Method for analyzing carbon source and sink rules in near-shore sea area based on global-regional migration learning
Patent Information
- Application Number
- CN202611182150.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-05
- Publication Date
- 2026-10-27
- Estimated Expiration
- 2046-08-05
AI Technical Summary
[0005]本发明提供一种基于全球-区域迁移学习的近岸海域碳源汇规律解析方法,以解决相关技术中,海气碳通量观测成本高、原位数据稀疏、时空分布不均导致近岸海域碳源汇规律解析不足等问题
[0024]本发明实施例可以融合全球海洋化学及案例区域实测数据等多源信息,挖掘常规易得指标与稀缺碳指标复杂映射关系,实现碳组分高分辨率预测、海-气CO2通量的快速估算与近岸海域碳源汇规律高精度解析,为海洋碳中和策略制定及近岸海域生态环境管理提供技术支撑。由此,解决了相关技术中,海气碳通量观测成本高、原位数据稀疏、时空分布不均导致近岸海域碳源汇规律解析不足等问题。
Smart Images

Figure CN122712087B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of marine environmental monitoring and big data analysis technology, and in particular to a method for analyzing the carbon source and sink patterns in nearshore waters based on global-regional transfer learning. Background Technology
[0002] Nearshore waters are a vital carrier of human development and an important carbon sink area in the ocean. With the intensification of climate change, accurate analysis of carbon sources and sinks in nearshore waters has become an important prerequisite for achieving the "dual carbon" goal, and is of great significance to the global carbon cycle balance and regional ecological security.
[0003] Among related technologies, nearshore marine carbon source and sink apportionment technology can conduct in-situ monitoring and laboratory testing to obtain high-quality data, providing basic support for carbon source and sink apportionment. At the same time, some technologies attempt to obtain carbon component-related data through remote sensing inversion methods, trying to make up for the limitations of in-situ and laboratory testing, and help to analyze the pattern and explore the laws of nearshore marine carbon sources and sinks.
[0004] However, among related technologies, nearshore marine carbon source and sink apportionment technology is mainly constrained by cross-interface carbon flux data. In daily monitoring, data on key carbon components such as CO2 (carbon dioxide) and DIC (Dissolved Inorganic Carbon) are scarce. Although in-situ monitoring and laboratory testing can obtain high-quality data, the data is sparse, costly, and has large sampling uncertainties. Although remote sensing inversion technology has the advantage of high spatial accuracy, it suffers from problems such as large workload for image interpretation and susceptibility to cloud cover and atmospheric aerosols. In terms of feature analysis, commonly used methods such as correlation analysis and principal component analysis are mostly based on limited discrete data and cannot obtain high-precision carbon source and sink patterns. Moreover, the meteorological and hydrological conditions in nearshore marine areas are complex and significantly affected by human activities, resulting in obvious spatiotemporal heterogeneity in carbon source and sink patterns, which urgently needs improvement. Summary of the Invention
[0005] This invention provides a method for analyzing the carbon source and sink patterns in nearshore waters based on global-regional transfer learning, in order to solve the problems in related technologies, such as high cost of air-sea carbon flux observation, sparse in-situ data, and uneven spatiotemporal distribution leading to insufficient analysis of carbon source and sink patterns in nearshore waters.
[0006] The first aspect of this invention provides a method for analyzing the carbon source and sink patterns in nearshore waters based on global-regional transfer learning, comprising the following steps: collecting global marine chemical data and measured data from a case study region; inputting the global marine chemical data into a preset global-scale carbon data-driven model, and using the measured data from the case study region to perform localized verification of the preset global-scale carbon data-driven model to obtain a verified model, and transferring the verified model to the nearshore waters of the case study region to output the target dissolved inorganic carbon and target carbon dioxide partial pressure of the nearshore waters of the case study region; calculating the sea-air carbon dioxide flux of the nearshore waters of the case study region based on the target dissolved inorganic carbon, the target carbon dioxide partial pressure, and preset meteorological data, and analyzing the carbon source and sink variation patterns of the nearshore waters of the case study region based on the sea-air carbon dioxide flux using at least one of spatiotemporal distribution, influencing factors, and gradient characteristics.
[0007] Optionally, in one embodiment of the present invention, the step of analyzing the carbon source and sink variation patterns of the nearshore waters of the case area based on the sea-air carbon dioxide flux using at least one of spatiotemporal distribution, influencing factors, and gradient characteristics includes: generating a distribution map of carbon dioxide partial pressure and sea-air carbon dioxide flux that meets preset high spatiotemporal resolution conditions based on the target carbon dioxide partial pressure and the sea-air carbon dioxide flux; performing statistical analysis on a monthly and annual scale based on the distribution map to obtain temporal analysis results, and performing statistical analysis on a global, sub-sea area, and hotspot area scale to obtain spatial analysis results, and then, according to the... The temporal and spatial analysis results are used to analyze the carbon source and sink variation patterns in the nearshore waters of the case area; the main meteorological and aquatic environmental factors affecting the sea-air carbon dioxide flux are identified, and the characteristic importance of the main meteorological and aquatic environmental factors is quantitatively analyzed to obtain characteristic analysis results. Based on the characteristic analysis results, the carbon source and sink variation patterns in the nearshore waters of the case area are analyzed; the distribution variation characteristics of the sea-air carbon dioxide flux are analyzed based on the salinity and nutrient gradients to obtain variation characteristic analysis results. Based on the variation characteristic analysis results, the carbon source and sink variation patterns in the nearshore waters of the case area are analyzed.
[0008] Optionally, in one embodiment of the present invention, before inputting the global ocean chemical data into a preset global-scale carbon data-driven model, the method further includes: acquiring temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate from the global ocean chemical data; and constructing the preset global-scale carbon data-driven model based on the temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate from the global ocean chemical data.
[0009] Optionally, in one embodiment of the present invention, the collection of global ocean chemical data and case area measured data includes: collecting at least one of global ship-based survey data, buoy observation data, and preset model output data, and determining the global ocean chemical data based on at least one of the global ship-based survey data, the buoy observation data, and the preset model output data; collecting daily environmental monitoring data, automatic water quality monitoring data, cruise supplementary monitoring data, and meteorological and hydrological data of the case area, and determining the case area measured data based on the daily environmental monitoring data, the automatic water quality monitoring data, the cruise supplementary monitoring data, and the meteorological and hydrological data of the case area.
[0010] Optionally, in one embodiment of the present invention, the step of using measured data from the case area to localize and validate the preset global-scale carbon data-driven model to obtain a validated model, and then applying the validated model to the nearshore waters of the case area to output the partial pressures of dissolved inorganic carbon and carbon dioxide in the target waters of the nearshore waters of the case area, includes: comparing the measured values of dissolved inorganic carbon, total alkalinity, and partial pressure of carbon dioxide in the nearshore waters of the case area with the simulated values to generate a comparison result, and performing localization and validation of the preset global-scale carbon data-driven model based on the comparison result to obtain the validated model; and applying the validated model to the nearshore waters of the case area to predict the partial pressures of dissolved inorganic carbon and carbon dioxide in the target waters of the nearshore waters of the case area using temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate from the global ocean chemical data.
[0011] Optionally, in one embodiment of the present invention, the step of calculating the sea-air carbon dioxide flux of the nearshore waters of the case area based on the dissolved inorganic carbon in the target water body, the target carbon dioxide partial pressure, and preset meteorological data includes: back-calculating the water carbon dioxide partial pressure in the carbon dioxide partial pressure based on the dissolved inorganic carbon in the target water body and the measured pH value; and calculating the sea-air carbon dioxide flux of the nearshore waters of the case area based on the water carbon dioxide partial pressure, the atmospheric carbon dioxide partial pressure, and the sea surface wind speed.
[0012] Optionally, in one embodiment of the present invention, the formula for calculating the sea-air carbon dioxide flux is:
[0013] in, This refers to the sea-air carbon dioxide flux. The carbon dioxide transport rate at the air-sea interface. This represents the solubility of carbon dioxide in water. The partial pressure of carbon dioxide in the water body. The partial pressure of atmospheric carbon dioxide is given.
