilation System (HRLDAS v3.6) and the China Meteorological Administration's Land Surface Data Assimilation System (CLDAS-V2.0). This paper uses the default Noah-MP scheme, with the specific parameterization combinations shown in Table 1. Regarding snow accumulation, the Noah model considers the soil and canopy as a whole. When the snow is thick, a large amount of energy is stored on the snow surface, causing the snow to melt and the surface temperature and soil temperature to drop. To address this problem, Noah-MP adopts a three-layer snow accumulation physics model and snow interception model, dividing the snow cover into three layers to represent the processes of infiltration, retention and refreezing in the snow, as well as energy transfer.

[0071] Table 1 Selection of Noah-MP parameterization schemes

[0072] Parameterized options Parameterized Scheme Dynamic vegetation Off (use leaf area index and maximum vegetation fraction) Canopy stomatal resistance Ball-Berry scheme Soil moisture factor on stomatal resistance Noah (using soil moisture) Runoff and groundwater Considering groundwater TOPMODEL Surface drag coefficient Noah's original plan Supercooled water No iteration Frozen soil permeability Linear, more permeable Radiative transfer Improved second-rate approximation Snow surface albedo CLASS Division of rain and snow Jordan (1991) used 2.5°C Lower boundary of soil temperature The bottom heat flux is zero, that is, no heat flux exchange Snow accumulation and soil temperature time plan Semi-secluded

[0073] 6. EnSRF+DI assimilation method

[0074] (1) EnSRF assimilation method

[0075] The assimilation algorithm is based on EnSRF (Ensemble Mean Square Filter). EnSRF is a new square root analysis scheme proposed by Whitaker et al. (2002) based on the standard ensemble Kalman filter. It does not require observation perturbations when calculating the analysis field ensemble, thus reducing or eliminating the sampling error caused by observation perturbations.

[0076] X a is the analytical field, X b is the background field, K is the Kalman gain, Y is the observation vector, and H is the observation operator.

[0077] X a =X b +K(Y-HX b );

[0078] The calculation method of the Kalman gain K is as follows, where P is the background error covariance matrix, R is the observation error matrix, and H is the observation operator.

[0079] K=PH T (HPH T +R) -1 ;

[0080] The background error covariance matrix P needs to be formed by constructing a set. The calculation formula of P is as follows, N is the number of sets, X′ b is a set of samples, X′ b It is a matrix obtained by subtracting the ensemble mean from each ensemble.

[0081]

[0082] X′ b =[X′1,X′2,X′3,X′4,…,X′ n ];

[0083] X′1=X j -∑ j＝1,n X j / n;

[0084] Satellite observations are of snow cover. In the land surface model, snow cover is a diagnostic measure of snow depth and snow water equivalent. Therefore, an observation operator H is required for snow cover assimilation. By assimilating the snow cover, the values ​​of snow depth and snow water equivalent are changed. The observation operator H in this embodiment is as follows:

[0085]

[0086] In the above formula, BDSNO, SWE, SD, FMELT, and FSNO represent snow density, snow water equivalent, snow depth, intermediate snow density, and snow cover, respectively.

[0087] (2) DI assimilation method

[0088] The basic idea of ​​the DI algorithm is to improve the incremental calculation method of snow water equivalent (state variable). The increment of snow water equivalent is calculated based on the relationship between observations and models. Observations and simulations (obs represents the observed snow cover, and model represents the simulated snow cover) are classified into the following relationships: (1) When obs has snow but model does not, a certain thickness of snow is added to the snow water equivalent of the model; (2) When obs has no snow but model does, the snow melting rate is set to reduce the snow. Specifically, if very little snow is observed on a specific day during the snow melt period, and there is a certain amount of snow in the model, a snow melt rate is assumed, and the corresponding state variable snow water equivalent is removed from the snow on a specific day. The snow melt rate is assumed to be uniform, and all the snow will melt within the specified number of days (but in reality, the snow melt rate is constantly decreasing due to the decrease in snow water equivalent). If the simulated snow water equivalent is less than 20 mm, all the snow will melt in one day. On the other hand, if the model simulates a snow water equivalent of less than 10 mm, and observations indicate that some snow cover is greater than 40%, snow cover is considered to be present, and the snow cover in the model is increased by 20 mm. Dynamic snow density is calculated (based on Noah's snow compaction algorithm) to convert the state variable snow water equivalent into snow depth.

[0089]

[0090] In the above formula, ΔSWE, SWE in 、SCF obs 、SCF model and β0 represent the change in snow water equivalent, input snow water equivalent, observed snow cover, model snow cover, and snowmelt rate, respectively.

