A method for assimilating FY-3 sea surface temperature data based on set optimal interpolation

Abstract

The application discloses a method for assimilating FY satellite sea surface temperature data based on set optimal interpolation, which comprises the following steps: obtaining multi-source sea temperature observation data and a numerical prediction model in the same region to construct a data set; calculating the correlation of each data point in the region with the sea temperature observation data and the numerical prediction model, and giving the data point a corresponding weight according to the correlation result; generating an interpolation result by using the optimal interpolation method to interpolate the sea temperature observation data based on the weight; and updating the numerical prediction model according to the interpolation result. The method has the advantages that: the multi-source sea temperature observation data are fused, the weight is generated according to the correlation of the sea temperature observation data and the numerical prediction model, the optimal interpolation method is used to interpolate the sea temperature observation data, the prediction deviation of the numerical model is corrected by using the data assimilation method, and a more reliable sea surface temperature distribution map is obtained.

Description

Technical Field

[0001] This invention relates to the field of marine data technology, and in particular to a method for assimilating Fengyun satellite sea surface temperature data based on ensemble optimal interpolation. Background Technology

[0002] Ocean data assimilation methods, by combining observational data and ocean models, can effectively improve the accuracy and predictive power of short-term climate forecasts. Satellite-sensed sea surface temperature (SST) data, with its advantages of high spatiotemporal resolution, global coverage, real-time acquisition, and long-term series, has increasingly become an indispensable data source for global and local SST data. Among them, the Fengyun-3 satellite, as my country's polar-orbiting meteorological satellite system, provides a large amount of high-resolution SST products, offering favorable conditions for improving the forecasting level of ocean data assimilation systems. However, the SST data acquired by the Fengyun satellite is subject to errors from various factors, such as cloud cover, atmospheric absorption, and sensor drift, which challenges its accuracy and reliability. To improve the quality of SST data, researchers have conducted research on satellite SST data assimilation, aiming to combine observational data with numerical model predictions to obtain more accurate SST distribution maps.

[0003] Currently, the most commonly used method for satellite sea surface temperature (SST) data assimilation is based on optimal interpolation. This method corrects and updates satellite SST data by weighted interpolation of observed data and numerical model predictions. However, existing optimal interpolation methods often use traditional nearest neighbor or linear interpolation, failing to fully utilize the information from both observed data and numerical models, resulting in limited improvement in the accuracy and precision of the interpolation results. Secondly, this method has limitations in its weight allocation of observed data and numerical models. Conventional weight allocation methods are mainly based on the spatial distribution of observed data or the error distribution of numerical models, lacking a comprehensive consideration of the interrelationship between the two. Furthermore, this method is insufficient in handling the correlation and uncertainty between data. Ocean surface temperature (OST) data exhibits certain spatial and temporal correlations, as well as observational errors and model uncertainties. Existing methods handle these correlations and uncertainties in a relatively simple manner, failing to fully utilize relevant information for data assimilation. Summary of the Invention

[0004] To address the aforementioned issues, this invention proposes a Fengyun satellite sea surface temperature (SST) data assimilation method based on ensemble optimal interpolation, which improves the accuracy and precision of satellite SST data and enhances the assimilation effect of ocean temperature data.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] A method for assimilating Fengyun satellite sea surface temperature data based on ensemble optimal interpolation includes the following steps:

[0007] Step 1: Obtain ocean temperature observation data and numerical prediction models from multiple sources in the same region to construct a dataset;

[0008] Step 2: Calculate the correlation between each data point in the region and the ocean temperature observation data and the numerical prediction model, and assign corresponding weights to the data points based on the correlation results;

[0009] Step 3: Based on the weights, interpolate the ocean temperature observation data using the optimal interpolation method to generate interpolation results;

[0010] Step 4: Update the numerical prediction model based on the interpolation results.

[0011] In some implementations, the ocean temperature observation data includes satellite sea surface temperature data from polar-orbiting meteorological satellite systems, as well as temperature and salinity observation data from the Global Temperature and Salinity Profile Project.