[0014] A second aspect of the present invention provides a device for analyzing nearshore marine carbon source and sink patterns based on global-regional transfer learning, comprising: a data acquisition module for acquiring global marine chemical data and measured data from a case study area; a verification module for inputting the global marine chemical data into a preset global-scale carbon data-driven model, and using the measured data from the case study area to perform localized verification of the preset global-scale carbon data-driven model to obtain a verified model, and transferring the verified model to the nearshore marine area of the case study area to output the target dissolved inorganic carbon and target carbon dioxide partial pressure of the nearshore marine area of the case study area; and an analysis module for calculating the sea-air carbon dioxide flux of the nearshore marine area of the case study area based on the target dissolved inorganic carbon, the target carbon dioxide partial pressure, and preset meteorological data, and analyzing the carbon source and sink variation patterns of the nearshore marine area of the case study area based on the sea-air carbon dioxide flux using at least one of spatiotemporal distribution, influencing factors, and gradient characteristics.
[0015] Optionally, in one embodiment of the present invention, the analysis module includes: a generation unit, configured to generate a distribution map of carbon dioxide partial pressure and sea-atmosphere carbon dioxide flux that meets preset high spatiotemporal resolution conditions based on the target carbon dioxide partial pressure and the sea-atmosphere carbon dioxide flux; and a first analysis unit, configured to perform statistical analysis on a monthly and annual scale based on the distribution map to obtain temporal analysis results, and perform statistical analysis on a global, sub-sea area, and hotspot area scale to obtain spatial analysis results, and analyze the nearshore sea area of the case region based on the temporal analysis results and the spatial analysis results. The system comprises three sub-units: a first sub-unit, which analyzes the carbon source and sink variation patterns of the region; a second sub-unit, which identifies the main meteorological and aquatic environmental factors affecting the sea-air carbon dioxide flux, performs quantitative analysis of the characteristic importance of the main meteorological and aquatic environmental factors, obtains the characteristic analysis results, and analyzes the carbon source and sink variation patterns of the nearshore waters of the case area based on the characteristic analysis results; and a third sub-unit, which analyzes the distribution variation characteristics of the sea-air carbon dioxide flux based on the salinity and nutrient gradients, obtains the variation characteristic analysis results, and analyzes the carbon source and sink variation patterns of the nearshore waters of the case area based on the variation characteristic analysis results.
[0016] Optionally, in one embodiment of the present invention, it further includes: an acquisition module, configured to acquire temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate from the global ocean chemical data before inputting the global ocean chemical data into a preset global-scale carbon data-driven model; and a construction module, configured to construct the preset global-scale carbon data-driven model based on the temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate from the global ocean chemical data.
[0017] Optionally, in one embodiment of the present invention, the acquisition module includes: a first acquisition unit, used to acquire at least one of global ship-based survey data, buoy observation data, and preset model output data, and to determine the global ocean chemical data based on at least one of the global ship-based survey data, the buoy observation data, and the preset model output data; and a second acquisition unit, used to acquire daily environmental monitoring data, automatic water quality monitoring data, cruise supplementary monitoring data, and meteorological and hydrological data of the case area, and to determine the measured data of the case area based on the daily environmental monitoring data, the automatic water quality monitoring data, the cruise supplementary monitoring data, and the meteorological and hydrological data of the case area.
[0018] Optionally, in one embodiment of the present invention, the verification module includes: a verification unit, used to compare the measured values of dissolved inorganic carbon, total alkalinity, and partial pressure of carbon dioxide in the nearshore waters of the case area with the simulated values, generate a comparison result, and perform localized verification of the preset global-scale carbon data-driven model based on the comparison result to obtain the verified model; and a prediction unit, used to apply the verified model to the nearshore waters of the case area to predict the dissolved inorganic carbon and target carbon dioxide partial pressure of the target water body in the nearshore waters of the case area using the temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate in the global ocean chemical data.
[0019] Optionally, in one embodiment of the present invention, the analysis module includes: a back-calculation unit, used to back-calculate the partial pressure of carbon dioxide in the water body based on the dissolved inorganic carbon in the target water body and the measured pH value; and a calculation unit, used to calculate the sea-air carbon dioxide flux in the nearshore waters of the case area based on the partial pressure of carbon dioxide in the water body, the partial pressure of carbon dioxide in the atmosphere, and the sea surface wind speed.
[0020] Optionally, in one embodiment of the present invention, the formula for calculating the sea-air carbon dioxide flux is:
[0021] in, This refers to the sea-air carbon dioxide flux. The carbon dioxide transport rate at the air-sea interface. This represents the solubility of carbon dioxide in water. The partial pressure of carbon dioxide in the water body. The partial pressure of atmospheric carbon dioxide is given.
[0022] A third aspect of the present invention provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the nearshore marine carbon source and sink pattern analysis method based on global-regional transfer learning as described in the above embodiments.
[0023] A fourth aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for analyzing nearshore marine carbon source-sink patterns based on global-regional transfer learning.
[0024] This invention integrates multi-source information, including global marine chemistry and field-measured data from case study areas, to uncover the complex mapping relationship between readily available and scarce carbon indicators. This enables high-resolution prediction of carbon components, rapid estimation of air-sea CO2 fluxes, and high-precision analysis of nearshore carbon source-sink patterns, providing technical support for marine carbon neutrality strategy formulation and nearshore marine ecological environment management. This solves problems in related technologies, such as high observation costs of air-sea carbon fluxes, sparse in-situ data, and uneven spatiotemporal distribution leading to insufficient analysis of nearshore carbon source-sink patterns.
[0025] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description
[0026] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A flowchart illustrating a method for analyzing nearshore marine carbon source and sink patterns based on global-regional transfer learning, according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating a method for analyzing nearshore marine carbon source-sink patterns based on global-regional transfer learning according to an embodiment of the present invention. Figure 3 To input variable sampling information and histograms of various indicators into a global ocean chemistry dataset according to an embodiment of the present invention; Figure 4 A comparative chart showing the performance of nine machine learning algorithms for estimating dissolved inorganic carbon concentration based on global-scale data according to an embodiment of the present invention; Figure 5 A comparative chart showing the performance of nine machine learning algorithms for estimating total alkalinity based on global-scale data according to an embodiment of the present invention; Figure 6 A comparative chart showing the performance of nine machine learning algorithms for estimating carbon dioxide fugacity based on global-scale data according to an embodiment of the present invention; Figure 7 A comparison chart of simulated carbon dioxide partial pressure values in water bodies using a global-scale data-driven model of optimal dissolved inorganic carbon according to an embodiment of the present invention, and locally measured values. Figure 8 This is a comparison chart of simulated carbon dioxide partial pressure values in water bodies from a global-scale total alkalinity optimal data-driven model according to an embodiment of the present invention, and locally measured values. Figure 9 A comparison chart of simulated carbon dioxide partial pressure in water bodies with locally measured values, based on global-scale optimal data-driven carbon dioxide fugacity according to an embodiment of the present invention; Figure 10 This is a monthly spatial distribution map of air-sea carbon dioxide flux in a nearshore area according to an embodiment of the present invention. Figure 11 A diagram illustrating the influencing factors and contributions of air-sea carbon dioxide flux in a nearshore marine area according to an embodiment of the present invention. Figure 12 This is a gradient distribution map of sea-air carbon dioxide flux, salinity, and eutrophication index in a nearshore sea area according to an embodiment of the present invention. Figure 13 This is a schematic diagram of a nearshore marine carbon source and sink pattern analysis device based on global-regional transfer learning according to an embodiment of the present invention; Figure 14 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present invention. Detailed Implementation
[0027] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.
[0028] The following description, with reference to the accompanying drawings, describes an embodiment of the present invention of a method for analyzing nearshore marine carbon source and sink patterns based on global-regional transfer learning. Addressing the problems mentioned in the background art, such as high costs of air-sea carbon flux observation, sparse in-situ data, and uneven spatiotemporal distribution leading to insufficient analysis of nearshore marine carbon source and sink patterns, this invention provides a method for analyzing nearshore marine carbon source and sink patterns based on global-regional transfer learning. This method integrates multi-source information, including global marine chemistry and measured data from case study regions, and uncovers the complex mapping relationship between readily available and scarce carbon indicators. It achieves high-resolution prediction of carbon components, rapid estimation of air-sea CO2 flux, and high-precision analysis of nearshore marine carbon source and sink patterns, providing technical support for the formulation of marine carbon neutrality strategies and the management of nearshore marine ecological environment. Thus, it solves the problems of high costs of air-sea carbon flux observation, sparse in-situ data, and uneven spatiotemporal distribution leading to insufficient analysis of nearshore marine carbon source and sink patterns in related technologies.
