A method for inverting chlorophyll-a concentration in estuaries based on multi-model fusion

Abstract

The present invention relates to a method for inverting chlorophyll a concentration in an estuary based on multi-model fusion, and relates to the technical field of water quality monitoring. ESTARFM is used to fuse MODIS remote sensing images and Landsat 8 remote sensing images to obtain high-temporal and spatial resolution remote sensing images, thereby solving the problem that existing remote sensing images cannot simultaneously achieve high-temporal and spatial resolution, resulting in poor accuracy in inverting chlorophyll a concentration in the estuary. The reflectivity of the band combination screened out from the high-temporal and spatial resolution remote sensing images is used as input, and the corresponding monitored chlorophyll a concentration is used as output to construct a random forest model and a GBR model, respectively. A TPE search weight method is used to establish a fusion model of the random forest model and the GBR model, thereby solving the problem that a single machine learning model in the existing technology cannot adapt to the complex environmental conditions of the estuary, resulting in low accuracy in inverting chlorophyll a concentration in the estuary.

Description

Technical Field

[0001] The present invention belongs to the technical field of water quality monitoring and relates to a chlorophyll a concentration inversion method for an estuary based on multi-model fusion. Background Art

[0002] In recent years, with economic development and population growth, estuaries have become increasingly eutrophic, severely impacting human habitability and sustainable development. Chlorophyll a concentration, as a key indicator of water eutrophication, can directly reflect the degree of eutrophication and assess water health. Therefore, inversion of chlorophyll a concentration in estuaries is crucial. There are generally two methods for inverting chlorophyll a concentration: measurement, which uses field sampling and laboratory analysis to obtain accurate water quality data. While this method provides more accurate results, it is time-consuming and labor-intensive for large-scale water quality monitoring. Remote sensing inversion, on the other hand, uses observational data and mathematical models to infer the distribution of chlorophyll a in water. This method is time-efficient and widely used. Commonly used inversion methods include empirical, analytical, semi-empirical, and machine learning.

[0003] However, existing remote sensing imagery cannot simultaneously achieve high spatial and temporal resolution. Low spatial resolution can lead to biased chlorophyll a retrieval results, while high spatial resolution remote sensing imagery often comes with low temporal resolution. This means it takes over ten days or even a month to acquire a single remote sensing image, and most images are subject to interference from weather conditions such as clouds and rain, resulting in a limited number of usable remote sensing images. Furthermore, existing technologies use machine learning models to simulate the relationship between remote sensing data and chlorophyll a concentration for retrieval. However, using a single machine learning model to accurately retrieve chlorophyll a concentration from estuarine remote sensing data is difficult, primarily due to the heterogeneity and complexity of estuarine data, including the influence of various environmental factors. Chlorophyll a concentration and remote sensing reflectance often exhibit a nonlinear relationship, which a single model cannot fully capture. Furthermore, issues such as inefficiency in feature selection and extraction, overfitting, and differences in feature distribution caused by spatiotemporal variations all affect prediction accuracy. The lack of flexibility and adaptability of a single model often makes it difficult to meet the needs of estuarine data analysis. Summary of the Invention

[0004] The present invention provides an estuarine chlorophyll-a concentration inversion method based on multi-model fusion. The method aims to use ESTARFM to fuse pre-processed MODIS and Landsat remote sensing images to obtain remote sensing images with high temporal and spatial resolution, so as to solve the problem that existing remote sensing images cannot simultaneously achieve high temporal and spatial resolution. The reflectivity of the band combination screened out from the remote sensing image is used as input, and the corresponding monitored chlorophyll-a concentration is used as output to construct a random forest model and a GBR model respectively. The TPE search weight method is used to establish a fusion model of the random forest model and the GBR model, so as to solve the problem that the single machine learning model in the existing technology has poor inversion accuracy for estuarine chlorophyll-a.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] This application provides a method for inverting chlorophyll-a concentration in estuaries based on multi-model fusion, including the following steps:

[0007] S1. Acquire estuary monitoring data for a specific time period, wherein the estuary monitoring data includes chlorophyll a concentration and MODIS remote sensing imagery and Landsat 8 remote sensing imagery data for the corresponding time period;

[0008] S2. Use ESTARFM to fuse MODIS remote sensing images and Landsat8 remote sensing images to obtain high temporal and spatial resolution remote sensing images;

[0009] S3, obtaining the reflectance of different band combinations of high temporal and spatial resolution remote sensing images, and calculating the Pearson correlation coefficient between the reflectance and chlorophyll a concentration;

[0010] S4. Determine the band combination with significant correlation between reflectance and chlorophyll a concentration based on the Pearson correlation coefficient, and record it as a highly correlated band combination;

[0011] S5. Using the reflectance of the highly correlated band combination as an explanatory variable and the chlorophyll a concentration as a response variable, a random forest model and a GBR model are established respectively;

[0012] S6. Use TPE search weight method to establish a fusion inversion model of random forest model and GBR model;

[0013] S7. Obtain high temporal and spatial resolution remote sensing images of the estuary to be monitored, input the reflectance of the highly correlated band combination into the fusion inversion model, and invert the chlorophyll a concentration of the estuary to be monitored.