[0012] In some implementations, the satellite sea surface temperature data is preprocessed before step 1.

[0013] In some implementations, the preprocessing includes:

[0014] Read the satellite sea surface temperature data output by the polar-orbiting meteorological satellite system, select the spatial point with the best quality label in the satellite sea surface temperature data as the reference point, and read the SST value of the reference point. Other spatial points that do not meet the requirements are set as invalid values. Then write the SST value of the reference point into a NETCDF format file.

[0015] Select grid points within a 10°*10° range in the file to form a candidate region, calculate the average value of all grid points in the candidate region, and remove grid points whose variance exceeds the average value by more than 2 standard deviations from the candidate region.

[0016] In some implementations, data on shallow areas with a water depth of less than 200 meters are removed from the file.

[0017] In some embodiments, the preprocessing is followed by quality control of the temperature and salinity observation data.

[0018] In some embodiments, the quality control includes:

[0019] The temperature and salinity observation data were compared with the topographic data, the ETOPO5 topography was interpolated to the location of the observation station, and locations with a water depth greater than 200 meters were removed.

[0020] It was confirmed that the measurement times for the temperature and salinity observation data were sequentially increasing.

[0021] It was confirmed that the measurement depths of the temperature and salinity observation data increased sequentially.

[0022] Temperature observations outside the 0-35℃ range and salinity observations outside the 0-40psu range were removed from the aforementioned temperature and salinity observation data.

[0023] Remove temperature and salinity elements from the temperature and salinity observation data whose error variance exceeds the threshold.

[0024] Confirm that the burrs in the temperature and salinity observation data are within the preset range.

[0025] In some implementations, the preprocessed satellite sea surface temperature data and the quality-controlled temperature and salinity observation data are fused and projected onto a global latitude and longitude grid to form visualized ocean temperature observation data.

[0026] The beneficial effects of this invention are as follows: by fusing ocean temperature observation data from multiple sources and using the correlation between ocean temperature observation data and numerical prediction models to generate weights, the optimal interpolation method is used to interpolate the ocean temperature observation data, and the data assimilation method is used to correct the prediction bias of the numerical model, resulting in a more reliable sea temperature distribution map. Attached Figure Description

[0027] Figure 1 This is a schematic diagram showing the root mean square error of the FY-3C raw sea surface temperature data and fusion product relative to the Reynolds sea surface temperature fusion product. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the content of this invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to this invention are shown in the accompanying drawings, not all of them.

[0029] This embodiment proposes a Fengyun satellite sea surface temperature data assimilation method based on ensemble optimal interpolation, including the following steps:

[0030] Step 1: Obtain ocean temperature observation data and numerical prediction models from multiple sources in the same region to construct a dataset;

[0031] In this embodiment, the ocean temperature observation data includes at least satellite sea surface temperature data from polar-orbiting meteorological satellite systems, as well as temperature and salinity observation data provided by the Global Temperature and Salinity Profile Project.

[0032] Before step 1, you can choose to fuse satellite sea surface temperature data with temperature and salinity observation data.

[0033] Before the two sets of data are fused, the quality of the satellite sea surface temperature (SST) data directly affects the fusion accuracy. Therefore, before the satellite SST data enters the fusion process, it is essential to ensure that the input data does not contain significant errors. Thus, the satellite SST data is first preprocessed.

[0034] This embodiment uses FY-3C satellite sea surface temperature data as an example. The specific preprocessing includes:

[0035] Reading satellite data and format conversion: Read the satellite sea surface temperature data output by the FY-3C polar-orbiting meteorological satellite system, select the spatial point with the best quality label in the satellite sea surface temperature data as the reference point, and read the SST value of the reference point. Other spatial points that do not meet the requirements are set as invalid values. Then write the SST value of the reference point into a NETCDF format file.