[0029] Specifically, Figure 1 This is a flowchart illustrating a method for analyzing nearshore marine carbon source-sink patterns based on global-regional transfer learning, provided in an embodiment of the present invention.
[0030] like Figure 1 As shown, this method for analyzing nearshore marine carbon source-sink patterns based on global-regional transfer learning includes the following steps: In step S101, global marine chemical data and measured data from the case study area are collected.
[0031] In this embodiment of the invention, multi-source data such as global ocean chemical data and measured data from case areas can be collected and processed to provide support for the construction and selection of subsequent global-scale carbon data-driven models.
[0032] Optionally, in one embodiment of the present invention, collecting global ocean chemical data and measured data of the case area includes: collecting at least one of global ship-based survey data, buoy observation data, and preset model output data, and determining global ocean chemical data based on at least one of global ship-based survey data, buoy observation data, and preset model output data; collecting daily environmental monitoring data, automatic water quality monitoring data, cruise supplementary monitoring data, and meteorological and hydrological data of the case area, and determining measured data of the case area based on the daily environmental monitoring data, automatic water quality monitoring data, cruise supplementary monitoring data, and meteorological and hydrological data of the case area.
[0033] It is understood that the global ocean chemical data and case area measured data collected in the embodiments of the present invention include data from various sources such as publicly available global ship surveys (global ship survey data), buoy observations and other preset model outputs, as well as data and information such as daily environmental monitoring, automatic water quality monitoring, cruise supplementary monitoring, meteorological and hydrological data of the case area.
[0034] Specifically, embodiments of the present invention can collect and process global ocean chemical data and measured data from case study areas. This includes collecting ocean chemical data from various sources globally, including publicly available ship-based surveys, buoy observations, and model outputs, and generating ocean chemical datasets from multiple sources. It also includes collecting data from daily water quality monitoring, automatic water quality monitoring, supplementary on-site cruise monitoring, meteorological and hydrological data, and geographic layers from case study areas. The global ocean chemical dataset includes DIC (Dissolved Inorganic Carbon), TA (Total Alkalinity), WT (Water Temperature), Sal (Salinity), pH, DO (Dissolved Oxygen), and NO3. - (nitrates), PO4 3- (Phosphate) and other indicators; water quality indicators in the case study area include WT, Sal, pH, DO, and NO3. - -N (nitrate nitrogen), NO2 - -N (nitrite nitrogen), NH4 + -N (ammonia nitrogen), PO4 3- -P (phosphate), Chla (chlorophyll a), etc., supplementary monitoring during the on-site cruise included DIC, TA, p The monitoring also includes scarce carbon components such as CO2 (Partial Pressure of Carbon Dioxide) and water quality physicochemical indicators that are monitored simultaneously.
[0035] This invention can collect global marine chemistry data and case study regional measured data. By fusing data from multiple global sources, a unified global marine chemistry dataset is constructed, and the case study regional measured data is organized to form a consistent case study regional dataset, providing a reliable data foundation for subsequent model training, testing and validation.
[0036] For example, this invention can first integrate four rigorously quality-controlled and non-overlapping comprehensive datasets from the globally publicly available OCADS (Ocean Carbon and Acidification Data System): GLODAP (Global Ocean Data Analysis Project), PACIFICA (Pacific Ocean Carbon Data Project), CARINA (Carbon in the North Atlantic), and CODAPNA (Carbon Dioxide in the Arctic and North Atlantic). These datasets include multi-source data from shipboard surveys, buoy observations, and model outputs, covering parameters such as inorganic carbon, total alkalinity, pH, dissolved oxygen, and nutrients, encompassing 1663 expeditions from 1972 to 2021, with a cumulative collection of 1.99 million water samples. This invention selects data with a quality control criterion of 2 (i.e., good data quality) from the Global Ocean Circulation Laboratory and removes negative, zero, and outlier values in indicators such as dissolved oxygen and nutrients to ensure the reliability and accuracy of the data used.
[0037] Meanwhile, this invention collected and organized data from the case area from 2017 to 2023, including daily water quality monitoring, automatic water quality monitoring, supplementary on-site cruise monitoring, meteorological and hydrological data, and geographic maps. The case area of this invention is the nearshore waters of the Guangdong-Hong Kong-Macao Greater Bay Area. Based on historical data collection and preliminary analysis, this invention conducted supplementary on-site cruise monitoring at key routine water quality monitoring points in the estuary and nearshore areas of the case area during different water seasons (high, normal, and low) in April, August, and November 2023. A total of 98 water samples were collected for laboratory testing. The tested physicochemical indicators included pH, DO (dissolved oxygen), WT (water temperature), Salinity, TA (total alkalinity), DIC (dissolved inorganic carbon), and DIN (dissolved inorganic nitrogen) (containing NO3). — N (Nitrogen), NO2 — N (Nitrite Nitrogen), NH4 + -N (Ammonium Nitrogen), PO4 3- -P (reactive phosphate), etc. In addition, in-situ monitoring of the atmosphere and surface seawater in the case study area... pA total of 37,562 valid data points were obtained for CO2. Among these, DIC and TA samples were collected from discrete locations, resulting in a limited sample size; water bodies... p CO2 is continuously monitored using a shipboard mobile monitoring method to acquire high spatial resolution data. p By matching CO2 data with environmental factors, a localized independent validation dataset was constructed, providing a reliable data foundation for subsequent localized model validation and migration applications.
[0038] This invention can analyze global marine chemistry datasets and existing monitoring data from case study areas to identify relatively readily available model input variables with large sample sizes and related to carbon composition, such as WT, Sal, pH, DO, and NO3. - PO4 3- Key sparsity indicators that need to be enhanced include DIC (dissolved inorganic carbon) and TA (total alkalinity). f CO2 (fugacity of carbon dioxide). Through data quality control, 337,973 sets of data were obtained where all six input variables met the criteria, matching DIC, TA, and [other parameters]. f The CO2 index data are 292543, 297772, and 38797 sets, respectively. Table 1 is a statistical characteristic information table of each input and output variable of the global ocean chemical dataset provided in this embodiment of the invention, as shown in Table 1:
[0039] Based on this, the present invention statistically analyzed the basic sampling information (sampling time, latitude and longitude, and sampling depth) of the input variables in the global-scale training and testing sets, as well as the histograms of each indicator, as shown below. Figure 2 As shown in the figure, the data is mainly concentrated in the spring and summer seasons of nearly 20 years from 1995 to 2015, and spatially highly concentrated in surface water bodies (especially those less than 200 m) in the mid-to-low latitudes of the Northern Hemisphere. The index values have a wide distribution range, with water temperature covering a cold to warm water range from -2.17 to 32.17 ℃, and salinity ranging from 0.12 to 40.61 PSU from freshwater to seawater. The nutrient indices nitrate nitrogen and phosphate exhibit a multi-peak distribution, covering water bodies with different trophic states from oligotrophic to eutrophic. The case study area itself also exhibits significant spatial heterogeneity in water quality, including bay areas with relatively good water quality and areas with water quality worse than Class IV seawater standards. The input variables for the case study area, including temperature, salinity, and nutrients, are generally within the range of the global dataset. Therefore, it is feasible to subsequently transfer the global data model to the case study area.
[0040] It should be noted that the water quality indicators in the global ocean chemical data of this invention are mostly in μmol / kg, while the routine environmental monitoring indicators are mostly in mg / L, which need to be converted using the relative atomic masses of the corresponding metrological elements.
[0041] Optionally, in one embodiment of the present invention, before inputting global ocean chemical data into a preset global-scale carbon data-driven model, the method further includes: acquiring temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate from the global ocean chemical data; and constructing a preset global-scale carbon data-driven model based on the temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate from the global ocean chemical data.
