A method, system, device and medium for generating a chla concentration profile

Abstract

This invention provides a method, system, device, and medium for generating Chla concentration distribution. The method includes: acquiring remote sensing observation data and field observation data of Chla concentration; generating a model dataset based on the remote sensing observation data and field observation data, wherein the remote sensing observation data includes CALIOP data and MODISChla data; establishing and training a Chla inversion model based on the model dataset; and inputting the CALIOP data into the Chla inversion model to obtain the Chla concentration distribution. This invention can generate the spatiotemporal distribution of Chla concentration based on CALIOP data.

Description

Technical Field

[0001] This invention relates to the field of remote sensing observation technology, and in particular to a method, system, device and medium for generating Chla concentration distribution. Background Technology

[0002] Currently, research on Arctic Ocean phytoplankton (chlorophyll a concentration) mainly employs three technical methods: on-site observation, satellite water color remote sensing, and satellite lidar remote sensing.

[0003] Field observations primarily include ship-based and buoy-based observations. Long-term ship-based observations of the Barents Sea in the Arctic Ocean revealed a significant increase in chlorophyll a concentration from 2010 to 2020 due to rising sea temperatures. In recent years, with technological advancements, Argo buoy observations have gradually expanded to higher latitude Arctic Ocean areas (such as Baffin Bay and the Greenland Sea). Studies based on Argo observation data have found that even under the ice sheet in winter, Arctic phytoplankton continue to grow and reproduce, indicating their adaptability to extremely low light conditions. Satellite ocean color remote sensing, with its advantages of large-scale, long-term, and rapid repeatable observations, has become an important technical means for studying the Arctic Ocean's marine ecological environment. Research has found a significant increase in Arctic chlorophyll a concentration, while simultaneously, the Arctic Ocean is becoming increasingly "Atlanticized." These changes have attracted widespread attention due to their profound ecological impacts. Compared to ocean color remote sensing, the development of lidar ocean remote sensing is relatively lagging, generally still in the system development and methodology research stage. Although there are no dedicated lidar satellites for ocean exploration internationally, lidar ocean remote sensing technology has developed rapidly in recent years, significantly expanding our scientific understanding of the ocean. In 2006 and 2018, the spaceborne lidar systems CALIOP and ICESat-2, used for aerosol / cloud observation and sea ice detection respectively, were put into orbit, promoting the development of spaceborne lidar ocean exploration technology. Currently, based on spaceborne lidar data, methods for measuring ocean particulate backscattering coefficient (bbp) and phytoplankton carbon content (C) have been developed. phyto Remote sensing inversion algorithms for parameters such as particulate organic carbon content and chlorophyll a concentration.

[0004] Existing technologies have the following drawbacks: First, in-situ observations mainly include ship-based and buoy-based observations, which are limited by the unique geographical location and climate conditions of the Arctic Ocean. Ship-based observation data is mostly concentrated from July to September each year, with observations during other periods, especially during the polar night, being relatively scarce. Argo buoy observations are concentrated only in areas such as Baffin Bay and the Greenland Sea, resulting in uneven spatial distribution. Second, satellite ocean color remote sensing uses the sun as a light source, making observations impossible during the 3-6 month polar night in the Arctic Ocean. Therefore, it is impossible to obtain complete seasonal succession information for phytoplankton, and there may be some uncertainty in characterizing long-term evolution. Third, related research has preliminarily demonstrated the technical feasibility of using spaceborne lidar to retrieve ocean parameters. For example, research has developed a CALIOP chlorophyll a concentration neural network inversion algorithm for global oceans. However, this algorithm is mainly applicable to open sea areas in mid- and low latitudes, and has significant uncertainties in the optically complex Arctic Ocean. Furthermore, in high-latitude sea areas (Arctic Ocean), there is an implicit and complex nonlinear relationship between lidar echo signals and chlorophyll a concentration, which cannot be accurately characterized by a simple linear model alone. Summary of the Invention

[0005] The purpose of this invention is to propose a method, system, device and medium for generating Chla concentration distribution, so as to generate the spatiotemporal distribution of Chla concentration based on CALIOP data.