[0036] Variance check: Select grid points within a 10°*10° range in the file to form a candidate region, calculate the average value of all grid points in the candidate region, and remove grid points whose variance exceeds the average value by 2 standard deviations from the candidate region;

[0037] Quality control: Remove data from shallow areas with a water depth of less than 200 meters in the file.

[0038] The temperature and salinity observation data provided by the Global Temperature and Salinity Profiling Project (GTSPP) underwent preliminary quality control before publication. However, due to limitations such as observational meteorological conditions and instrument accuracy, the data contains errors from various sources. Therefore, preprocessing and quality control are necessary before use in the fusion of satellite sea surface temperature data, and the data is interpolated to a 2-meter depth to generate a set of in-situ observed temperature data suitable for fusion with FY-3C satellite sea surface temperature data. Thus, after preprocessing, quality control of the temperature and salinity observation data is also required.

[0039] Quality control includes:

[0040] Location check: Compare temperature and salinity observation data with topographic data, interpolate ETOPO5 topography to the location of the observation station, and remove locations with water depth greater than 200 meters;

[0041] Observation time check: Confirm that the measurement times for temperature and salinity observation data are sequentially increasing;

[0042] Observation depth check: Confirm that the measurement depths of temperature and salinity observation data increase sequentially;

[0043] Climate threshold check: Remove temperature observations outside the 0-35℃ range and salinity observations outside the 0-40psu range from the temperature and salinity observation data.

[0044] Variance check: Remove temperature and salinity elements from the temperature and salinity observation data whose error variance exceeds the threshold.

[0045] Burr inspection: Confirms that the burrs in the temperature and salinity observation data are within the preset range. Vertical

[0046] spike = abs(v2-(v3+v1) / 2)-abs(v1-v3) / 2. If spike_temp>2 or spike_salt>0.3, remove the layer. V1 represents the layer above the check layer, V2 represents the check layer, and V3 represents the layer below the check layer.

[0047] Finally, the preprocessed satellite sea surface temperature (SST) data, along with quality-controlled temperature and salinity observation data, are fused and projected onto a global isotropic grid to form visualized ocean temperature observation data. For example, surface SST data from satellite SST data and quality-controlled temperature and salinity observations are projected onto a global isotropic grid to generate a fused FY-3C satellite SST product from February 2015 to December 2016, with a temporal resolution of monthly and a spatial resolution of 0.25°. This is then compared and analyzed with the US Reynolds fused SST product (e.g.,...). Figure 1 As shown in the figure, the average error of the fused SST is 0.63℃.

[0048] Step 2: Calculate the correlation between each data point within the region and the ocean temperature observation data and the numerical prediction model, and assign corresponding weights to the data points based on the correlation results. In one example, the weight of each data point can be calculated by considering the spatial distribution of the observation data, the error distribution of the numerical model, and the interrelationships between the data.

[0049] Step 3: Based on the weights, the optimal interpolation method is used to interpolate the ocean temperature observation data, generating interpolation results. Unlike traditional nearest neighbor or linear interpolation, this method considers the weight distribution of the data, making more effective use of information from ocean temperature observation data and numerical prediction models, thus improving the accuracy and precision of the interpolation results.

[0050] Step 4: Update the numerical prediction model based on the interpolation results. By fusing information from observational data, the prediction bias of the numerical model can be corrected more accurately, resulting in a more reliable sea surface temperature distribution map.

[0051] After the above assimilation steps, the advantages of the updated numerical prediction model include:

[0052] Consistent Model Physical Processes and Resolution: The ocean model MOM4_L40 used in this invention maintains consistent model physical processes and resolution with the short-term climate prediction system of the Climate Center. This means that during the ocean data assimilation process, the model used has good matching with the short-term climate prediction system, providing a dynamically coordinated ocean initial field and effectively improving the climate forecasting level of my country's short-term climate model.