[0042] It is understood that, based on the global ocean chemical dataset obtained above, the embodiments of the present invention construct conventional indicators such as temperature and salinity, as well as DIC, p This invention provides global-scale carbon data-driven models for comparing and selecting scarce carbon indicators such as CO2. The embodiments of this invention can also mine data from conventional water quality indicators such as temperature, salinity, and nutrients, as well as DIC (Dissipative Carbon Dioxide). p The complex mapping relationships of scarce carbon indicators such as CO2 are explored through specific steps including input factor screening and preprocessing, machine learning model construction, and model performance evaluation and comparison. This invention can comprehensively utilize multi-source data, including global marine chemical data and measured data from case areas, to construct a data enhancement model for scarce carbon components in nearshore waters that combines transferability and mechanistic interpretability. It can predict DIC, TA, and other parameters using readily available data such as temperature, salinity, and nutrients. p Key scarce carbon indicators such as CO2 are enhanced by shifting from high-frequency to scarce data.
[0043] In practical implementation, embodiments of the present invention can construct and compare global-scale carbon data-driven models, using routinely monitored indicators such as temperature, salinity, pH, dissolved oxygen, and nitrogen and phosphorus nutrients as inputs, and DIC, TA, etc. pUsing scarce carbon components such as CO2 as outputs, a global-scale carbon data-driven model is constructed. This invention can remove negative values, zero values, and outliers from the input data. The outlier removal method uses a factor of 3 (3 standard deviations) to convert the dataset into a new dataset with a mean of 0 and a standard deviation of 1. The training and validation sets are randomly divided in an 8:2 ratio. Nine machine learning algorithms are selected to construct a global-scale data-driven model: MLR (Multiple Linear Regression), KNN (K-Nearest Neighbors), RF (Random Forest), AdaBoost (Adaptive Boosting), XGBoost (Extreme Gradient Boosting), LightGBM (Light Gradient Boosting Machine), SVM (Support Vector Machine), MLP (Multi-Layer Perceptron), and CNN (Convolutional Neural Network). The main parameters of each algorithm are set. Table 2 shows the main parameter settings for the machine learning algorithms provided in this invention.
[0044] The embodiments of the present invention use R 2 Multiple metrics, including coefficient of determination (CRD), root mean square error (RMSE), and mean absolute percentage error (MAPE), are used to evaluate the performance of different algorithms on DIC, TA, and [other parameters]. f The performance evaluation results of the CO2 simulation are as follows: Figures 2 to 5 As shown in the figure. The results show that Random Forest (RF) and K-Nearest Neighbors (KNN) algorithms outperform in all metrics, especially in predicting DIC and TA with extremely high accuracy. Among them, Random Forest (RF) performs best in all prediction tasks, with its R-values on the training set and test set being the highest. 2 All values were above 0.97, and the MAPE ranged from 0.05% to 3.08%, demonstrating excellent stability and generalization ability. Therefore, based on six routinely monitored aquatic environmental factors, machine learning algorithms such as random forests can be applied to achieve the analysis of key parameters (DIC, TA, and...) of the seawater carbonate system. fHigh-precision and robust simulation of CO2. In particular, the random forest algorithm, with its high explained variance and low relative error, can efficiently and cost-effectively estimate key carbon parameters in marine areas where conventional carbon monitoring is not conducted, using relatively easily obtainable aquatic environmental factors. This provides a feasible technical path for dynamic monitoring of global and regional marine carbon indicators and carbon cycle research. To reduce the uncertainty introduced by the randomness of dataset partitioning in machine learning models, this invention, based on model performance evaluation, selects appropriate parameters for DIC, TA, and... f The two best-performing algorithms for CO2 (based on R values on the test set) 2 (Sorting), and further robustness and reliability verification are performed using 10-fold cross-validation. Table 3 shows the 10-fold cross-validation results of the two best-performing algorithms provided in this embodiment of the invention, as shown in Table 3:
[0045] Random Forest (RF) algorithm in DIC, TA and f In CO2 predictions, not only R 2 The mean is the highest, and its R... 2 The variance is generally low, indicating optimal overall performance and strongest stability. Therefore, this invention adopts the random forest model as the preferred method for high-precision simulation of marine carbon concentration.
[0046] This invention utilizes the PyTorch deep learning library on the Python platform and GPU (Graphics Processing Unit) acceleration to achieve efficient algorithm computation. It employs R... 2 Multiple metrics such as (coefficient of determination), RMSE (root mean square error), MAPE (mean absolute percentage error), and MAE (mean absolute error) are used to evaluate the performance of machine learning algorithms. 2 The optimal machine learning algorithm for prediction was selected based on RMSE, and the two optimal algorithms were cross-validated using a 10-fold cross-validation method to reduce the risk of overfitting caused by the uncertainty of dataset partitioning. Considering that CO2 observation data in the global ocean chemistry dataset is relatively scarce, this invention also utilizes the carbonate system chemical relationship (CO2SYS chemical thermodynamic model) for cross-validation. Specifically, the dissolved inorganic carbon or alkalinity obtained by the machine learning algorithm is combined with the measured pH, and the CO2SYS model is used to estimate the... p CO2, and centrally measured data from global ocean chemistry datasets. f CO2 (carbon dioxide fugacity, which can be determined by the fugacity coefficient) and... p The CO2 conversion data was then subjected to secondary verification. f CO2 and p The CO2 conversion formula is shown below:
[0047] in, The partial pressure of carbon dioxide in water, uatm, is obtained in this invention through the aforementioned machine learning method and CO2SYS model; T Water temperature (Celsius), °C.
[0048] In step S102, global ocean chemical data is input into a preset global-scale carbon data-driven model, and the preset global-scale carbon data-driven model is localized and validated using measured data from the case area to obtain a validated model. The validated model is then applied to the nearshore waters of the case area to output the partial pressure of dissolved inorganic carbon and carbon dioxide in the target water body of the nearshore waters of the case area.
[0049] In practical implementation, this invention can perform localized verification and transfer applications of global-scale models. It utilizes optimal machine learning algorithms and limited measured data from the case study region for independent localized verification and error analysis, ensuring the reliability of model transfer to the case study region. Verification metrics include DIC, TA, and [other metrics not specified in the original text]. p CO2, like the global model, uses R 2 Error analysis was performed using RMSE, MAPE, and MAE. This invention inputs various sources of marine chemical data, including globally publicly available ship surveys, buoy observations, and model outputs, into a pre-set global-scale carbon data-driven model. The pre-set global-scale carbon data-driven model is then localized and validated using measured data from the case area, including daily environmental monitoring, automatic water quality monitoring, cruise supplementary monitoring, and meteorological and hydrological data. The validated model is then applied to the nearshore waters of the case area to output the partial pressures of dissolved inorganic carbon and carbon dioxide in the target water bodies of the nearshore waters of the case area.
[0050] The embodiments of the present invention can independently verify and transfer the global-scale carbon data-driven model established above using localized measured data. This includes verifying the selected global-scale carbon data model in a case area and transferring it to the case area. It also utilizes local high-frequency temperature, salinity and other data to enhance the scarce carbon components, thereby realizing data enhancement and sample expansion of scarce indicators by inverting high-frequency conventional observation data.
[0051] Optionally, in one embodiment of the present invention, a preset global-scale carbon data-driven model is locally validated using measured data from the case area to obtain a validated model. The validated model is then transferred and applied to the nearshore waters of the case area to output the partial pressures of dissolved inorganic carbon and carbon dioxide in the target waters of the nearshore waters of the case area. This includes: comparing the measured values of dissolved inorganic carbon, total alkalinity, and partial pressure of carbon dioxide in the nearshore waters of the case area with the simulated values to generate comparison results; and performing local validation of the preset global-scale carbon data-driven model based on the comparison results to obtain a validated model. The validated model is then transferred and applied to the nearshore waters of the case area to predict the partial pressures of dissolved inorganic carbon and carbon dioxide in the target waters of the nearshore waters of the case area using temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate from global marine chemical data.
[0052] It is understood that the embodiments of the present invention can be verified using limited measured data, which verifies limited dissolved inorganic carbon, total alkalinity and carbon dioxide partial pressure. Then, a machine learning model is used for enhancement, and the output of the target dissolved inorganic carbon and target carbon dioxide partial pressure of the water body are the unobserved time and location, that is, the data-enhanced dissolved inorganic carbon and carbon dioxide partial pressure of the water body.