[0091] (3) Improvement of assimilation efficiency

[0092] Considering the different snow cover in different periods, the different numbers of observations to be assimilated, and the fact that a large number of ensembles will lead to slow operation, the EnSRF assimilation method is optimized (see Figure 4 ), realize configurable block operation, so that the normal assimilation operation time is completed in about 1 minute (see Figure 5 ).

[0093] (4) Construction and selection of sample sets

[0094] In this embodiment, ensemble samples are generated by randomly perturbing the state variables to construct the background error covariance matrix. Therefore, the number of ensemble samples is important for the assimilation results. Therefore, 10 / 20 / 25 / 30 / 40 / 50 / 60 / 80 / 100 ensembles are used for snow assimilation experiments, and the time period is from December 1 to 31, 2021. The assimilation results are evaluated using station observations, and the bias, root mean square error, correlation coefficient, unbiased root mean square error, and TS scores at different magnitudes are calculated (see the results). Figures 6 and 7 From the perspective of bias, the deviation of snow cover assimilation under different sample sets is basically the same, and the deviation begins to decrease slightly starting from the 50th set. From the perspective of root mean square error and unbiased root mean square error, the RMSE of the 10-40 sets is basically around 3.5cm, and the root mean square error and unbiased root mean square error begin to decrease starting from the 50th set. From the perspective of correlation coefficient, the correlation coefficient begins to improve starting from the 50th set. From the perspective of TS scores for different snow cover levels (greater than 0cm, greater than 0.5cm, greater than 3cm, and greater than 5cm), the TS scores all show a certain improvement starting from the 50th set. Therefore, considering computational efficiency, this embodiment uses 50 sets for snow cover assimilation.

[0095] (5) Comparison of different assimilation methods

[0096] The choice of assimilation method is crucial to the assimilation effect. Therefore, this example uses DI, EnSRF, and "EnSRF+DI" different assimilation methods to verify the snow assimilation effect. The time period is selected from December 1 to 31, 2021 (see the results). Figure 8 ), in terms of bias, the snow depth of the three methods all have positive bias, among which the DI method has the largest bias. The bias of EnSRF and "EnSRF+DI" is basically similar from December 1st to 20th, and the bias of EnSRF is smaller than that of "EnSRF+DI" after the 12th and 20th. In terms of root mean square error, the DI method has the largest root mean square error, and the root mean square errors of EnSRF and "EnSRF+DI" are basically similar. In comparison, the root mean square error of "EnSRF+DI" method is smaller. In terms of correlation coefficient, the correlation coefficients of "EnSRF+DI" and DI method are both greater than those of EnSRF, among which the correlation coefficient of "EnSRF+DI" is slightly greater than that of DI method. In terms of hit rate, "EnSRF+DI" and DI method are basically similar, with the hit rate basically close to 1, which is significantly better than EnSRF. In terms of false alarm rate, "EnSRF+DI" has a lower false alarm rate than both DI and EnSRF. In terms of TS score, "EnSRF+DI" has the highest TS score, followed by the DI method, and EnSRF has the lowest TS score. Overall, "EnSRF+DI" has the best effect.

[0097] In order to further analyze the effects of DI, EnSRF and “EnSRF+DI” assimilation methods, the snow cover case in November 2021 was selected for analysis (see Figure 9 ). The left figure shows a case where snow is observed at stations and satellites in northwest and southeast Gansu, but not in the model. DI and "EnSRF+DI" show snow in northwest and southeast Gansu, while EnSRF fails to reflect the presence of snow. The right figure shows a case where snow is observed and satellites are present, but the model does. The DI method removes some snow, but some snow accumulation remains. Both EnSRF and "EnSRF+DI" show no snow. Therefore, from this example, DI performs better when snow is observed but not in the model, while EnSRF performs better when snow is observed but not in the model. "EnSRF+DI" combines the advantages of both methods and can effectively complement the model when snow is observed but not in the model, or vice versa.

Claims

1. A snow cover land surface assimilation method based on "EnSRF+DI", characterized by: The steps include: Step (1), land surface model simulation: using atmospheric driving data and surface parameters as input data to simulate snow accumulation in the land surface model to obtain snow accumulation simulation data; Step (2), construction of a snow cover super observation field: constructing a snow cover super observation field using satellite remote sensing observation data and snow cover station observation data; Step (3), snow cover assimilation: The snow cover simulation data obtained by land surface model simulation is combined with the snow cover super observation field, and the "EnSRF+DI" assimilation method is used to perform snow cover land surface assimilation to obtain the snow cover land surface assimilation results.

2. The snow cover land surface assimilation method based on "EnSRF+DI" according to claim 1 is characterized in that: In step (1), the atmospheric driving data include temperature, air pressure, wind speed, precipitation, humidity and incident ground solar shortwave radiation data provided by the China Meteorological Administration's Land Data Assimilation System (CLDAS); the surface parameters include soil texture, land cover type, leaf area index and topography; and the snow simulation data include snow cover simulation data, snow depth simulation data and snow water equivalent simulation data.