[0006] To achieve the above objectives, in a first aspect, embodiments of the present invention provide a method for generating Chla concentration distribution, the method comprising:

[0007] Acquire remote sensing and field observation data of Chla concentration, and generate a model dataset based on the remote sensing and field observation data, wherein the remote sensing data includes CALIOP data and MODISChla data;

[0008] A Chla inversion model is established and trained based on the model dataset. The CALIOP data is then input into the Chla inversion model to obtain the Chla concentration distribution.

[0009] Furthermore, the step of generating the model dataset based on the remote sensing observation data and the field observation data includes:

[0010] The remote sensing observation data and the field observation data are divided according to the preset spatial distribution and time period;

[0011] The CALIOP data, which are in the same spatial distribution and time period, are matched with the MODISChla data and the field observation data to generate a model dataset.

[0012] Furthermore, the acquisition of remote sensing observation data on Chla concentration includes:

[0013] The CALIOP data is transiently corrected using a deconvolution algorithm.

[0014] Furthermore, the transient correction of the CALIOP data using the deconvolution algorithm includes:

[0015] The attenuated backscattered signal in the CALIOP data is deconvolved with the transient response function.

[0016] Further, the step of establishing and training the Chla inversion model based on the model dataset includes:

[0017] The forward propagation algorithm is used to optimize the model weights and biases of the Chla inversion model.

[0018] Furthermore, the acquisition of remote sensing observation data on Chla concentration includes:

[0019] The total depolarization ratio of the ocean and the backscattering from the ocean subsurface lidar were calculated based on the transiently corrected CALIOP data as remote sensing observation data.

[0020] Furthermore, the matching of the CALIOP data, which are spatially distributed and within the same time period, with the MODISChla data and the field observation data, respectively, includes:

[0021] If the MODISChla data and field observation data exist in the same spatial distribution and time period, only the field observation data will be retained.

[0022] Secondly, embodiments of the present invention provide a Chla concentration distribution generation system, the system comprising:

[0023] The model dataset generation module is used to acquire remote sensing observation data and field observation data of Chla concentration, and generate a model dataset based on the remote sensing observation data and field observation data. The remote sensing observation data includes CALIOP data and MODISChla data.

[0024] The Chla concentration distribution generation module is used to establish and train a Chla inversion model based on the model dataset, and input the CALIOP data into the Chla inversion model to obtain the Chla concentration distribution.

[0025] Thirdly, embodiments of the present invention also provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0026] Fourthly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.

[0027] This invention provides a method, system, device, and medium for generating Chla concentration distribution. The method includes: acquiring remote sensing observation data and field observation data of Chla concentration; generating a model dataset based on the remote sensing observation data and field observation data, wherein the remote sensing observation data includes CALIOP data and MODISChla data; establishing and training a Chla inversion model based on the model dataset; and inputting the CALIOP data into the Chla inversion model to obtain the Chla concentration distribution. This invention can generate the spatiotemporal distribution of Chla concentration based on CALIOP data. Attached Figure Description

[0028] Figure 1 This is a schematic flowchart of a method for generating Chla concentration distribution according to an embodiment of the present invention;

[0029] Figure 2 This is a flowchart illustrating the construction process of the star and star-ground spatiotemporal big data set provided in this embodiment of the invention;

[0030] Figure 3 This is a schematic diagram of the spatiotemporal distribution of the Arctic Ocean Chla seasonal cycle provided in an embodiment of the present invention;

[0031] Figure 4 This is a schematic diagram illustrating the model accuracy evaluation of the Chla inversion model provided in this embodiment of the invention;

[0032] Figure 5 This is a schematic diagram illustrating the accuracy verification of the CALIOPChla product based on an independent verification dataset, provided in an embodiment of the present invention.