[0014] Furthermore, the MODIS remote sensing image is downloaded from the NASA official website and preprocessed, including:

[0015] Screening of remote sensing images: Download MODIS remote sensing images from NASA's official website and select remote sensing images with little cloud cover and no more than 7 days from the monitoring time;

[0016] Remote sensing image radiometric calibration and geometric correction: Download the MODIS processing plug-in MTCK in ENVI, then import MODIS remote sensing data in MTCK and select the WGS-84 coordinate system;

[0017] Water body extraction: The estuary water area was extracted based on the NDWI water body index, and band operations were performed using Band math in ENVI.

[0018] Furthermore, the NDWI water index is calculated as follows:

[0019]

[0020] Where Green represents the green band of the MODIS remote sensing image, which corresponds to the fourth band among the 36 bands; NIR represents the near-infrared band of the MODIS remote sensing image, which corresponds to the second band among the 36 bands; NDWI is the water body index.

[0021] Furthermore, the Landsat remote sensing image is acquired in the geospatial data cloud and pre-processed, including:

[0022] Screening of remote sensing images: Download Landsat8 remote sensing images from the Geospatial Data Cloud and select remote sensing images with minimal cloud cover and the closest time to the MODIS remote sensing images;

[0023] Remote sensing image radiometric calibration: converting the DN value recorded by the sensor into a radiometric brightness value. The DN value is the grayscale value of the recorded object.

[0024] Atmospheric correction of remote sensing images: Perform atmospheric correction on Landsat8 remote sensing images using the FLAASH Atmospheric Correction tool in ENVI to eliminate the images produced by electromagnetic waves passing through the atmosphere and the influence of atmospheric substances on the reflection of ground objects.

[0025] Water body extraction: The estuary water area was extracted based on the NDWI water body index, and band operations were performed using Band math in ENVI.

[0026] Furthermore, in step S2, the use of ESTARFM to fuse the MODIS remote sensing image and the Landsat8 remote sensing image to obtain a high temporal and spatial resolution remote sensing image includes the following steps:

[0027] Select reference images: select Landsat 8 remote sensing images as high-resolution images and MODIS remote sensing images as low-resolution images;

[0028] By constructing a spatiotemporal model, the spectral information of MODIS remote sensing images and the spatial information of Landsat 8 remote sensing images are fused;

[0029] Generate high temporal and spatial resolution images that retain the high spatial resolution of Landsat 8 remote sensing images while introducing the timeliness of MODIS remote sensing images.

[0030] Furthermore, in step S3, the Pearson correlation coefficient is calculated using the following formula:

[0031]

[0032] Where p represents the Pearson correlation coefficient between chlorophyll a concentration and reflectance of different band combinations of high temporal and spatial resolution remote sensing images; R(x i ) represents the chlorophyll a concentration of the i-th sample value; R(y i ) represents the i-th sample value of the reflectance of different band combinations of high temporal and spatial resolution remote sensing images; R(x) is the average value of all chlorophyll a concentration samples; R(y) represents the average value of all reflectance samples of different band combinations of high temporal and spatial resolution remote sensing images.

[0033] Furthermore, in step S5, the random forest model includes the following construction steps:

[0034] Data preparation: Collect reflectance data and corresponding chlorophyll a concentration data of different band combinations of high temporal and spatial resolution remote sensing images; divide the dataset into training and test sets;

[0035] Parameter setting: determine the number of decision trees in the random forest, set the maximum depth of the decision tree, and determine the minimum number of sample splits;

[0036] Model training: The random forest model is trained using the training set data, with the reflectance data of highly correlated band combinations as input features and the chlorophyll-a concentration data as the target variable. During the training process, multiple decision trees are constructed based on the set parameters. Each decision tree grows on a randomly selected sample set and randomly selected features to learn the relationship between reflectance and chlorophyll-a concentration.

[0037] Model evaluation: The reflectance data in the test set is passed into the model to obtain the chlorophyll a concentration value predicted by the model; evaluation indicators are calculated, including the mean square error, mean absolute error, and determination coefficient.

[0038] Furthermore, in step S5, the GBR model includes the following steps:

[0039] Data preparation: Collect reflectance data of different band combinations of high temporal and spatial resolution remote sensing images and the corresponding chlorophyll a concentration data; divide the dataset into training and test sets;

[0040] Parameter initialization: Determine the initial values ​​of model parameters, including learning rate, number of weak learners, maximum depth of decision tree, and minimum number of sample splits;

[0041] Model training: The GBR model is trained using the training set data, with reflectance data of highly correlated band combinations as input features and chlorophyll a concentration data as the target variable to construct the GBR model. In each iteration, the model calculates the gradient of the loss function based on the current prediction error. Based on the gradient information, a new decision tree is trained as a weak learner to reduce the loss function. The new decision tree is added to the GBR model and works together with the previous weak learner to gradually improve the model's performance. The model training process is repeated until the preset number of iterations is reached or the stopping condition is met.

[0042] Model evaluation: Use the test set data to evaluate the trained GBR model.

[0043] Furthermore, in step S6, the TPE search weight method is used to establish a fusion inversion model of the random forest model and the GBR model, including the following steps:

[0044] Define an objective function: Define an objective function that receives a set of parameters, including weights and hyperparameters for the random forest model and the GBR model. Within the objective function, use the given weights to fuse the predictions of the random forest model and the GBR model. Train the random forest model and the GBR model using the training set and evaluate the performance of the fused model on the test set. Return the value of the evaluation metric as the output of the objective function.