3. The snow cover land surface assimilation method based on "EnSRF+DI" according to claim 1 is characterized in that: In step (1), at least one of the CLM land surface model, the Noah land surface model and the NoahMP land surface model is used to perform land surface model simulation.

4. The snow cover land surface assimilation method based on "EnSRF+DI" according to claim 1 is characterized in that: In step (2), the snow depth observation data are derived from the quality-controlled national snow depth observation data provided by TianQing, a meteorological big data platform of the National Meteorological Information Center.

5. The snow cover land surface assimilation method based on "EnSRF+DI" according to claim 1 is characterized in that: In step (2), the satellite remote sensing observation data includes observation data provided by the Fengyun-4 geostationary meteorological satellite FY4A and satellite observation data provided by the Fengyun-3 polar-orbiting meteorological satellite FY3D.

6. The snow cover land surface assimilation method based on "EnSRF+DI" according to claim 5 is characterized in that: In step (2), the method for constructing the snow cover super observation field includes the following steps: Step (2-1), pre-processing the projection of the snow cover data retrieved from the Fengyun-4 geostationary meteorological satellite FY4A; Step (2-2), preprocessing the snow depth data provided by the Fengyun-3 polar-orbiting meteorological satellite FY3D; Step (2-3), performing spatiotemporal matching of the pre-processed FY4A snow cover data and FY3D snow depth data to obtain satellite snow cover data; Steps (2-4) use monthly vegetation cover data, real-time precipitation data, and ground analysis fields of snow cover station observations to conduct collaborative quality control of satellite snow cover data; Steps (2-5) use the quality-controlled satellite snow cover data and snow station observation data to construct a snow cover super observation field.

7. The snow cover land surface assimilation method based on "EnSRF+DI" according to claim 1 is characterized in that: In step (3), the "EnSRF+DI" assimilation method includes the following steps: Step (3-1), calculate the difference between the snow cover super observation field and the snow cover simulated by the land surface model, denoted as Y-HX b ; Step (3-2): Use the random perturbation method based on normal distribution to perturb the snow cover simulated by the land surface model to obtain 50 sets of snow cover data and construct the background error covariance matrix P in the EnSRF assimilation method; Step (3-3), Y-HX obtained by step (3-1) and step (3-2) b and P, using the EnSRF assimilation method to achieve snow-covered land surface assimilation; Steps (3-4) perform DI judgment based on the relationship between whether the land surface model simulates snow accumulation and whether the snow-covered super observation field has snow accumulation. If the land surface model simulates snow accumulation and the snow-covered super observation field has no snow accumulation, set the melting rate to remove excess snow. If the land surface model simulates no snow accumulation and the snow-covered super observation field has snow accumulation, add some snow accumulation. Steps (3-5) use the assimilated snow cover to update the snow depth and snow water equivalent based on the assimilated snow cover, combined with the relationship between snow cover, snow water equivalent, and snow depth.

8. The snow cover land surface assimilation method based on "EnSRF+DI" according to claim 7 is characterized in that: In step (3-1), the following formula is used to calculate Y-HX b : X a =X b +K(Y-HX b ); Where, X a is the analytical field, X b is the background field, K is the Kalman gain, Y is the observation vector, and H is the observation operator; Where K = PH T (HPH T +R) -1 ; R is the observation error matrix, P is the background error covariance matrix; The observation operator H is in, BDSNO, SWE, SD, FMELT, and FSNO represent snow density, snow water equivalent, snow depth, intermediate snow density, and snow cover, respectively; In step (3-2), the background error covariance matrix P is calculated as follows: In the above formula, X′ b is the set sample, and N is the number of set samples, that is, 50.

9. The snow cover land surface assimilation method based on "EnSRF+DI" according to claim 7 is characterized in that: In steps (3-4), when the snow cover super observation site observes snow but the land surface model simulates no snow, snow is added to the snow water equivalent simulated by the land surface model. When the snow cover super observation site observes no snow but the land surface model simulates snow, the snow melting rate is set to reduce the snow. The calculation formula is: In the above formula, ΔSWE, SWE in 、SCF obs 、SCF model and β0 represent the change in snow water equivalent, input snow water equivalent, observed snow cover, model snow cover, and snowmelt rate, respectively.

10. The snow cover land surface assimilation method based on "EnSRF+DI" according to claim 1 is characterized in that: The snow land surface assimilation results obtained in step (3) include the snow depth assimilation analysis results, snow cover assimilation analysis results, and snow water equivalent assimilation analysis results; when the "EnSRF+DI" assimilation is run, a configurable block operation is implemented.

Images