A method for estimating sea surface temperature
By matching the space-time with the second satellite equipped with 6.9GHz channel, using neural networks and regression models to simulate the bright temperature value of the 6.9GHz channel, the ocean surface temperature inversion accuracy on Fengyun-3 satellite is improved, and the lack of microwave imager is solved. Problems with temperature-sensitive channels for ocean surfaces.
Patent Information
- Application Number
- CN202110684135.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-06-21
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2041-06-21
AI Technical Summary
The microwave imager on the Fengyun-3 satellite lacks a 6.9GHz channel that is sensitive to ocean surface temperature, resulting in limited accuracy of ocean surface temperature estimation.
By matching time and space with a second satellite (such as Aqua satellite) equipped with a 6.9GHz channel, a neural network is used to establish a calculation relationship of bright temperature observations, simulate the bright temperature value of the 6.9GHz channel, and use a regression model to perform real-time ocean surface temperature. Inversion.
The accuracy of the microwave imager's ocean surface temperature inversion is improved, and the problem of lack of sensitive channels to ocean surface temperature is solved.
Smart Images

Figure CN113255232B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of remote sensing technology, and in particular to a method for estimating ocean surface temperature. Background Art
[0002] The MicroWave Radiometer Imager (MWRI) is an important remote sensing instrument carried on the Fengyun-3 (FY-3) polar-orbiting series satellites. The instrument is a 10-channel dual-polarization microwave imaging instrument with central observation frequencies set to 10.65, 18.7, 23.8, 36.5 and 89.0 GHz. Each frequency point has two detection channels, vertical (V) and horizontal (H). The brightness temperature data obtained by the instrument can be used to quantitatively obtain geophysical parameter information such as atmospheric precipitation, water vapor, sea surface wind speed, sea temperature, sea ice distribution, soil moisture and land surface temperature.
[0003] The present inventors' "FY-3C Microwave Imager Sea Surface Temperature Product Algorithm and Precision Verification", Journal of Remote Sensing, 2018, Vol. 22, Issue (5): 1993-2002, DOI: 10.11834 / jrs.20187217, discloses an algorithm for inverting sea surface temperature (SST) using China's FY-3C microwave imager. The FY-3C MWRI SST product uses a statistical algorithm. First, the pixels containing precipitation and sea ice are removed using the MWRI precipitation and sea ice products. Then, the FY-3C MWRI observed brightness temperatures at a time interval of 0.2h and 0.2° and 100km away from the coast are selected to match the buoy observations. Then, the world is divided into four latitude bands in space and 12 months in time, and divided into ascending and descending orbits. The statistical relationship between the buoy sea temperature observation results and the MWRI brightness temperature is established to achieve the estimation of SST. However, the article does not mention the processing of the 6.9GHz channel because the FY-3C satellite does not have this channel. Summary of the invention
[0004] The technical problem to be solved by the present application is to provide a method for estimating ocean surface temperature.
[0005] In order to achieve the above technical objectives, this application adopts the following technical solutions:
[0006] A method for estimating sea surface temperature, comprising:
[0007] S101: Performing spatial and temporal matching of the first satellite and the second satellite to obtain matching sample points;
[0008] S102: using the brightness temperature observation values of the first satellite in the horizontal and vertical polarization modes of each channel and the brightness temperature observation values of the second satellite in the horizontal and vertical polarization modes of the preset channel at the matching sample point, a calculation relationship for obtaining brightness temperature simulation values of the first satellite in the horizontal and vertical polarization modes of the preset channel is established;
[0009] S103: Calculate the regression coefficient of the inversion model of the sea surface temperature of the first satellite;
[0010] S104: Based on the inversion model, the ocean surface temperature of the first satellite is inverted in real time using the brightness temperature observation values of the horizontal and vertical polarization modes of each channel of the first satellite and the calculation relationship to obtain an inversion value of the ocean surface temperature.