[0045] Use TPE for parameter search: Define the parameter space in Python, which includes the weights of the random forest model and the GBR model and the range of hyperparameters; use the fmin function of hyperopt for parameter search, specify the search algorithm as TPE, and set the maximum number of evaluations;

[0046] Construct a fusion inversion model: Based on the optimal weight parameters obtained by the search, construct a fusion inversion model; retrain the random forest model and GBR model using the training set; during prediction, use the optimal weight parameters to fuse the prediction results of the two models;

[0047] Model evaluation and application: Use test set data to evaluate the fusion inversion model and analyze the model performance based on the evaluation results.

[0048] Furthermore, the hyperparameters include the number of decision trees in the random forest, the maximum depth of the decision tree, the minimum number of samples required for the decision tree to split, the minimum number of samples required for leaf nodes, the learning rate in the GBR model, the number of weak learners, the maximum depth of the decision tree, and the proportion of samples used in each iteration.

[0049] Furthermore, the prediction results of the random forest model and the GBR model are fused using given weights, specifically using a weighted average method.

[0050] Beneficial effects of the present invention:

[0051] (1) Using ESTARFM to fuse MODIS remote sensing images and Landsat8 remote sensing images to obtain high temporal and spatial resolution remote sensing images, this solves the problem that existing remote sensing images cannot achieve high temporal and spatial resolution at the same time, resulting in poor accuracy in inverting chlorophyll a concentration in estuaries;

[0052] (2) The reflectivity of the band combination screened out from the high temporal and spatial resolution remote sensing image is used as input, and the corresponding monitored chlorophyll a concentration is used as output to construct a random forest model and a GBR model respectively; the TPE search weight method is used to establish a fusion model of the random forest model and the GBR model to solve the problem that the existing single machine learning model cannot adapt to the complex environmental conditions of the estuary, resulting in low accuracy in the inversion of chlorophyll a concentration in the estuary. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] To facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0054] Figure 1 This is a flow chart of a method for inverting chlorophyll-a concentration in an estuary based on multi-model fusion in the present invention.

[0055] Figure 2 The flowchart of obtaining high temporal and spatial resolution remote sensing images in one embodiment of the present invention.

[0056] Figure 3 The figure is a flow chart of constructing a fusion inversion model in one embodiment of the present invention. DETAILED DESCRIPTION

[0057] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0058] See also Figure 1-Figure 3 , this application provides a method for inverting chlorophyll-a concentration in an estuary based on multi-model fusion, comprising the following steps:

[0059] S1. Acquire estuary monitoring data for a specific time period, wherein the estuary monitoring data includes chlorophyll a concentration and MODIS remote sensing imagery and Landsat 8 remote sensing imagery data for the corresponding time period;

[0060] In this embodiment, the specific time period is at least one year or more, and the chlorophyll a concentration is obtained from an estuary water quality monitoring station. To improve the accuracy and quality of the data, the data needs to be preprocessed, including outlier processing and data normalization. MODIS remote sensing images are downloaded from the NASA official website, and the MODIS remote sensing images are preprocessed by radiometric calibration, geometric correction, and cropping, including the following steps:

[0061] (1) Screening of remote sensing images: Download MODIS remote sensing images from NASA's official website and select remote sensing images with little cloud cover and no more than 7 days from the monitoring time; you can directly check the cloud cover situation on NASA's official website and select the appropriate image to download.

[0062] (2) Radiometric calibration and geometric correction of remote sensing images: Download the MODIS processing plug-in MTCK in ENVI, then import MODIS remote sensing data into MTCK. The parameter settings are as follows:

[0063] Output Type: Select Strict Georeferencing;

[0064] Processing Options: Select Remote Sensing Reflectance;

[0065] Resampling method: Select the nearest neighbor method to resample and try to retain the original data;

[0066] Use double-precision format data: If you select yes, the reflectivity data will have more decimal places;

[0067] Coordinate System: Select the WGS-84 coordinate system to facilitate the subsequent determination of the reflectance corresponding to the chlorophyll concentration at the monitoring station. After preprocessing, the pixel represents the reflectance.

[0068] (3) Water body extraction: After completing the above steps, the required remote sensing image is obtained. Then it is necessary to eliminate non-water body areas. The estuary water area can be extracted based on the NDWI water body index. This step requires band calculation in Bandmath in ENVI. The calculation formula is as follows:

[0069]

[0070] Where Green represents the green band of the MODIS remote sensing image, which corresponds to the fourth band among the 36 bands; NIR represents the near-infrared band of the MODIS remote sensing image, which corresponds to the second band among the 36 bands; NDWI is the water body index.

[0071] Landsat data is acquired from the geospatial data cloud, and remote sensing images are pre-processed by radiometric calibration, atmospheric correction, and cropping, including the following steps:

[0072] (1) Remote sensing image selection: Download Landsat 8 remote sensing images from the Geospatial Data Cloud and select images with minimal cloud cover and closest to the MODIS timeframe. Most data can be viewed directly for cloud cover, while a small portion requires downloading and viewing in ENVI.

[0073] (2) Remote sensing image radiometric calibration: This step mainly converts the DN values ​​recorded by the sensor into radiometric brightness values. The DN value is the grayscale value of the recorded ground object. This is done in the Radiometric Calibration tool of ENVI.

[0074] (3) Atmospheric correction of remote sensing images: This step is mainly to eliminate the images produced by electromagnetic waves passing through the atmosphere and the influence of atmospheric substances on the reflection of ground objects. It is performed in the FLAASH Atmospheric Correction tool in ENVI.