[0033] Figure 6 This is a system block diagram of a Chla concentration distribution generation system provided in an embodiment of the present invention;

[0034] Figure 7 This is an internal structural diagram of the computer device in an embodiment of the present invention. Detailed Implementation

[0035] To make the objectives, technical solutions, and beneficial effects of this application clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Obviously, the embodiments described below are only part of the embodiments of the present invention and are used to illustrate the present invention, but are not intended to limit the scope of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0036] In one embodiment, such as Figure 1 As shown, a method for generating Chla concentration distribution is provided, the method comprising:

[0037] S11. Obtain remote sensing observation data and field observation data of Chla concentration, and generate a model dataset based on the remote sensing observation data and field observation data. The remote sensing observation data includes CALIOP data and MODISChla data.

[0038] The flowchart for constructing the star and star-ground spatiotemporal big data set in this embodiment is as follows: Figure 2 As shown. Specifically, Arctic Ocean BGC-Argo buoy observation data (2010-2021, N approximately 6000) were collected as measured data, obtained from http: / / www.coriolis.eu.org / Data-Products / DataDelivery / Data-selection. To obtain accurate Arctic Ocean surface Chla measured data, the obtained BGC-Argo data were processed, with the main steps being:

[0039] Median filtering is used to smooth the data in order to remove the original Chla noise and peaks;

[0040] The system deviation caused by the sensor being mounted on the buoy is corrected using OffsetCorrection.

[0041] The reduction in Chla fluorescence signal was corrected using Non-Photochemical Quenching (NPQ) Correction.

[0042] Collect remote sensing data. Download CALIOP Level 1B Version 4.1 data from 2007 to 2021 and MODIS Chla Level-39km monthly average data products acquired by the Aqua satellite, obtained from https: / / subset.larc.nasa.gov / calipso / login.php and https: / / oceancolor.gsfc.nasa.gov, respectively. CALIPSO, as part of the A-Train satellite constellation, orbits the Earth every 96 minutes with a revisit period of 16 days. The time interval between CALIPSO and Aqua satellites passing the same location is approximately 1-2 minutes; therefore, they can be considered to be observing the Earth synchronously. CALIOP is the main instrument on CALIPSO, with a horizontal resolution of 1 / 3 km and a vertical resolution of approximately 22.5 meters in water. This instrument can emit laser pulses of two wavelengths: 532 nm and 1064 nm. The 532 nm channel is further divided into a vertical channel and a horizontal channel.

[0043] Collect auxiliary data. From CopernicusMarineEnvironmentMonitoringService(CMEMS: http: / / marine.copernicus.eu / Monthly average data were obtained for Arctic Ocean sea surface temperature (SST) (2006-2021), mixed layer depth (MLD) (2006-2021), and nitrate concentration (NO3) (2007-2020, reanalysis data); data were obtained from the National Snow and Ice Data Center. https: / / nsidc.org / data / g02202 / Daily sea ice concentration (SIC) data for the Arctic Ocean (2006–2021) were obtained; data from the Arctic Great Rivers ( https: / / arcticgreatrivers.org / Daily runoff data (2006-2021) for the six major Arctic rivers (Mackenzie, Yukon, Kolyma, Lena, Yenisey, and Ob) were obtained. A schematic diagram of the spatiotemporal distribution of the Arctic Ocean Chla seasonal cycle in this embodiment is shown below. Figure 3 As shown.

[0044] To obtain the most accurate underwater signals possible, this embodiment performs quality control on CALIOP data according to screening principles. These principles include:

[0045] 1) Extract profiles with peak values ​​within ±120m (±4bin) of the “Surface_Elevation” parameter to identify the location of sea surface signals;

[0046] 2) Remove saturated sea surface backscattered signals (ieSurface_Saturation_Flag_532=0) to reduce scattered signal errors;

[0047] 3) Extracted attenuated backscattering coefficient integral (IAB) < 0.17sr -1 To screen for profiles of clean air;

[0048] 4) Calculate the average value (approximately 5 km) of 15 consecutive laser pulses at the same tilt angle (Off_Nadir_Angle) to reduce random errors.

[0049] 5) Using δ T A conservative threshold of ≤0.05 was used to eliminate the impact of sea ice cover.

[0050] 6) Extract wind speed 3ms -1 ≤w<8ms -1Profile data within the range is used to avoid signal contamination from bubbles, foam, and white waves, as well as potential errors caused by strong specular reflections from the ocean surface.