[0053] In actual implementation, embodiments of the present invention can conduct localized verification and migration applications, using case regions DIC, TA, and p The measured values of CO2, the simulated values of DIC and TA by the global-scale random forest model, and the values obtained by combining the CO2SYS model (PyCO2SYS v1.8) p Simulated CO2 values were used to conduct independent local validation of the model. The comparison results between simulated and measured DIC values are as follows: Figure 6 As shown, the results indicate that its MAPE is 8.29%, and pH and DIC were obtained using the CO2SYS model. p R-squared values of CO2 simulation results 2 With a total alkalinity (TA) of 0.77 and a MAPE of 17.32%, the random forest model based on global-scale data shows generally acceptable performance in the nearshore waters of the case study area. The simulation comparison results for total alkalinity (TA) are as follows: Figure 7 As shown, MAPE is 3.39%, and pH and TA were obtained using the CO2SYS model. p R-squared values of CO2 simulation results 2 The total alkalinity simulation results are good, with a value of 0.74 and a MAPE of 21.06%. p The accuracy of CO2 is slightly too high. It was simulated directly using a global-scale random forest model. f CO2, then converted through the fugacity coefficient p Comparison of CO2 simulation results with measured values ( Figure 8 ), R2 The simulation results are not very accurate, with a value of 0.24 and a MAPE of 41.95%. This is presumably due to issues with global data. f The quantity and diversity of CO2 samples are insufficient to support reliable application in the nearshore waters of the Guangdong-Hong Kong-Macao Greater Bay Area. Therefore, this invention selects a global-scale DIC random forest model and combines it with the CO2SYS model to predict the target DIC and target [data missing] in the nearshore waters of the case study area. p CO2, and selected the nearshore waters of the case area based on the global-scale TA random forest model for prediction.
[0054] In step S103, based on the dissolved inorganic carbon in the target water body, the partial pressure of the target carbon dioxide, and the preset meteorological data, the sea-air carbon dioxide flux in the nearshore waters of the case area is calculated. Based on the sea-air carbon dioxide flux, the carbon source and sink variation patterns in the nearshore waters of the case area are analyzed using at least one of the spatiotemporal distribution, influencing factors, and gradient characteristics.
[0055] It is understood that the sea-air carbon dioxide flux in the embodiments of the present invention refers to the sea-air CO2 flux in nearshore waters.
[0056] In practical implementation, embodiments of the present invention can use a gas transport dynamics model to calculate the sea-air carbon dioxide flux in the nearshore waters of the case area based on dissolved inorganic carbon, target carbon dioxide partial pressure, and preset meteorological data. Based on the sea-air carbon dioxide flux, the invention reveals the high-precision carbon source and sink variation patterns in nearshore waters from multiple dimensions such as spatiotemporal distribution, influencing factors, and gradient characteristics. The present invention proposes a new technical path for low-cost and high-efficiency carbon composition prediction, effectively improving the accuracy of nearshore carbon source and sink pattern analysis, and providing technical support for the formulation of marine carbon neutrality strategies and the management of nearshore marine ecological environment.
[0057] The high-precision carbon source and sink pattern analysis in nearshore waters in this embodiment of the invention includes the spatiotemporal distribution of carbon sources and sinks, influencing factors of carbon source and sink patterns, and analysis of salinity and nutrient gradient characteristics. By integrating multi-source information such as global marine chemistry and measured data from case areas, this invention has uncovered the complex mapping relationship between commonly available indicators and scarce carbon indicators, achieving high-resolution prediction of carbon components, rapid estimation of air-sea CO2 flux, and high-precision analysis of nearshore carbon source and sink patterns, providing technical support for the formulation of marine carbon neutrality strategies and the management of nearshore marine ecological environment.
[0058] Optionally, in one embodiment of the present invention, the formula for calculating the sea-air carbon dioxide flux is:
[0059] in, This refers to the sea-atmosphere carbon dioxide flux. The carbon dioxide transport rate at the air-sea interface. This represents the solubility of carbon dioxide in water. This refers to the partial pressure of carbon dioxide in water. This is the partial pressure of atmospheric carbon dioxide.
[0060] In this embodiment of the invention, the sea-to-air CO2 flux in nearshore waters can be calculated using a gas transport kinetic model. The flux is determined by the gas exchange rate, gas solubility, and the partial pressure difference of CO2 between the sea and air. A positive sea-to-air CO2 flux indicates that the ocean releases CO2 into the atmosphere, acting as a "carbon source" for the atmosphere; a negative sea-to-air CO2 flux indicates that the ocean absorbs atmospheric CO2, acting as a "carbon sink" for the atmosphere. The formula for calculating the sea-to-air CO2 flux is as follows:
[0061] in, For air-sea CO2 flux, ; For the CO2 transport rate at the air-sea interface, ; The partial pressure of carbon dioxide in water, uatm; The atmospheric carbon dioxide partial pressure, uatm, is used in this invention as the average measured value of the case area. This represents the solubility of carbon dioxide in water. The calculations are based on Weiss's research, and the specific calculations are shown in the following formula:
[0062]
[0063] in, Tk is the solubility of carbon dioxide in water, K is the water temperature (K), and Sal is the salinity of the water body (PSU).
[0064] Gas transport rate at the air-sea interface wind speed 10 m above the water surface The functional relationship between them can be expressed as follows:
[0065] in, The instantaneous wind speed at 10 m above the water surface, in m / s. The Schmidt number for measuring water is calculated by dividing the kinematic viscosity of water by the molecular diffusion coefficient of the gas. This parameter is dimensionless and depends on the molar mass of the gas, the salinity of the water, and the temperature, as shown in the following formula:
[0066] in, To measure the Schmidt number of a water body, , The Schmidt numbers are for seawater (salinity equal to 35 PSU) and freshwater (salinity equal to 0 PSU), respectively. T (Related to water temperature (Celsius), °C), the calculation formula is shown below:
[0067] in, , where is the Schmidt number for seawater or freshwater, dimensionless; A~E are the equation coefficients, based on experimental values of the gas in different water bodies, referencing the research results of Wanninkhof R (2014); , , , The meanings are respectively Celsius water temperature, Celsius water temperature quadratic term, Celsius water temperature cubic term, and Celsius water temperature quartic term.
[0068] Optionally, in one embodiment of the present invention, the sea-air carbon dioxide flux in the nearshore waters of the case area is calculated based on the dissolved inorganic carbon in the target water body, the target carbon dioxide partial pressure, and preset meteorological data. This includes: back-calculating the water carbon dioxide partial pressure in the carbon dioxide partial pressure based on the dissolved inorganic carbon in the target water body and the measured pH value; and calculating the sea-air carbon dioxide flux in the nearshore waters of the case area based on the water carbon dioxide partial pressure, the atmospheric carbon dioxide partial pressure, and the sea surface wind speed.
[0069] For example, embodiments of the present invention can calculate the air-sea CO2 flux in a case study area. The case study area can be the nearshore waters of the Guangdong-Hong Kong-Macao Greater Bay Area. Monthly routine monitoring data (surface water sampling) of 76 stations in 2023 for six environmental factors (water temperature, salinity, nutrient concentration, etc.) in the case study area are input into a constructed random forest model to estimate the DIC concentration in the nearshore waters of the case study area. Then, combined with measured pH data, the CO2SYS model is used to back-calculate the concentration. p CO2water (partial pressure of carbon dioxide in water), introduced in the embodiments of this invention p CO2air (atmospheric carbon dioxide partial pressure) observation data and sea surface wind speed were used to calculate the results based on the air-sea interface gas exchange dynamics model. F CO2 (ocean-air CO2 flux). Further, the IDW (Inverse Distance Weighting) spatial interpolation method was employed to represent the nearshore marine station scale of the case study area. F The CO2 calculation results were interpolated onto a spatially continuous grid, ultimately generating a spatiotemporal distribution map of ocean-atmosphere CO2 flux with a resolution of 300 m × 300 m, as shown below. Figure 9 As shown.