[0011] Preferably, in step S101, an interval distance is preset in space and an interval time is preset in time to obtain matching sample points after the first satellite and the second satellite are matched in time and space.
[0012] Preferably, in step S102, the brightness temperature observation values of the first satellite in the horizontal and vertical polarization modes of each frequency channel at the matching sample point and the brightness temperature observation values of the second satellite in the horizontal and vertical polarization modes of the preset frequency channel at the matching sample point are input, and a calculation relationship is established using a neural network calculation method.
[0013] Preferably, in step S104, the real-time inversion includes:
[0014] The inversion value of the ocean surface temperature is calculated in real time based on the inversion model according to the brightness temperature observation values of the first satellite in the horizontal and vertical polarization modes in multiple frequency channels and the brightness temperature simulation value of the first satellite in the preset frequency channel obtained according to the calculation relationship.
[0015] Preferably, in step S103, the regression coefficient of the inversion model is calculated based on the following formula:
[0016] in
[0017] t i =T Bi -150
[0018] (Brightness temperature for vertical or horizontal polarization of 6.9, 10.65, 18.7, 36.5 GHz channels)
[0019] t i =-ln(290-T Bi )
[0020] (Brightness temperature for vertical or horizontal polarization of 23.8 GHz channel)
[0021] Where SST0 is the sea surface temperature obtained from buoy observations, i represents the i-th frequency channel, T Bi is the brightness temperature observation value of the first satellite in the horizontal and vertical polarization modes of each frequency channel or the brightness temperature simulation value in the preset frequency channel, a0, a i , b i is the regression coefficient.
[0022] Preferably, based on the calculation relationship in step S102, the brightness temperature simulation value is calculated in real time and used as the brightness temperature value of the first satellite in the preset frequency channel, and the brightness temperature observation values of the first satellite in the horizontal and vertical polarization modes of each frequency channel are substituted into the regression model to calculate the regression coefficient.
[0023] Preferably, step S104 is repeated to directly use the inversion model to calculate the inversion value of the ocean surface temperature in real time.
[0024] Preferably, in step S104, the inversion model is as follows:
[0025] in
[0026] t i =T Bi -150
[0027] (Brightness temperature for vertical or horizontal polarization of 6.9, 10.65, 18.7, 36.5 GHz channels)
[0028] t i =-ln(290-T Bi )
[0029] (Brightness temperature for vertical or horizontal polarization of 23.8 GHz channel)
[0030] In the above formula, SST is the real-time calculated sea surface temperature, i represents the frequency channel, and T Bi is the brightness temperature observation value of the first satellite in the horizontal and vertical polarization modes of channel i or the brightness temperature simulation value of the first satellite in the horizontal and vertical polarization modes of the preset channel, a0, a i , b i is the regression coefficient calculated in step 103.
[0031] Preferably, the preset frequency channel is a channel that is sensitive to the sea surface temperature.
[0032] The present application has the following technical effects: utilizing the brightness temperature observation value of the second satellite at a preset channel of 6.9 GHz, obtaining the brightness temperature simulation value of the first satellite at the preset channel through a simulation method, and using it for real-time calculation of the MWRI SST inversion value, thereby improving the MWRI SST inversion accuracy and better solving the problem that the MWRI carried by the first satellite lacks a channel sensitive to SST. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] Figure 1 The present invention is a flowchart of a method for estimating sea surface temperature according to some embodiments of the present application. DETAILED DESCRIPTION
[0034] The technical content of this application is described in detail below with reference to the accompanying drawings and specific embodiments.