[0051] Secondly, this embodiment also performs transient correction on the CALIOP data using a deconvolution algorithm. Specifically, the attenuated backscattered signal in the CALIOP data is deconvolved with the transient response function. After transient correction, the total depolarization ratio of the ocean and the backscattered signal from the ocean subsurface lidar are calculated based on the transiently corrected CALIOP data as remote sensing observation data. The formula for calculating the transient response function is as follows:

[0052]

[0053] The 12 bins represent the distance from the bin before the surface echo peak to the 10th bin after the peak. i With Z j Let β(Z) represent the heights at the i-th and j-th bins, respectively. i ) and β(Z j ) represent the attenuated backscattering at the i-th and j-th bins, respectively. i and j represent the positions corresponding to each bin, and p is the location of the land surface signal, i.e., the bin position of the peak signal.

[0054] The convolution of the target's reflected signal and the response function yields the output signal. Similarly, the attenuated ocean backscattered signal β′ observed by CALIOP... m (Z) is the correct attenuation of the backscattered signal β′. c The result of the convolution between (Z) and the transient response function F(Z). The specific calculation formula is as follows:

[0055] β′ m (Z)=F(Z)*β′ c (Z)

[0056] Further expressed as a matrix equation:

[0057]

[0058] Based on the above solution F(Z) and the known β′ m (Z) The CALIOP is transiently corrected by deconvolution, and the correct CALIOP attenuated backscattered signal is obtained. The deconvolution formula is as follows:

[0059] β′ c (Z)=[F(Z)] -1 *β′ m (Z)

[0060] Finally, ocean parameters, namely the total depolarization ratio (δ), are calculated based on the correct CALIOP attenuated backscattered signal. T The calculation formula for backscattering (γ) from marine subsurface lidar is as follows:

[0061]

[0062]

[0063] Where p is the location of the signal on the ocean surface, i.e., the bin position of the peak signal; θ represents the incident angle of the CALIOP lidar system, which was adjusted from 0.3° to 3° after November 28, 2007. β′ cr_corr (Z i ) represents the attenuated backscattering of the corrected vertical channel; β′ co_corr (Z i ) represents the attenuated backscattering of the parallel channel after correction; σ 2 δ represents the wave slope variance, which is a function of the wind speed data. w The subsurface deflection ratio is defined as the ratio of the integrals of the vertical channel and the horizontal channel below sea level, typically δ. w Take 0.1.

[0064] Due to the transient response of the CALIOP photomultiplier tube (PMT), the attenuated backscattered signal is affected by the strong signal reflected from the ocean surface. Therefore, this embodiment uses transient correction, selecting a hard surface as a good target for the CALIOP transient response function, thereby effectively removing the influence of the non-ideal recovery of the PMT and improving the accuracy of the data.

[0065] In this embodiment, the process of generating the model dataset includes: dividing the remote sensing observation data and field observation data according to a preset spatial distribution and time period; matching the CALIOP data within the same spatial distribution and time period with the MODISChla data and the field observation data respectively to generate the model dataset. If both MODISChla data and field observation data exist within the same spatial distribution and time period, only the field observation data is retained. Specifically, outliers in MODIS and measured data are removed based on the 3sigma principle; CALIOP, MODIS, and measured data are gridded to ensure better consistency among the three. If each type of data contains multiple valid values ​​within the same window, such as a 2°×2° spatial window and a monthly time window, their average is taken; otherwise, a null value is set. Based on time and latitude / longitude, the gridded CALIOP data is spatiotemporally matched with MODIS and measured data according to a 2°×2° spatial window and a monthly time window respectively. It should be noted that if valid values ​​exist in both MODIS and measured data at the same time location, the measured data is taken as the true value of the model, accounting for approximately 5% of all measured data. The spatiotemporal matching dataset is divided into a model dataset and an independent validation dataset. The model dataset is randomly divided into two independent subsets: approximately 90% for the training dataset and approximately 10% for the test dataset. The independent validation dataset includes MODIS validation data and measured validation data. Compared to MODISChla data, field observation data is more accurate due to actual observations. Therefore, in this embodiment, when both exist in the same spatiotemporal location, only the field observation data is retained, which allows the generated dataset to better reflect the actual chlorophyll a distribution.