[0075] (4) Water body extraction: The water body method of Landsat8 is the same as that of MODIS. Band math in ENVI is used to perform band operations. Due to the different satellites used, the green band and near-infrared band represent the 2nd and 4th bands in Landsat8, respectively.

[0076] S2. Use ESTARFM to fuse MODIS remote sensing images and Landsat8 remote sensing images to obtain high temporal and spatial resolution remote sensing images;

[0077] In this embodiment, ESTARFM (Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model) is used to fuse MODIS and Landsat 8 remote sensing images to obtain remote sensing images with high temporal and spatial resolution. This process is implemented in the LDL workbench in ENVI. Before fusion, the remote sensing image data of MODIS and Landsat 8 must be prepared to ensure that the temporal and spatial ranges of the two data sets overlap. At the same time, pay attention to the format and band of the data. Landsat 8 images usually have a higher spatial resolution (30 meters), while MODIS images provide a higher temporal resolution (once a day) and a spatial resolution of 1000 meters. The ESTARFM model requires the calculation of some basic parameters, including:

[0078] Spectral Reflectance: Extract spectral reflectance data from MODIS and Landsat 8.

[0079] Time Difference: Calculate the difference in acquisition time between MODIS and Landsat 8 images so as to interpolate the time when fusion is performed.

[0080] When performing ESTARFM fusion, follow these steps:

[0081] (1) Select reference images: Select Landsat8 as the high-resolution image and MODIS as the low-resolution image.

[0082] (2) By constructing a spatiotemporal model, the spectral information of MODIS and the spatial information of Landsat 8 are integrated.

[0083] (3) Generate high temporal and spatial resolution images: The generated images will retain the high spatial resolution of Landsat 8 while introducing the timeliness of MODIS.

[0084] S3, obtaining the reflectance of different band combinations of high temporal and spatial resolution remote sensing images, and calculating the Pearson correlation coefficient between the reflectance and chlorophyll a concentration;

[0085] Obtaining the reflectance of different band combinations of high-temporal-resolution remote sensing images and calculating the Pearson correlation coefficient between these values ​​and chlorophyll-a concentration is crucial for studying aquatic environments. Data preparation begins with collecting high-temporal-resolution remote sensing imagery and chlorophyll-a concentration data for the corresponding areas. The remote sensing images are then preprocessed, including radiometric correction to eliminate sensor and atmospheric effects, obtain accurate reflectance values, and geometric correction to ensure alignment with the actual geographic location. Different band combinations are then selected based on the research objectives, and the reflectance for each band combination is calculated. The reflectance data and chlorophyll-a concentration data are then collated to ensure a one-to-one correspondence. The Pearson correlation coefficient is calculated using statistical software or a programming language. The formula comprehensively considers the relationship between individual data points and the average value. The correlation coefficient ranges from -1 to 1, with values ​​close to 1 indicating a positive correlation, -1 indicating a negative correlation, and 0 indicating no linear relationship. This coefficient can be used to determine the degree of correlation between reflectance and chlorophyll-a concentration, providing a strong basis for subsequent research and applications, and contributing to a better understanding of aquatic ecosystems.

[0086] Furthermore, the Pearson correlation coefficient is calculated as follows:

[0087]

[0088] Where p represents the Pearson correlation coefficient between chlorophyll a concentration and reflectance of different band combinations of high temporal and spatial resolution remote sensing images, R(x i ) represents the chlorophyll a concentration of the i-th sample value; R(y i ) represents the i-th sample value of the reflectance of different band combinations of high temporal and spatial resolution remote sensing images, R(x) is the average value of all chlorophyll a concentration samples; R(y) represents the average value of all reflectance samples of different band combinations of high temporal and spatial resolution remote sensing images.

[0089] S4. Determine the band combination with significant correlation between reflectance and chlorophyll a concentration based on the Pearson correlation coefficient, and record it as a highly correlated band combination;

[0090] Generally speaking, when the absolute value of the Pearson correlation coefficient reaches 0.7 or above, it can be considered to be highly correlated. A correlation coefficient between 0.7 and 1 indicates that there is a strong positive correlation between the variables; between -0.7 and -1, it indicates that there is a strong negative correlation. However, the definition of "high correlation" is not absolute. In different specific application scenarios, the threshold value of the correlation coefficient may be adjusted according to the actual situation. In this embodiment, the enumeration method is used in python to combine the bands in the chlorophyll a inversion data, and create two-band, three-band, and four-band combinations respectively, and a total of more than 2w combinations are obtained. Pearson correlation analysis is then performed in SPSS, and it is found that the correlation between the three-band and four-band combinations and chlorophyll a is higher than that between the single-band and two-band combinations. Therefore, the three-band and four-band combinations with high correlation are selected for the subsequent establishment of chlorophyll a concentration inversion.

[0091] S5. Using the reflectance of the highly correlated band combination as an explanatory variable and the chlorophyll a concentration as a response variable, a random forest model and a GBR model are established respectively;

[0092] In this embodiment, the random forest model plays an important role in the study of the relationship between the reflectance of the highly correlated band combination and the chlorophyll a concentration. The random forest is essentially an integrated learning model composed of many decision trees. In the data preparation stage, we use the reflectance of the highly correlated band combination as the explanatory variable and the chlorophyll a concentration as the response variable, and divide the data set into a training set and a test set. When constructing a random forest, multiple sample sets are first extracted from the original data by random sampling, and each sample set is used to construct a decision tree. In addition, in the process of constructing each decision tree, a part of the features are randomly selected from the numerous features for splitting, so that each decision tree has unique randomness and diversity. Finally, the prediction results of these decision trees are combined, usually by taking an average or majority vote to obtain the final prediction result.