[0070] Optionally, in one embodiment of the present invention, based on the sea-air carbon dioxide flux, the carbon source and sink variation patterns of the nearshore waters of the case area are analyzed using at least one of the following: based on the target carbon dioxide partial pressure and the sea-air carbon dioxide flux, a distribution map of carbon dioxide partial pressure and sea-air carbon dioxide flux that meets the preset high spatiotemporal resolution conditions is generated; based on the distribution map, statistical analysis is performed on a monthly and annual scale to obtain temporal analysis results, and statistical analysis is performed on a spatial scale according to the whole, sub-sea areas, and hotspot areas to obtain spatial analysis results; and the carbon source and sink variation patterns of the nearshore waters of the case area are analyzed based on the temporal and spatial analysis results; the main meteorological and water environment factors affecting the sea-air carbon dioxide flux are identified, and the characteristic importance of the main meteorological and water environment factors is quantitatively analyzed to obtain characteristic analysis results; and the carbon source and sink variation patterns of the nearshore waters of the case area are analyzed based on the characteristic analysis results; the distribution variation characteristics of the sea-air carbon dioxide flux are analyzed based on the salinity and nutrient change gradients to obtain change characteristic analysis results; and the carbon source and sink variation patterns of the nearshore waters of the case area are analyzed based on the change characteristic analysis results.
[0071] It is understood that the distribution maps of carbon dioxide partial pressure and ocean-atmosphere carbon dioxide flux that meet the preset high spatiotemporal resolution conditions in the embodiments of the present invention can be high spatiotemporal resolution maps. p CO2 and F Distribution map of CO2.
[0072] In practical implementation, the embodiments of the present invention can perform high-precision analysis of carbon source and sink patterns in nearshore waters. First, the surface seawater at the monitoring station scale is augmented. p CO2 and F CO2 time series results were interpolated to a spatially continuous grid using the inverse distance weighted method, ultimately generating a high spatiotemporal resolution dataset. p CO2 and FThe distribution map of CO2. On this basis, combined with meteorological and water quality data, the variation law of carbon sources and sinks in the coastal waters of the case area is revealed from multiple dimensions such as spatiotemporal distribution, influencing factors, and gradient characteristics. When analyzing the spatiotemporal distribution, statistical analysis is carried out at monthly and annual time scales, and at the spatial scale of the whole area, sub-sea areas and hotspots; when analyzing the influencing factors, Spearman correlation analysis and random forest method are used to identify the main meteorological and water environmental factors affecting the sea-air CO2 flux and conduct quantitative analysis of feature importance; when analyzing the gradient characteristics, the distribution variation characteristics of the sea-air CO2 flux are analyzed from the variation gradients of salinity and nutrients. According to salinity, the water in the coastal waters is divided into three gradients: fresh water (Sal<2 PSU), brackish water-seawater mixture (2≤Sal≤25 PSU) and seawater (Sal>25 PSU). The comprehensive eutrophication index is used to characterize the nutrient concentration level, and three indicators, namely chemical oxygen demand, inorganic nitrogen and active phosphate, are used to calculate the EI (Eutrophication Index), with the calculation formula as follows:
[0073] Further, the embodiment of the present invention divides the eutrophication status of the coastal waters into four gradients: oligotrophic (EI<1), light eutrophication (1≤EI≤3), moderate eutrophication (3<EI≤9) and severe eutrophication (EI>9).
[0074] The embodiment of the present invention can significantly improve the acquisition capability and refined inversion capability of key scarce carbon indicators such as DIC, p CO2, realize high-precision analysis of carbon source and sink laws in coastal waters with low cost and high efficiency, which can provide a reliable data basis and technical support for coastal water carbon budget accounting, carbon cycle process research and ecological environment management and control. The present invention ensures that through high-precision carbon source and sink law analysis, scientific and technical support is provided for coastal water environment management and climate mitigation policy formulation.
[0075] For example, in the embodiment of the present invention, the high-precision carbon source and sink variation law in coastal waters is revealed from multiple dimensions such as spatiotemporal distribution, influencing factors, and gradient characteristics. The analysis results of spatiotemporal distribution characteristics show that the overall CO2 emission intensity to the atmosphere in the case area is , which is 13 times the defined value of strong source, and has spatiotemporal heterogeneity: temporally, it is high in summer and autumn and low in spring and winter, with the emission intensity reaching the peak in September ; spatially, it presents a pattern that the western sea area is a strong carbon source and the eastern sea area is a weak carbon sink , and the hotspots appear at the estuaries of major polluted rivers such as Humen of the Pearl River and the Shenzhen River.
[0076] The results of influencing factors and feature importance are shown in Figure 10As shown, Spearman correlation analysis reveals... F CO2 and p The strong positive correlation with CO2 (r=0.86) indicates that... p CO2 is controlled F The most direct driving force of CO2. Among the features of random forests... p CO2 scored the highest in importance, contributing 0.456, further validating its dominant role in influencing the distribution characteristics of ocean-atmosphere CO2 flux. p CO2 itself is regulated by the carbonate system and is significantly correlated with pH (pH, r=-0.89), total alkalinity (TA, r=-0.80), and salinity (Sal, r=-0.57), indicating that the chemical balance of water bodies is regulated by the carbonate system. p CO2 indirect effects F CO2. Nutrients and F The positive correlation between CO2 and water quality (r ranges from 0.29 to 0.48) suggests that eutrophication may increase CO2 levels in water bodies by enhancing respiration. p CO2, which in turn indirectly promotes CO2 emissions.
[0077] Salinity and nutrient gradient analysis results are as follows Figure 11 As shown, along the salinity gradient, F CO2 exhibits a significant nonlinear negative correlation (R0). 2 =0.71, p <0.001), through F Piecewise linear fitting of CO2 and salinity was used to find the salinity "inflection point" of 20 PSU, constrained by minimizing the mean square error. This salinity of 20 PSU represents a crucial transition point in the nearshore waters of the case study area, shifting from a CO2 "source" to a weak "sink" or near-equilibrium. Regarding the eutrophication gradient... F CO2 and the eutrophication index show a significant linear positive correlation (R0). 2 =0.35, p When the eutrophication index was <0.01 and the eutrophication index was 2.21, the nearshore waters of the case area showed an inflection point effect, transitioning from a CO2 "source" to a weak "sink." As the degree of eutrophication intensified, the waters in the slightly to moderately eutrophic areas... F The gradual rise in CO2 reflects the possibility that increased nutrients may exacerbate net carbon emissions by promoting respiration in nearshore ecosystems. In severely eutrophic waters, F The fact that CO2 levels reached their peak further illustrates that under high nutrient loads, the mineralization process dominates the carbon cycle, leading to the release of CO2 into the atmosphere.
[0078] Furthermore, Figure 12This is a flowchart illustrating the method for analyzing nearshore marine carbon source and sink patterns based on global-regional transfer learning, as described in an embodiment of the present invention.
[0079] like Figure 12 As shown, it includes the following steps: Step S1201: Global and case-area multi-source data collection and processing.
[0080] Step S1202: Global-scale carbon data-driven model construction and comparison.
[0081] Step S1203: Localization verification and migration application of the global-scale model.
[0082] Step S1204: Calculation of sea-air carbon dioxide flux in the case area.
[0083] Step S1205: High-precision analysis of carbon source and sink patterns in the case area.
[0084] The nearshore marine carbon source and sink pattern analysis method based on global-regional transfer learning proposed in this invention can construct a global-scale data-driven model based on multi-source data collection and fusion, and employ various machine learning algorithms to mine six conventional water quality indicators such as temperature, salinity, and nutrients, as well as DIC (dissolved inorganic carbon). p The complex mapping relationships of scarce carbon indicators such as CO2 (partial pressure of carbon dioxide) are explored. Through model performance evaluation, 10-fold cross-validation, and localized measured data verification, the results are applied to the monitoring of sparse nearshore sea areas. Furthermore, this invention reveals the changing patterns of carbon sources and sinks from multiple dimensions, including spatiotemporal distribution, influencing factors, and gradient characteristics. This invention significantly improves the accuracy of DIC (dissipative carbon dioxide) monitoring. p The ability to acquire and refine key scarce carbon indicators such as CO2 has enabled high-precision analysis of carbon source and sink patterns in nearshore waters at low cost and high efficiency. This provides a reliable data foundation and technical support for nearshore marine carbon budget accounting, carbon cycle research, and ecological environment management. This solves the problems of high cost, scarcity of observational data, and uneven spatiotemporal distribution in in-situ monitoring of nearshore carbon fluxes, which lead to insufficient precision in analyzing carbon source and sink patterns in related technologies.