[0035] Since the MWRI carried by the Fengyun-3 satellite lacks the 6.9GHz channel that is sensitive to SST, the AMSR-2 carried by the US Aqua satellite has a total of 7 frequency channels, namely 6.9GHz, 7.3GHz, 10.65GHz, 18.7GHz, 23.8GHz, 36.5GHz and 89GHz. Therefore, the present invention matches the Fengyun-3 satellite with the Aqua satellite in time and space, adopts a neural network calculation method, and uses the brightness temperature values of each channel of the matched Fengyun-3 satellite's MWRI to simulate the brightness temperature value of the 6.9GHz channel. By introducing the simulated brightness temperature value of the 6.9GHz channel and using a regression model, the inversion accuracy of the SST of the Fengyun-3 satellite MWRI is improved. This application takes the Fengyun-3 satellite as an example to introduce the SST inversion method of the FY3C MWRI. However, the present application is not limited thereto, and the method is also applicable to satellites such as FY3B, FY3D and FY3F.
[0036] Figure 1 Flow chart of the method for estimating sea surface temperature of this embodiment, which specifically includes the following steps.
[0037] S101, performing time and space matching of the first satellite and the second satellite
[0038] Here, firstly, the interval distance is preset in space, and the interval time is preset in time, and the matching sample points after space-time matching are obtained. For example, the preset distance (spatial resolution) is 5 km, 10 km, 20 km, etc., and the above preset time is every 10 minutes, or every 12 minutes, etc. The brightness temperature values of the first satellite (in this embodiment, the Fengyun-3 satellite) in the horizontal and vertical polarization modes of the 10.65 GHz, 18.7 GHz, 23.8 GHz, and 36.5 GHz channels at the matching sample points, and the brightness temperature values of the second satellite (such as the American Aqua satellite) at the preset frequency channels at the matching sample points are obtained.
[0039] Time-space matching means that the time interval and distance interval between two satellites observing the same location are limited to a certain range.
[0040] In step S101, the aforementioned preset frequency channel is the 6.9 GHz channel. This is because this channel is not available in the MWRI carried by the first satellite (Fengyun-3), so it needs to be obtained through the advanced microwave scanning radiometer carried by the second satellite (Aqua satellite of the United States).
[0041] S102, using the brightness temperature observation values of the first satellite in the plurality of frequency channels and the brightness temperature observation values of the second satellite in the preset frequency channel of the matching sample point, a calculation relationship is established for obtaining the brightness temperature simulation values of the first satellite in the horizontal and vertical polarization modes in the preset frequency channel
[0042] Input the brightness temperature observation values of the first satellite at the matching sample point in the horizontal and vertical polarization modes of multiple frequency channels 10.65GHz, 18.7GHz, 23.8GHz, and 36.5GHz, and the brightness temperature observation values of the second satellite (such as the US Aqua satellite) at the matching sample point in the horizontal and vertical polarization modes of the preset frequency channel (6.9GHz), and use the neural network calculation method to establish a calculation relationship for obtaining the brightness temperature simulation values of the first satellite in the horizontal and vertical polarization modes of the 6.9GHz channel.
[0043] By using this calculation relationship, based on the brightness temperature observation values of the first satellite in each channel (excluding the 6.9 GHz channel), the brightness temperature simulation values of the first satellite in the horizontal and vertical polarization modes in the 6.9 GHz channel can be obtained. In this way, the brightness temperature values of the first satellite in all channels (including the 6.9 GHz channel) can be obtained, of which the brightness temperature simulation values in the 6.9 GHz channel are brightness temperature simulation values, and the brightness temperature observation values in the other channels are brightness temperature observation values.
[0044] S103, calculating the regression coefficient of the inversion model of the sea surface temperature of the first satellite
[0045] Based on the brightness temperature values of the first satellite in the horizontal and vertical polarization modes of all channels obtained in the previous step, they are matched with the buoy observation values in time and space. The spatial distance is 5 km, 10 km, 20 km, etc., and the time interval is every 10 minutes, or every 12 minutes, etc., to obtain the brightness temperature values and buoy observation values of each matching sample point, and then calculate the regression coefficient in the following formula:
[0046] in
[0047] t i =T Bi -150
[0048] (Brightness temperature for vertical or horizontal polarization of 6.9, 10.65, 18.7, 36.5 GHz channels)
[0049] t i =-ln(290-T Bi )
[0050] (Brightness temperature for vertical or horizontal polarization of 23.8 GHz channel)
[0051] In the above formula, SST0 is the sea surface temperature obtained from buoy observations, i represents the frequency channel, and T Bi is the brightness temperature of the first satellite in the horizontal and vertical polarization modes of channel i (i.e., the brightness temperature of the first satellite in all channels obtained in step 102), a0, a i , b i is the regression coefficient.