[0093] The random forest model offers many significant advantages. First, it is highly resistant to overfitting. Because it is composed of multiple decision trees, even if a single decision tree may overfit, the combined results of multiple trees can greatly improve the model's generalization ability. Second, it is insensitive to outliers and noise. When dealing with data containing outliers or noise, random forests are relatively stable and are not easily affected by these interfering factors. Third, it can handle high-dimensional data well. For data with a large number of features, such as the reflectance of multiple highly correlated bands here, random forests can effectively handle it and automatically perform feature selection to identify the most important features for predicting chlorophyll a concentration. Fourth, it is highly interpretable. Although the random forest model is relatively complex, we can understand the contribution of each feature to the prediction by calculating feature importance.

[0094] In practical applications, data preparation is first performed, followed by determining the parameters of the random forest model, such as the number of trees, maximum depth, and minimum number of sample splits. The choice of these parameters affects model performance and can be optimized through methods such as cross-validation. The random forest model is then trained using the training set data, allowing it to learn the relationship between reflectance and chlorophyll a concentration. The performance of the random forest model is then evaluated using the test set data. Common evaluation metrics include mean squared error, mean absolute error, and coefficient of determination. Based on the evaluation results, further parameter adjustments or feature engineering can be performed to improve model performance. Finally, the optimized and evaluated random forest model can be used to predict chlorophyll a concentration in new remote sensing imagery data.

[0095] Furthermore, in step S5, the random forest model includes the following construction steps:

[0096] (1) Data preparation

[0097] Collect reflectance data and corresponding chlorophyll-a concentration data for different band combinations of high-temporal and spatial resolution remote sensing images. Ensure data accuracy and completeness, and cover different environmental conditions and regional ranges.

[0098] Divide the dataset into a training set and a test set. This is typically done in a certain ratio, such as 80% for training and 20% for testing. This ensures that the model can both learn the patterns in the data and be effectively validated on unseen data during subsequent model training and evaluation.

[0099] (2) Parameter setting

[0100] Determine the number of decision trees in the random forest (n_estimators). Generally speaking, a larger number of trees may improve model performance, but the computational cost will also increase. You can try starting with a small number, such as 100, and then gradually adjust the number based on the model's performance.

[0101] Set the maximum depth of the decision tree (max_depth). Limiting the depth of the tree can prevent overfitting, but if the depth is too small, it may cause the model to underfit. You can use methods such as cross-validation to determine an appropriate maximum depth value.

[0102] Determine the minimum number of samples to split (min_samples_split). This parameter controls the minimum number of samples required for the decision tree to split. If it is set too small, it may lead to overfitting; if it is set too large, it may prevent the tree from growing fully, affecting the performance of the model.

[0103] Other parameters can also be considered, such as the minimum number of leaf node samples (min_samples_leaf), feature selection method, etc.

[0104] (3) Model training

[0105] Import random forest related libraries, such as RandomForestRegressor in scikit-learn.

[0106] The random forest model is trained using the training set data. The reflectance data for the highly correlated band combinations is used as the input features, and the chlorophyll a concentration data is used as the target variable. These are passed to the model's fit method.

[0107] During the model training process, multiple decision trees are constructed based on the set parameters. Each decision tree is grown on a randomly selected set of samples and randomly selected features to learn the relationship between reflectance and chlorophyll a concentration.

[0108] (4) Model evaluation

[0109] Use the test set data to evaluate the trained random forest model. Pass the reflectance data in the test set to the model's predict method to obtain the chlorophyll a concentration value predicted by the model.

[0110] Calculate the evaluation indicators, the common ones are mean square error (MSE), mean absolute error (MAE) and coefficient of determination (R 2 ) etc. The mean square error and mean absolute error measure the error between the predicted value and the true value, while the coefficient of determination indicates how well the model fits the data.

[0111] Analyze the model's performance based on the evaluation results. If the model performance is not ideal, adjust the parameters, retrain and reevaluate, or perform feature engineering, such as selecting a more appropriate band combination or performing data standardization.

[0112] (5) Application and optimization

[0113] The optimized random forest model can be used to predict chlorophyll a concentrations for new remote sensing imagery data. By inputting the reflectance of the new data into the model, the corresponding chlorophyll a concentration prediction is obtained. The model's performance is continuously monitored. As new data accumulates, the model can be regularly updated and optimized to improve its accuracy and adaptability.

[0114] In this embodiment, another deep learning model, the GBR model, the gradient boosting regression model, plays an important role in processing tasks with the reflectance of a highly correlated band combination as an explanatory variable and the chlorophyll a concentration as a response variable. In principle, the GBR model uses an iterative approach to gradually build a strong learner. It starts with an initial simple model. During each iteration, a new weak learner is trained in a targeted manner based on the prediction error generated by the previous model. These weak learners are usually decision trees. With continuous iterations, new weak learners are added to the model in turn, working together to gradually reduce the prediction error. The entire process is achieved by continuously optimizing the loss function, with the aim of enabling the model to fit the given data more accurately.