[0085] Next, referring to the accompanying drawings, we describe the nearshore marine carbon source and sink pattern analysis device based on global-regional transfer learning proposed in an embodiment of the present invention.
[0086] Figure 13 This is a schematic diagram of the structure of the nearshore marine carbon source and sink pattern analysis device based on global-regional transfer learning according to an embodiment of the present invention.
[0087] like Figure 13 As shown, the nearshore marine carbon source and sink pattern analysis device 10 based on global-regional transfer learning includes: a data acquisition module 100, a verification module 200, and an analysis module 300.
[0088] Specifically, the acquisition module 100 is used to acquire global marine chemical data and measured data from case study areas.
[0089] The verification module 200 is used to input global ocean chemical data into a preset global-scale carbon data-driven model, and to perform localized verification of the preset global-scale carbon data-driven model using measured data from the case area, thereby obtaining the verified model. The verified model is then applied to the nearshore waters of the case area to output the partial pressure of dissolved inorganic carbon and target carbon dioxide in the target water body of the nearshore waters of the case area.
[0090] The analysis module 300 is used to calculate the sea-air carbon dioxide flux in the nearshore waters of the case area based on the dissolved inorganic carbon in the target water body, the partial pressure of the target carbon dioxide, and preset meteorological data. Based on the sea-air carbon dioxide flux, it analyzes the carbon source and sink variation patterns in the nearshore waters of the case area using at least one of the spatiotemporal distribution, influencing factors, and gradient characteristics.
[0091] Optionally, in one embodiment of the present invention, the parsing module 300 includes: a generation unit, a first parsing unit, a second parsing unit, and a third parsing unit.
[0092] The generation unit is used to generate a distribution map of carbon dioxide partial pressure and sea-air carbon dioxide flux that meets the preset high spatiotemporal resolution conditions, based on the target carbon dioxide partial pressure and sea-air carbon dioxide flux.
[0093] The first analytical unit is used to perform statistical analysis on a monthly and annual scale based on the distribution map to obtain temporal analysis results, and to perform statistical analysis on a spatial scale according to the whole, sub-sea areas and hotspot areas to obtain spatial analysis results. Based on the temporal and spatial analysis results, the carbon source and sink change patterns of the nearshore sea area of the case area are analyzed.
[0094] The second analysis unit is used to identify the main meteorological and water environmental factors affecting the sea-air carbon dioxide flux, and to perform quantitative analysis of the characteristic importance of the main meteorological and water environmental factors to obtain the characteristic analysis results. Based on the characteristic analysis results, the carbon source and sink change patterns in the nearshore waters of the case area are analyzed.
[0095] The third analysis unit is used to analyze the distribution and variation characteristics of sea-air carbon dioxide flux based on the salinity and nutrient gradient, obtain the variation characteristic analysis results, and analyze the carbon source and sink variation patterns in the nearshore waters of the case area based on the variation characteristic analysis results.
[0096] Optionally, in one embodiment of the present invention, the nearshore marine carbon source and sink pattern analysis device 10 based on global-regional transfer learning further includes: an acquisition module and a construction module.
[0097] The acquisition module is used to acquire temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate from global ocean chemical data before inputting the global ocean chemical data into a preset global-scale carbon data-driven model.
[0098] The module is used to build a pre-defined global-scale carbon data-driven model based on temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate data from global ocean chemistry data.
[0099] Optionally, in one embodiment of the present invention, the acquisition module 100 includes: a first acquisition unit and a second acquisition unit.
[0100] The first acquisition unit is used to acquire at least one of global ship survey data, buoy observation data and preset model output data, and to determine global ocean chemical data based on at least one of global ship survey data, buoy observation data and preset model output data.
[0101] The second data acquisition unit is used to collect daily environmental monitoring data, automatic water quality monitoring data, cruise supplementary monitoring data, and meteorological and hydrological data of the case area, and to determine the measured data of the case area based on the daily environmental monitoring data, automatic water quality monitoring data, cruise supplementary monitoring data, and meteorological and hydrological data of the case area.
[0102] Optionally, in one embodiment of the present invention, the verification module 200 includes a verification unit and a prediction unit.
[0103] The verification unit is used to compare the measured values of dissolved inorganic carbon, total alkalinity and partial pressure of carbon dioxide in the nearshore waters of the case area with the simulated values, generate comparison results, and perform localized verification of the preset global-scale carbon data-driven model based on the comparison results to obtain the verified model.
[0104] The prediction unit is used to transfer the validated model to the nearshore waters of the case area to predict the partial pressure of dissolved inorganic carbon and target carbon dioxide in the target waters of the case area using temperature, salinity, pH, dissolved oxygen, nitrate and phosphate from global ocean chemistry data.
[0105] Optionally, in one embodiment of the present invention, the parsing module 300 includes a reverse calculation unit and a calculation unit.
[0106] The reverse calculation unit is used to reverse calculate the partial pressure of carbon dioxide in the water body based on the dissolved inorganic carbon in the target water body and the measured pH value.
[0107] The calculation unit is used to calculate the sea-air carbon dioxide flux in the nearshore waters of the case area based on the partial pressure of carbon dioxide in water, the partial pressure of carbon dioxide in the atmosphere, and the wind speed at the sea surface.
[0108] Optionally, in one embodiment of the present invention, the formula for calculating the sea-air carbon dioxide flux is:
[0109] in, This refers to the sea-atmosphere carbon dioxide flux. The carbon dioxide transport rate at the air-sea interface. This represents the solubility of carbon dioxide in water. This refers to the partial pressure of carbon dioxide in water. This is the partial pressure of atmospheric carbon dioxide.
[0110] It should be noted that the foregoing explanation of the embodiment of the nearshore marine carbon source and sink pattern analysis method based on global-regional transfer learning also applies to the nearshore marine carbon source and sink pattern analysis device based on global-regional transfer learning in this embodiment, and will not be repeated here.
[0111] The nearshore marine carbon source-sink pattern analysis device based on global-regional transfer learning proposed in this invention can integrate multi-source information such as global marine chemistry and measured data from case study regions. It uncovers the complex mapping relationship between readily available and scarce carbon indicators, achieving high-resolution prediction of carbon components, rapid estimation of air-sea CO2 fluxes, and high-precision analysis of nearshore marine carbon source-sink patterns. This provides technical support for the formulation of marine carbon neutrality strategies and the management of nearshore marine ecological environment. Therefore, it solves the problems in related technologies, such as high cost of air-sea carbon flux observation, sparse in-situ data, and uneven spatiotemporal distribution leading to insufficient analysis of nearshore marine carbon source-sink patterns.
[0112] Figure 14 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. The electronic device may include: The memory 1401, the processor 1402, and the computer program stored on the memory 1401 and executable on the processor 1402.
[0113] When the processor 1402 executes the program, it implements the nearshore marine carbon source and sink pattern analysis method based on global-regional transfer learning provided in the above embodiments.
[0114] Furthermore, electronic devices also include: Communication interface 1403 is used for communication between memory 1401 and processor 1402.
[0115] The memory 1401 is used to store computer programs that can run on the processor 1402.
[0116] The memory 1401 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage.
[0117] If the memory 1401, processor 1402, and communication interface 1403 are implemented independently, then the communication interface 1403, memory 1401, and processor 1402 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be divided into address buses, data buses, control buses, etc. For ease of representation, Figure 14 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.
[0118] Optionally, in a specific implementation, if the memory 1401, processor 1402, and communication interface 1403 are integrated on a single chip, then the memory 1401, processor 1402, and communication interface 1403 can communicate with each other through an internal interface.
[0119] The processor 1402 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention.
[0120] This embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for analyzing nearshore marine carbon source-sink patterns based on global-regional transfer learning.
[0121] This invention also provides a computer program product storing a computer program that, when executed by a processor, implements the above-described method for analyzing nearshore marine carbon source-sink patterns based on global-regional transfer learning.
[0122] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.
[0123] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.
[0124] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.
[0125] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.
[0126] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any of the following techniques known in the art, or a combination thereof: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.
[0127] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.
[0128] Furthermore, the functional units in the various embodiments of the present invention can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.