[0052] Using the calculation relationship obtained in S102, based on the brightness temperature observation values of the first satellite in the horizontal and vertical polarization modes in each channel (10.65GHz, 18.7GHz, 23.8GHz, 36.5GHz), the brightness temperature simulation value is obtained as the brightness temperature simulation value of the first satellite in the horizontal and vertical polarization modes at 6.9GHz. Substituting the brightness temperature observation values of the first satellite in each channel (10.65GHz, 18.7GHz, 23.8GHz, 36.5GHz) and the brightness temperature simulation value in the 6.9GHz channel into the above formula, based on the SST0 value obtained according to the buoy observation value, the regression coefficients a0 and a i , b i After the regression coefficient is determined, the regression inversion model is established.
[0053] S104, perform real-time inversion of the first satellite SST
[0054] According to the brightness temperature observation values of 10.65, 18.7, 23.8, 36.5 GHz horizontal and vertical polarization modes observed by the microwave imager of the first satellite, and the brightness temperature simulation values of the first satellite at 6.9 GHz horizontal and vertical polarization modes obtained according to the calculation relationship in step S102, the SST value is calculated in real time based on the following formula, so as to perform real-time inversion of the SST value of the first satellite. At this time, because the regression coefficient has been determined, the following formula can be directly used for real-time calculation to obtain the SST inversion value.
[0055] in
[0056] t i =T Bi -150
[0057] (Brightness temperature for vertical or horizontal polarization of 6.9, 10.65, 18.7, 36.5 GHz channels)
[0058] t i =-ln(290-T Bi )
[0059] (Brightness temperature for vertical or horizontal polarization of 23.8 GHz channel)
[0060] In the above formula, SST is the real-time calculated sea surface temperature, i represents the frequency channel, and T Bi is the brightness temperature observation value of the first satellite in the horizontal and vertical polarization modes of channel i (for 10.65 GHz, 18.7 GHz, 23.8 GHz, and 36.5 GHz channels) or the brightness temperature simulation value of the first satellite in the horizontal and vertical polarization modes of the preset channel (for the 6.9 GHz channel), a0, a i , b i is the regression coefficient calculated in step 103. The brightness temperature simulation values of the first satellite in the horizontal and vertical polarization modes of the 6.9 GHz channel are obtained in real time using the calculation relationship determined in step S102.
[0061] The above method solves the problem that the first satellite does not have a preset frequency channel, such as a 6.9 GHz channel sensitive to SST. The brightness temperature value observed by the second satellite with the preset frequency channel is used to improve the SST inversion accuracy of the first satellite.
[0062] It should be noted that the present invention is not limited to using the above regression model to calculate SST, and other calculation models may also be used; moreover, the present invention may also be applicable to various methods of inverting and calculating SST using brightness temperature values. These changes are all within the scope of protection of the present invention.
[0063] The present application is described in detail above. For a person skilled in the art, any obvious modification made to the present application without departing from the essence of the present application will constitute an infringement of the patent right of the present application and the person shall bear the corresponding legal liability.