[0115] The GBR model offers many key features and advantages. First, it boasts high accuracy, often achieving excellent prediction results for complex nonlinear relationships. By automatically learning the interactions between features, the GBR model can more deeply capture the inherent patterns in the data, thereby providing more accurate predictions. Second, it offers significant flexibility, capable of handling a wide range of data types, including continuous and discrete variables. Furthermore, the GBR model adapts well to data of varying distributions and noise levels. Furthermore, the model is robust to outliers and noise. This is due to its training process, which gradually optimizes the loss function. This process provides a degree of resilience to outliers and noise, ensuring the model's stability and reliability. Furthermore, while not as intuitive as simpler models, the GBR model still provides a degree of explanation by calculating feature importance. This is very helpful for understanding the model's decision-making process and determining the importance of individual variables in predictions.

[0116] In practical applications, data preparation is the first step. Similar to the random forest model, reflectance data for highly correlated band combinations is prepared as the explanatory variable, and chlorophyll a concentration data is prepared as the response variable. The dataset is then divided into training and test sets. Next comes parameter setting. Determining parameters for the GBR model is crucial, such as the learning rate, number of trees, and maximum depth. The learning rate controls the update step size for each iteration, while the number of trees and maximum depth determine the complexity of the weak learners. Choosing these parameters appropriately directly impacts model performance and can be optimized using methods such as grid search and random search. Next comes the model training phase, where the GBR model is trained using the training set data. In each iteration, a new weak learner is trained based on the prediction error of the previous model and added to the model. This process continues until the preset number of iterations is reached or other stopping conditions are met. Next, model evaluation is performed using the test set data to evaluate the performance of the GBR model. Common evaluation metrics include mean squared error, mean absolute error, and coefficient of determination. Based on the evaluation results, further parameter adjustments or feature engineering can be performed to improve model performance. Finally, the optimized and evaluated GBR model can be used to predict chlorophyll a concentration in new remote sensing image data.

[0117] Furthermore, the GBR model includes the following steps:

[0118] (1) Data preparation

[0119] Collect reflectance data and corresponding chlorophyll-a concentration data from different band combinations of high-temporal and spatial resolution remote sensing images. Ensure the accuracy, completeness, and representativeness of the data, covering different environmental conditions and regional scopes.

[0120] Divide the dataset into a training set and a test set. This is typically done in a certain ratio, such as 80% for training and 20% for testing. This ensures that the model can fully learn the patterns in the data and effectively validate on unseen data during subsequent model training and evaluation.

[0121] (2) Parameter initialization

[0122] Determine the initial values ​​of some important parameters. For example, the learning rate can be initialized to a small value, such as 0.1. The learning rate controls the magnitude of the model update at each iteration. A smaller learning rate generally makes model training more stable, but may require more iterations to achieve good performance.

[0123] Set the initial number of weak learners (usually the number of decision trees), and start with a relatively small value, such as 100.

[0124] Determine the maximum depth of the decision tree, the minimum number of sample splits, and other parameters. These parameters will affect the complexity and performance of the decision tree, and the initial values ​​can be determined based on experience or through some preliminary experiments.

[0125] (3) Model training

[0126] Import GBR related libraries, such as GradientBoostingRegressor in scikit-learn.

[0127] The GBR model is trained using the training set data. The reflectance data of the highly correlated band combination is used as the input feature, and the chlorophyll a concentration data is used as the target variable. These data are passed to the model's fit method.

[0128] At each iteration, the model calculates the gradient of the loss function based on the current prediction error. Then, based on this gradient information, a new decision tree is trained as a weak learner to reduce the loss function. This new decision tree is then added to the model, working together with the previous weak learners to gradually improve model performance.

[0129] This process is repeated until the preset number of iterations is reached or other stopping conditions are met.

[0130] (4) Model evaluation

[0131] Use the test set data to evaluate the trained GBR model. Pass the reflectance data in the test set to the model's predict method to obtain the chlorophyll a concentration value predicted by the model.

[0132] Calculate evaluation metrics, such as mean squared error (MSE), mean absolute error (MAE), and coefficient of determination (R²). MSE and MAE measure the error between the predicted value and the true value, while the coefficient of determination indicates how well the model fits the data.

[0133] Analyze the model's performance based on the evaluation results. If the model performance is not ideal, adjust the parameters, retrain and reevaluate, or perform feature engineering, such as selecting a more appropriate band combination or performing data standardization.

[0134] (5) Application and optimization

[0135] The optimized GBR model can be used to predict chlorophyll a concentrations for new remote sensing imagery data. By inputting the reflectance of the new data into the model, the corresponding chlorophyll a concentration prediction is obtained. The model's performance is continuously monitored. As new data accumulates, the model can be regularly updated and optimized to improve its accuracy and adaptability.

[0136] S6. Use TPE search weight method to establish a fusion inversion model of random forest model and GBR model;

[0137] The TPE search weight method is an algorithm based on Bayesian optimization. Its core principle is to estimate the performance of different parameter combinations by constructing a probabilistic model. Based on this model, the next set of parameters is selected for evaluation, and the optimal parameter combination is gradually optimized to be found. In fusion inversion models, the definition of the parameter space is crucial. This includes the weights of the random forest model and the GBR model, and may also include other hyperparameters such as the number of trees in the random forest model and the learning rate in the GBR model. By properly defining the parameter space, a comprehensive exploration of different model combinations and hyperparameter settings is possible, laying the foundation for finding the optimal fusion inversion model. During the search process, the TPE algorithm gradually evaluates different parameter combinations in the parameter space. For each parameter set, it uses the fusion inversion model to make predictions and judges the quality of the parameter combination based on performance metrics such as mean squared error (MSE). As the evaluation continues, the TPE algorithm continuously updates the probabilistic model, gradually focusing on parameter regions with good performance and reducing exploration of regions with poor performance.