[0053] Global Data Assimilation Capability: The Fengyun satellite sea surface temperature data assimilation method based on ensemble optimal interpolation developed in this invention has global data assimilation capability. This means that this method can process satellite sea surface temperature data, satellite altimeter data, and satellite wind field data from all over the world, and is not limited to specific regions or specific data sources. This capability makes the method more flexible and adaptable, and can meet the assimilation requirements of different regions and different data needs.

[0054] Scalability as a subsystem: The Fengyun satellite sea surface temperature data assimilation method based on ensemble optimal interpolation of this invention can be integrated as a subsystem into other large-scale data assimilation systems or climate prediction systems. This means that the system has good integration capabilities and can cooperate with other systems to provide them with near real-time ocean data assimilation capabilities. This scalability enables the technology of this invention to be promoted and applied on a wider scale, improving the application capabilities of my country's Fengyun meteorological satellite data in climate models.

[0055] In summary, the Fengyun satellite sea surface temperature data assimilation method based on ensemble optimal interpolation, developed using the National Climate Center's ocean model MOM4_L40, possesses model physics processes and resolution consistent with short-term climate prediction systems. It can provide dynamically coordinated ocean initial fields, effectively improving the climate forecasting capabilities of my country's short-term climate models. This system has global data assimilation capabilities and can be integrated as a subsystem into other large-scale data assimilation systems or climate prediction systems, enhancing the application capabilities of my country's Fengyun meteorological satellite data in climate models.

[0056] The above embodiments are merely illustrative of the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. All equivalent changes or modifications made based on the essence of the content of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for assimilating Fengyun satellite sea surface temperature (SST) data based on ensemble optimal interpolation, using the National Climate Center's ocean model MOM4_L40 and FY-3C satellite SST data, characterized in that... Includes the following steps: Step 1: Obtain ocean temperature observation data and numerical prediction models from multiple sources in the same region, and construct a dataset. The ocean temperature observation data includes satellite sea surface temperature data from polar-orbiting meteorological satellite systems, as well as temperature and salinity observation data provided by the Global Temperature and Salinity Profile Project. Before step 1, the process includes preprocessing the satellite sea surface temperature (SST) data. This preprocessing includes: reading the SST data output by the polar-orbiting meteorological satellite system; selecting the spatial point with the best quality identifier from the SST data as a reference point; reading the SST value of the reference point; setting other unsatisfactory spatial points as invalid values; and then writing the SST value of the reference point into a NETCDF format file. A 10°*10° grid area within the file is selected to form a candidate region. The average value of all grid points in the candidate region is calculated, and grid points with variances exceeding two standard deviations of the average value are removed from the candidate region. Data from shallow water areas with a depth less than 200 meters in the file are removed. After the preprocessing, the process also includes quality control of the temperature and salinity observation data. This quality control includes: adjusting the temperature and salinity data... Salinity observation data is compared with topographic data. ETOPO5 topographic data is interpolated to the location of the observation station, and locations with a water depth greater than 200 meters are removed. It is confirmed that the measurement time and depth of the temperature and salinity observation data are sequentially increasing. Temperature observations outside the 0-35℃ range and salinity observations outside the 0-40 psu range are removed. Temperature and salinity elements with variances exceeding a threshold in each layer are removed. It is confirmed that the spikes in the temperature and salinity observation data are within a preset range. The preprocessed satellite sea surface temperature data and the quality-controlled temperature and salinity observation data are fused and projected onto a global latitude and longitude grid to form visualized ocean temperature observation data. Step 2: Calculate the correlation between each data point in the region and the ocean temperature observation data and the numerical prediction model, and assign corresponding weights to the data points based on the correlation results; Step 3: Based on the weights, interpolate the ocean temperature observation data using the optimal interpolation method to generate interpolation results; Step 4: Update the numerical prediction model based on the interpolation results.

2. The Fengyun satellite sea surface temperature data assimilation method based on ensemble optimal interpolation as described in claim 1, characterized in that, The preprocessed satellite sea surface temperature data and the quality-controlled temperature and salinity observation data are fused together and projected onto a global latitude and longitude grid to form visualized ocean temperature observation data.

Images