[0066] S12. Establish and train the Chla inversion model based on the model dataset, and input the CALIOP data into the Chla inversion model to obtain the Chla concentration distribution.

[0067] In this embodiment, a BPNN-Chla inversion model is constructed using a neural network approach, comprising one input layer, six hidden layers, and one output layer (Chla). The model employs the Levenberg-Marquardt backpropagation algorithm as the training function to optimize model weights and biases, and selects Mean Square Error (MSE) as the loss function. Tanh (Tansig) is used as the activation function between the model input and the hidden layers, and Purelin is selected as the transfer function for the output layer. Input parameters include: CALIOP parameters (CALIOP vertical attenuation backscattering coefficients); time parameters (year and month); and spatial parameters (longitude and latitude). In the actual training process, because the forward propagation algorithm does not require storing all intermediate states during backpropagation, this embodiment uses the forward propagation algorithm as the training function to effectively improve the generation efficiency of Chla concentration distribution by optimizing model weights and biases.

[0068] A BPNN-Chla model was established based on CALIOP satellite data, obtaining a complete and seamless long-term (2007-2021) CALIOPChla data product for the Arctic Ocean. The study found that the loss curve of the BPNN-Chla model gradually decreases, and both training and testing errors follow a normal distribution, indicating that the model has good fitting and generalization capabilities.

[0069] Table 1. Accuracy Evaluation of BPNN-Chla Model

[0070]

[0071]

[0072] The model accuracy evaluation of the Chla inversion model of this invention is shown in Table 1 and Figure 4 As shown, the vast majority of points are distributed along the 1:1 line, N Train =19031, R Train =0.86, MAPD Train =26%; N Test =2115, R Test =0.87, MAPD Test =27%, with slightly lower inversion accuracy for spring data (MAPD) Train =32% and MAPD Test =34%) in the fall (MAPD) Train =22% and MAPD Test =23%). The model inversion accuracy remains basically consistent across different latitude ranges. That is, the Chla inversion model constructed in this invention has high Chla inversion accuracy.

[0073] Secondly, the accuracy of the CALIOPChla product was verified based on independent validation datasets (MODIS validation and field testing). Figure 5 As shown in the table, the gray area represents the number of samples included in MAPD < 35%.

[0074] Table 2. Independent Validation of CALIOPChla (2007-2009 & 2020-2021)

[0075]

[0076] Both validation results showed good accuracy (MAPD). MODIS-Val =37% and MAPD In-situ-Val =45%), and the vast majority of points are distributed near the 1:1 line. In the actual measurement verification, the gray area represents MAPD. In-situ-Val <35% of the sample distribution (N=382), which accounts for 75% of the total (N=515), indicates that the accuracy of the vast majority of points is consistent with the MODIS validation results. Furthermore, the study found (Table 1) that the inversion accuracy of summer data was the highest (MAPD). In-situ-Val =33%), while the inversion accuracy of spring data is lower (MAPD). In-situ-Val =67%). Comparing the validation results at different latitudes, it was found that the validation accuracy of CALIOPChla was basically consistent.

[0077] Compared with existing technologies, this invention selects chlorophyll a concentration as a characteristic parameter of the marine ecosystem and employs big data deep learning methods to identify the complex nonlinear relationship between lidar echoes and sea surface chlorophyll a concentration from space-ground spatiotemporal big data. It establishes an AI-assisted spaceborne lidar remote sensing inversion model for Arctic Ocean chlorophyll a concentration. Based on this, it develops a long-term, spatiotemporally complete monthly average remote sensing product for Arctic Ocean sea surface chlorophyll a concentration, systematically revealing the spatiotemporal distribution characteristics and mechanisms of Arctic Ocean chlorophyll a concentration, particularly the spatial distribution characteristics during polar night, as well as the complete seasonal succession characteristics and long-term evolution trends under polar night conditions. This invention can invert chlorophyll a concentration using CALIOP data, without being limited by the complex geographical conditions of high-latitude seas or the scarcity of observational data during polar night, and also overcomes the influence of uneven spatial distribution of buoy observations. Furthermore, this invention can accurately characterize the nonlinear relationship between lidar echo signals and chlorophyll a concentration in high-latitude sea areas. This invention enables high-precision remote sensing inversion of chlorophyll a concentration using a spaceborne lidar, and can obtain a pan-Arctic annual chlorophyll a concentration dataset. Based on this dataset, the spatiotemporal distribution of chlorophyll a concentration during the polar night can be understood, and the high-precision spatiotemporal distribution of chlorophyll a concentration in the Arctic Ocean throughout the year can be obtained, effectively filling the gap that water color satellites cannot conduct winter half-year observations in the polar regions.