[0138] The TPE search weight method offers numerous advantages. The first is efficiency. Compared to traditional grid search or random search methods, TPE can more effectively explore the parameter space, avoiding unnecessary parameter evaluations and thus saving computational resources and time. Second is adaptability. It can automatically adapt to different problems and datasets without requiring special preprocessing or assumptions about the parameter space. Finally, accuracy. By constructing a probabilistic model, TPE can more accurately estimate the performance of different parameter combinations, greatly increasing the likelihood of finding the globally optimal parameter combination. In summary, the TPE search weight method provides a powerful and efficient means for finding the optimal model weights and hyperparameters for fusion inversion models, significantly improving model performance and generalization capabilities.

[0139] In this embodiment, the TPE search weight method is used to establish a fusion inversion model of the random forest model and the GBR model, which can fully combine the advantages of the two different models and improve the prediction accuracy of the target variable.

[0140] First, during the preparation phase, ensure that the necessary Python libraries are installed, such as scikit-learn for building random forest and GBR models, and hyperopt for parameter search. Also, prepare reflectance data from different band combinations of high-resolution remote sensing imagery as explanatory variables, and chlorophyll a concentration data as the response variable. Next, defining the objective function is a crucial step. This objective function accepts a set of parameters, including weights for the random forest and GBR models, as well as possible hyperparameters. Within the objective function, the predictions from the random forest and GBR models are first fused according to the given weights, for example, using a weighted average. Assuming that the random forest model rf and the GBR model gbr have been defined, first perform predictions on the test set, generating rf_preds and gbr_preds, respectively. The fused predictions, combined_preds, are then weighted averaged. Next, the performance of the fused model is measured using an appropriate evaluation metric, such as mean squared error (MSE), which is returned as the output of the objective function.

[0141] Next, use the fmin function from the hyperopt library to perform a parameter search. During this process, it's necessary to define the parameter space, which includes the weight ranges for the random forest model and the GBR model. Typically, you can use functions such as hp.uniform provided by hyperopt to define continuous parameter ranges. In addition, you can also include other possible hyperparameter ranges, such as the number of trees in the random forest model and the learning rate in the GBR model. Specify the search algorithm as TPE. TPE is an efficient parameter search algorithm that estimates the performance of different parameter combinations by building a probabilistic model and gradually optimizes the parameter selection. At the same time, set parameters such as the maximum number of evaluations to control the search time and resource consumption.

[0142] Finally, a fused inversion model is constructed based on the optimal weight parameters found. First, the random forest model and the GBR model are retrained using the training set to ensure that the models fully learn the characteristics and patterns of the training data. When making predictions, the predictions from the two models are fused using the optimal weights to obtain the final predicted value. In this way, the fused inversion model leverages the advantages of the random forest model's ability to resist overfitting and the GBR model's high precision, improving the prediction performance of chlorophyll a concentration and providing more accurate results for research and applications in related fields.

[0143] In summary, the TPE search weight method is used to establish a fusion inversion model of the random forest model and the GBR model, including the following steps:

[0144] (1) Define the objective function

[0145] Define an objective function that receives a set of parameters, including the weights of the random forest model and the GBR model, and possibly other hyperparameters.

[0146] Within the objective function, the prediction results of the random forest model and the GBR model are fused using given weights, such as weighted averaging.

[0147] Use the training set to train two models and evaluate the performance of the fused model on the test set. Usually, mean square error (MSE), mean absolute error (MAE), etc. can be used as evaluation indicators.

[0148] Returns the value of the evaluation metric as the output of the objective function.

[0149] (2) Parameter search using TPE

[0150] Import the hyperopt library, which is a library for hyperparameter optimization.

[0151] Define the parameter space, including the weight ranges for the random forest model and the GBR model and possible other hyperparameter ranges.

[0152] Use the fmin function of hyperopt for parameter search, specify the search algorithm as TPE, and set parameters such as the maximum number of evaluations.

[0153] (3) Constructing a fusion inversion model

[0154] Based on the optimal weight parameters obtained through search, a fusion inversion model is constructed.

[0155] Retrain the random forest model and GBR model using the training set.

[0156] During prediction, the prediction results of the two models are fused using the optimal weights.

[0157] (4) Model evaluation and application

[0158] The fused inversion model is evaluated using the test data set, and evaluation metrics such as mean squared error, mean absolute error, and coefficient of determination are calculated. The model's performance is analyzed based on the evaluation results. If the model performance is unsatisfactory, parameters can be adjusted and retrained and searched. The optimized fused inversion model can be used to predict chlorophyll a concentrations in new remote sensing imagery data.

[0159] Furthermore, when using the TPE search weight method to establish a fusion inversion model of the random forest model and the GBR model, the hyperparameters that may be involved are mainly the following:

[0160] (1) Random Forest Model Hyperparameters

[0161] n_estimators: The number of decision trees in the random forest. A larger number may improve the performance of the model, but also increase the computational cost and the risk of overfitting.

[0162] max_depth: The maximum depth of the decision tree. Limiting the depth can prevent overfitting, but if it is set too small, it may cause the model to underfit.

[0163] min_samples_split: The minimum number of samples required for the decision tree to split. Higher values ​​can prevent overfitting but may prevent the tree from growing sufficiently.