Claims
1. A method for estimating sea surface temperature, characterized in that: include: S101: Performing spatial and temporal matching of the first satellite and the second satellite to obtain matching sample points; S102: using the brightness temperature observation values of the horizontal and vertical polarization modes of each frequency channel of the first satellite at the matching sample point and the brightness temperature observation values of the horizontal and vertical polarization modes of the second satellite at the preset frequency channel, a calculation relationship for obtaining brightness temperature simulation values of the first satellite at the horizontal and vertical polarization modes of the preset frequency channel is established; S103: Calculating the regression coefficient of the inversion model of the sea surface temperature of the first satellite; S104: Based on the inversion model, using the brightness temperature observation values of the horizontal and vertical polarization modes of each frequency channel of the first satellite and the calculation relationship, perform real-time inversion of the ocean surface temperature of the first satellite to obtain an inversion value of the ocean surface temperature. Wherein, in the step S102, the brightness temperature observation values of the first satellite in the horizontal and vertical polarization modes of the respective frequency channels at the matching sample point and the brightness temperature observation values of the second satellite in the horizontal and vertical polarization modes of the preset frequency channel at the matching sample point are input, and the calculation relationship is established using a calculation method of a neural network. The preset frequency channel is a channel that is sensitive to the ocean surface temperature.
2. The method for estimating sea surface temperature according to claim 1, wherein: In the step S102, the brightness temperature observation values of the first satellite at the matching sample point in the horizontal and vertical polarization modes of multiple frequency channels, and the brightness temperature observation values of the second satellite at the matching sample point in the horizontal and vertical polarization modes of the preset frequency channel are input; By using a neural network calculation method, brightness temperature simulation values of the first satellite in horizontal and vertical polarization modes in a preset frequency channel are obtained.
3. The method for estimating sea surface temperature according to claim 2, wherein: In the step S101, an interval distance is preset in space and an interval time is preset in time to obtain matching sample points after the first satellite and the second satellite perform the space-time matching.
4. The method for estimating sea surface temperature according to claim 2, wherein: In step S104, the real-time inversion includes: The inversion value of the ocean surface temperature is calculated in real time based on the inversion model according to the brightness temperature observation values in the horizontal and vertical polarization modes of the multiple frequency channels observed by the first satellite and the brightness temperature simulation value of the first satellite in the preset frequency channel obtained according to the calculation relationship.
5. The method for estimating sea surface temperature according to claim 4, characterized in that: In step S103, the regression coefficient of the inversion model is calculated based on the following formula: ,in Indicates the brightness temperature value of vertical or horizontal polarization mode for 6.9, 10.65, 18.7, and 36.5 GHz channels; Indicates the brightness temperature value of the vertical or horizontal polarization mode for the 23.8GHz channel; Where SST0 is the sea surface temperature obtained from buoy observations, i represents the i-th frequency channel, T Bi is the brightness temperature observation value of the first satellite in the horizontal and vertical polarization modes of the respective frequency channels or the brightness temperature simulation value in the preset frequency channels, a0, a i 、b i is the regression coefficient.
6. The method for estimating sea surface temperature according to claim 4, characterized in that: Based on the calculation relationship in step S102, the brightness temperature simulation value is calculated in real time and used as the brightness temperature value of the first satellite in the preset frequency channel, and the brightness temperature observation values of the first satellite in the horizontal and vertical polarization modes of each frequency channel are substituted into the inversion model to calculate the regression coefficient.
7. The method for estimating sea surface temperature according to claim 4, wherein: Repeat step S104 to directly use the inversion model to calculate the inversion value of the ocean surface temperature in real time.
8. The method for estimating sea surface temperature according to claim 7, wherein: In the step S104, the inversion model is as follows: ,in Indicates the brightness temperature value of vertical or horizontal polarization mode for 6.9, 10.65, 18.7, and 36.5 GHz channels; Indicates the brightness temperature value of the vertical or horizontal polarization mode for the 23.8GHz channel; In the above formula, SST is the real-time calculated sea surface temperature, i represents the frequency channel, and T Bi is the brightness temperature observation value of the first satellite in the horizontal and vertical polarization modes of channel i or the brightness temperature simulation value of the first satellite in the horizontal and vertical polarization modes of the preset frequency channel, a0, a i 、b i is the regression coefficient calculated in step 103.