[0078] Based on the above-described method for generating Chla concentration distribution, this invention also provides a Chla concentration distribution generation system, such as... Figure 6 As shown, the system includes:

[0079] Dataset generation module 1 is used to acquire remote sensing observation data and field observation data of Chla concentration, and generate a model dataset based on the remote sensing observation data and field observation data. The remote sensing observation data includes CALIOP data and MODISChla data.

[0080] Chla concentration distribution generation module 2 is used to establish and train a Chla inversion model based on the model dataset, and input the CALIOP data into the Chla inversion model to obtain the Chla concentration distribution.

[0081] For specific limitations regarding a Chla concentration distribution generation system, please refer to the limitations regarding a Chla concentration distribution generation method described above, which will not be repeated here. Each module in the above system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0082] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described above.

[0083] Figure 7 This diagram illustrates the internal structure of a computer device in one embodiment, which may specifically be a terminal or a server. The computer device includes a processor, memory, a network interface, a display, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. The display screen may be a liquid crystal display (LCD) or an e-ink display. The input devices may be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0084] Those skilled in the art will understand that Figure 7The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specifically, the computing device may include more or fewer components than shown in the diagram, or combine certain components, or have the same component arrangement.

[0085] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0086] In summary, this invention provides a method, system, device, and medium for generating Chla concentration distribution. The method includes: acquiring remote sensing observation data and field observation data of Chla concentration; generating a model dataset based on the remote sensing observation data and field observation data, wherein the remote sensing observation data includes CALIOP data and MODISChla data; establishing and training a Chla inversion model based on the model dataset; and inputting the CALIOP data into the Chla inversion model to obtain the Chla concentration distribution. This invention can generate the spatiotemporal distribution of Chla concentration based on CALIOP data.

[0087] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0088] The embodiments described above are merely preferred embodiments of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the technical principles of this invention, and these improvements and substitutions should also be considered within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the scope of the claims.

Claims

1. A method for generating Chla concentration distribution, characterized in that, The method includes: Remote sensing and field observation data of Chla concentration in the Arctic Ocean were acquired, and the remote sensing data was quality controlled based on screening principles. A space-ground spatiotemporal dataset was generated based on the quality-controlled remote sensing and field observation data. The remote sensing data includes CALIOP data and MODISChla data, and the field observation data includes Arctic Ocean BGC-Argo buoy observation data. The step of generating a space-ground spatiotemporal dataset based on quality-controlled remote sensing observation data and field observation data includes: dividing the remote sensing observation data and field observation data according to a preset spatial distribution and time period; matching the CALIOP data in the same spatial distribution and time period with the MODISChla data and the field observation data respectively to generate a space-ground spatiotemporal dataset. The matching of the CALIOP data, which are spatially distributed and within the same time period, with the MODISChla data and the field observation data, respectively, further includes: Outliers in MODIS and measured data were removed based on the 3sigma principle; the CALIOP, MODIS and measured data after outlier removal were then gridded. Based on time and latitude and longitude, the gridded CALIOP data is spatiotemporally matched with MODIS and measured data according to a 2°×2° spatial window and a monthly time window; The space-ground spatiotemporal big data set includes a model dataset and an independent validation dataset; the model dataset consists of 90% training data and 10% test data. A Chla inversion model for the Arctic Ocean is established based on the aforementioned model dataset. The inversion model employs the Levenberg-Marquardt forward propagation algorithm as the training function to optimize model weights and biases, and selects mean squared error as the loss function. Tanh is used as the activation function between the input and hidden layers of the Chla inversion model, and Purelin is selected as the transfer function for the output layer. The input parameters of the Chla inversion model include the CALIOP vertical attenuation backscattering coefficient, temporal parameters, and spatial parameters. The accuracy of the inversion model is tested using the aforementioned test dataset, and its accuracy is validated using the aforementioned independent validation dataset. The acquired CALIOP data of the Arctic Ocean is input into the Chla inversion model to obtain the Chla concentration distribution of the Arctic Ocean.