[0129] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.
Claims
1. A method for analyzing nearshore marine carbon source-sink patterns based on global-regional transfer learning, characterized in that, Includes the following steps: Collect global marine chemical data and field-measured data from case study areas; The global ocean chemical data is input into a preset global-scale carbon data-driven model, and the preset global-scale carbon data-driven model is localized and validated using measured data from the case area to obtain a validated model. The validated model is then applied to the nearshore waters of the case area to output the partial pressures of dissolved inorganic carbon and carbon dioxide in the target water body of the nearshore waters of the case area. Based on the dissolved inorganic carbon in the target water body, the partial pressure of the target carbon dioxide, and preset meteorological data, the sea-air carbon dioxide flux in the nearshore waters of the case area is calculated. Based on the sea-air carbon dioxide flux, the carbon source and sink variation patterns in the nearshore waters of the case area are analyzed using at least one of the spatiotemporal distribution, influencing factors, and gradient characteristics. The step of analyzing the carbon source and sink variation patterns in the nearshore waters of the case area based on the sea-air carbon dioxide flux using at least one of spatiotemporal distribution, influencing factors, and gradient characteristics includes: Based on the target carbon dioxide partial pressure and the sea-atmosphere carbon dioxide flux, a distribution map of carbon dioxide partial pressure and sea-atmosphere carbon dioxide flux that meets the preset high spatiotemporal resolution conditions is generated. Based on the distribution map, statistical analysis was performed on a monthly and annual scale to obtain temporal analysis results. Statistical analysis was also performed spatially, categorized by overall area, sub-sea areas, and hotspot areas to obtain spatial analysis results. The carbon source and sink variation patterns in the nearshore waters of the case area were then analyzed based on the temporal and spatial analysis results. Identify the main meteorological and aquatic environmental factors affecting the sea-air carbon dioxide flux, and perform quantitative analysis of the characteristic importance of the main meteorological and aquatic environmental factors to obtain the characteristic analysis results. Based on the characteristic analysis results, analyze the carbon source and sink variation patterns of the nearshore sea area in the case area. The distribution and variation characteristics of the sea-air carbon dioxide flux were analyzed based on the salinity and nutrient gradients, and the results of the variation characteristic analysis were obtained. Based on the results of the variation characteristic analysis, the carbon source and sink variation patterns of the nearshore waters of the case area were analyzed.
2. The method for analyzing nearshore marine carbon source and sink patterns based on global-regional transfer learning according to claim 1, characterized in that, Before inputting the global ocean chemical data into a pre-defined global-scale carbon data-driven model, the process also includes: Acquire temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate from the global ocean chemical data; Based on the temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate data from the global ocean chemical data, the preset global-scale carbon data-driven model is constructed.
3. The method for analyzing nearshore marine carbon source and sink patterns based on global-regional transfer learning according to claim 1, characterized in that, The collected global marine chemical data and case-area measured data include: Collect at least one of global ship survey data, buoy observation data, and preset model output data, and determine the global ocean chemical data based on at least one of the global ship survey data, the buoy observation data, and the preset model output data; Daily environmental monitoring data, automatic water quality monitoring data, cruise supplementary monitoring data, and meteorological and hydrological data of the case area are collected, and the measured data of the case area are determined based on the daily environmental monitoring data, the automatic water quality monitoring data, the cruise supplementary monitoring data, and the meteorological and hydrological data of the case area.
4. The method for analyzing nearshore marine carbon source and sink patterns based on global-regional transfer learning according to claim 1, characterized in that, The process involves using measured data from the case study area to localize and validate the preset global-scale carbon data-driven model, obtaining a validated model, and then applying the validated model to the nearshore waters of the case study area to output the partial pressures of dissolved inorganic carbon and carbon dioxide in the target water bodies of the nearshore waters of the case study area, including: The measured values of dissolved inorganic carbon, total alkalinity, and partial pressure of carbon dioxide in the nearshore waters of the case area are compared with the simulated values to generate comparison results. Based on the comparison results, the preset global-scale carbon data-driven model is localized and validated to obtain the validated model. The validated model was then applied to the nearshore waters of the case area to predict the partial pressure of dissolved inorganic carbon and target carbon dioxide in the target waters of the case area using temperature, salinity, pH, dissolved oxygen, nitrate, and phosphate from the global ocean chemical data.
5. The method for analyzing nearshore marine carbon source and sink patterns based on global-regional transfer learning according to claim 4, characterized in that, The calculation of the sea-atmosphere carbon dioxide flux in the nearshore waters of the case area based on the dissolved inorganic carbon in the target water body, the target carbon dioxide partial pressure, and preset meteorological data includes: Based on the dissolved inorganic carbon in the target water body and the measured pH value, the partial pressure of carbon dioxide in the water body is calculated by reverse calculation. Based on the partial pressure of carbon dioxide in water, the partial pressure of carbon dioxide in the atmosphere, and the wind speed at the sea surface, the sea-atmosphere carbon dioxide flux in the nearshore waters of the case area is calculated.
6. The method for analyzing nearshore marine carbon source and sink patterns based on global-regional transfer learning according to claim 5, characterized in that, The formula for calculating the air-sea carbon dioxide flux is: in, This refers to the sea-air carbon dioxide flux. The carbon dioxide transport rate at the air-sea interface. This represents the solubility of carbon dioxide in water. The partial pressure of carbon dioxide in the water body. The partial pressure of atmospheric carbon dioxide is given.
7. A device for analyzing nearshore marine carbon source and sink patterns based on global-regional transfer learning, employing the nearshore marine carbon source and sink pattern analysis method based on global-regional transfer learning as described in any one of claims 1-6, characterized in that, include: The data acquisition module is used to collect global marine chemical data and field-measured data from case study areas; The verification module is used to input the global ocean chemical data into a preset global-scale carbon data-driven model, and to perform localized verification of the preset global-scale carbon data-driven model using the measured data of the case area to obtain the verified model. The verified model is then applied to the nearshore waters of the case area to output the partial pressure of dissolved inorganic carbon and target carbon dioxide in the target water body of the nearshore waters of the case area. The analysis module is used to calculate the sea-air carbon dioxide flux in the nearshore waters of the case area based on the dissolved inorganic carbon in the target water body, the partial pressure of the target carbon dioxide, and preset meteorological data, and to analyze the carbon source and sink variation patterns in the nearshore waters of the case area based on the sea-air carbon dioxide flux using at least one of the spatiotemporal distribution, influencing factors, and gradient characteristics. The step of analyzing the carbon source and sink variation patterns in the nearshore waters of the case area based on the sea-air carbon dioxide flux using at least one of spatiotemporal distribution, influencing factors, and gradient characteristics includes: Based on the target carbon dioxide partial pressure and the sea-atmosphere carbon dioxide flux, a distribution map of carbon dioxide partial pressure and sea-atmosphere carbon dioxide flux that meets the preset high spatiotemporal resolution conditions is generated. Based on the distribution map, statistical analysis was performed on a monthly and annual scale to obtain temporal analysis results. Statistical analysis was also performed spatially, categorized by overall area, sub-sea areas, and hotspot areas to obtain spatial analysis results. The carbon source and sink variation patterns in the nearshore waters of the case area were then analyzed based on the temporal and spatial analysis results. Identify the main meteorological and aquatic environmental factors affecting the sea-air carbon dioxide flux, and perform quantitative analysis of the characteristic importance of the main meteorological and aquatic environmental factors to obtain the characteristic analysis results. Based on the characteristic analysis results, analyze the carbon source and sink variation patterns of the nearshore sea area in the case area. The distribution and variation characteristics of the sea-air carbon dioxide flux were analyzed based on the salinity and nutrient gradients, and the results of the variation characteristic analysis were obtained. Based on the results of the variation characteristic analysis, the carbon source and sink variation patterns of the nearshore waters of the case area were analyzed.
8. An electronic device, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for analyzing nearshore marine carbon source and sink patterns based on global-regional transfer learning as described in any one of claims 1-6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the nearshore marine carbon source-sink pattern analysis method based on global-regional transfer learning as described in any one of claims 1-6.
Citation Information
Patent Citations
Method for evaluating carbon sink increase potential of offshore area
CN118917559A
KR20260097289A