2. The method for generating Chla concentration distribution according to claim 1, characterized in that, The acquisition of remote sensing data on Chla concentration in the Arctic Ocean includes: The CALIOP data is transiently corrected using a deconvolution algorithm.

3. The method for generating Chla concentration distribution according to claim 2, characterized in that, The transient correction of the CALIOP data using a deconvolution algorithm includes: The attenuated backscattered signal in the CALIOP data is deconvolved with the transient response function.

4. The method for generating Chla concentration distribution according to claim 1, characterized in that, The step of establishing and training the Chla inversion model for the Arctic Ocean based on the model dataset includes: The forward propagation algorithm is used to optimize the model weights and biases of the Chla inversion model.

5. The method for generating Chla concentration distribution according to claim 1, characterized in that, The acquisition of remote sensing data on Chla concentration in the Arctic Ocean includes: The total depolarization ratio of the ocean and the backscattering from the ocean subsurface lidar were calculated based on the transiently corrected CALIOP data as remote sensing observation data.

6. The method for generating Chla concentration distribution according to claim 1, characterized in that, The matching of the CALIOP data, which are in the same spatial distribution and time period, with the MODISChla data and the field observation data includes: If the MODISChla data and field observation data exist in the same spatial distribution and time period, only the field observation data will be retained.

7. A Chla concentration distribution generation system, characterized in that, The system includes: The dataset generation module is used to acquire remote sensing observation data and field observation data of Chla concentration in the Arctic Ocean, and to perform quality control on the remote sensing observation data based on screening principles, and generate a space-ground spatiotemporal big data set based on the quality-controlled remote sensing observation data and field observation data. The remote sensing data includes CALIOP data and MODISChla data; the field observation data includes Arctic Ocean BGC-Argo buoy observation data. The step of generating a space-ground spatiotemporal dataset based on quality-controlled remote sensing observation data and field observation data includes: dividing the remote sensing observation data and field observation data according to a preset spatial distribution and time period; matching the CALIOP data in the same spatial distribution and time period with the MODISChla data and the field observation data respectively to generate a space-ground spatiotemporal dataset. The matching of the CALIOP data, which are spatially distributed and within the same time period, with the MODISChla data and the field observation data, respectively, further includes: Outliers in MODIS and measured data were removed based on the 3sigma principle; the CALIOP, MODIS and measured data after outlier removal were then gridded. Based on time and latitude and longitude, the gridded CALIOP data is spatiotemporally matched with MODIS and measured data according to a 2°×2° spatial window and a monthly time window; The space-ground spatiotemporal big data set includes a model dataset and an independent validation dataset; the model dataset consists of 90% training data and 10% test data. The Chla concentration distribution generation module is used to establish a Chla inversion model of the Arctic Ocean based on the model dataset. The inversion model employs a forward propagation algorithm as the training function to optimize model weights and biases, and selects mean squared error as the loss function. Tanh is used as the activation function between the input and hidden layers of the Chla inversion model, and Purelin is selected as the transfer function for the output layer. The input parameters of the Chla inversion model include the CALIOP vertical attenuation backscattering coefficient, temporal parameters, and spatial parameters. The inversion model is tested for accuracy using the test dataset and validated for accuracy using the independent validation dataset. The acquired CALIOP data of the Arctic Ocean is input into the Chla inversion model to obtain the Chla concentration distribution of the Arctic Ocean.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images