[0164] min_samples_leaf: The minimum number of samples required for a leaf node. Again, a higher value can prevent overfitting.

[0165] (2) GBR model hyperparameters

[0166] learning_rate: The learning rate controls the update step size of the model in each iteration. A smaller learning rate usually makes the model training more stable, but may require more iterations to achieve good performance.

[0167] n_estimators: The number of weak learners (decision trees). Similar to random forests, a larger number may result in better performance, but the computational cost will also increase.

[0168] max_depth: The maximum depth of the decision tree, which affects the complexity and performance of the model.

[0169] subsample: The sample ratio used in each iteration. Smaller values ​​can increase the randomness of the model and prevent overfitting.

[0170] S7. Obtain high temporal and spatial resolution remote sensing images of the estuary to be monitored, input the reflectance of the highly correlated band combination into the fusion inversion model, and invert the chlorophyll a concentration of the estuary to be monitored.

[0171] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A multi-model fusion-based chlorophyll-a concentration inversion method for estuaries, characterized by: The following steps are involved: S1. Acquire estuary monitoring data for a specific time period, wherein the estuary monitoring data includes chlorophyll a concentration and MODIS remote sensing imagery and Landsat 8 remote sensing imagery data for the corresponding time period; S2. Use ESTARFM to fuse MODIS remote sensing images and Landsat8 remote sensing images to obtain high temporal and spatial resolution remote sensing images; S3, obtaining the reflectance of different band combinations of high temporal and spatial resolution remote sensing images, and calculating the Pearson correlation coefficient between the reflectance and chlorophyll a concentration; S4. Determine the band combination with significant correlation between reflectance and chlorophyll a concentration based on the Pearson correlation coefficient, and record it as a highly correlated band combination; S5. Using the reflectance of the highly correlated band combination as an explanatory variable and the chlorophyll a concentration as a response variable, a random forest model and a GBR model are established respectively; S6. Use TPE search weight method to establish a fusion inversion model of random forest model and GBR model; S7, obtaining a high temporal and spatial resolution remote sensing image of the estuary to be monitored, inputting the reflectance of the highly correlated band combination thereof into the fusion inversion model, and inverting the chlorophyll a concentration of the estuary to be monitored; Among them, in step S6, the use of the TPE search weight method to establish a fusion inversion model of the random forest model and the GBR model includes the following steps: Define an objective function: Define an objective function that receives a set of parameters, including weights and hyperparameters for the random forest model and the GBR model. Within the objective function, use the given weights to fuse the predictions of the random forest model and the GBR model. Train the random forest model and the GBR model using the training set and evaluate the performance of the fused model on the test set. Return the value of the evaluation metric as the output of the objective function. Use TPE for parameter search: Define the parameter space in Python, which includes the weights of the random forest model and the GBR model and the range of hyperparameters; use the fmin function of hyperopt for parameter search, specify the search algorithm as TPE, and set the maximum number of evaluations; Constructing a fusion inversion model: Based on the optimal weight parameters obtained through the search, a fusion inversion model is constructed; the random forest model and the GBR model are retrained using the training set; during prediction, the prediction results of the two models are fused using the optimal weight parameters, specifically using a weighted average method; Model evaluation and application: Use test set data to evaluate the fusion inversion model and analyze the model performance based on the evaluation results.

2. The method for inverting chlorophyll-a concentration in an estuary based on multi-model fusion according to claim 1, characterized in that: The MODIS remote sensing images were downloaded from NASA's official website and preprocessed, including: Screening of remote sensing images: Download MODIS remote sensing images from NASA's official website and select remote sensing images with little cloud cover and no more than 7 days from the monitoring time; Remote sensing image radiometric calibration and geometric correction: Download the MODIS processing plug-in MTCK in ENVI, then import MODIS remote sensing data in MTCK and select the WGS-84 coordinate system; Water body extraction: The estuary water area was extracted based on the NDWI water body index, and band operations were performed using Band math in ENVI.

3. The method for inverting chlorophyll-a concentration in an estuary based on multi-model fusion according to claim 2, characterized in that: The NDWI water index is calculated as follows: ， Where, Green The green band representing the MODIS remote sensing image corresponds to the 4th band among the 36 bands; NIR Represents the second band of the 36 bands corresponding to the near-infrared band of MODIS remote sensing images; NDWI The water index.

4. The method for inverting chlorophyll-a concentration in an estuary based on multi-model fusion according to claim 1, characterized in that: The Landsat8 remote sensing images are acquired in the geospatial data cloud and pre-processed, including: Screening of remote sensing images: Download Landsat8 remote sensing images from the Geospatial Data Cloud and select remote sensing images with minimal cloud cover and the closest time to the MODIS remote sensing images; Remote sensing image radiometric calibration: converting the DN value recorded by the sensor into a radiometric brightness value. The DN value is the grayscale value of the recorded object. Remote sensing image atmospheric correction: Use the FLAASH Atmospheric Correction tool in ENVI to perform atmospheric correction on Landsat8 remote sensing images to eliminate the effects of electromagnetic waves passing through the atmosphere and the reflection of objects on the ground. Water body extraction: The estuary water area was extracted based on the NDWI water body index, and band operations were performed using Band math in ENVI.

5. The method for inverting chlorophyll-a concentration in an estuary based on multi-model fusion according to claim 1, characterized in that: In step S2, the use of ESTARFM to fuse the MODIS remote sensing image and the Landsat8 remote sensing image to obtain a high temporal and spatial resolution remote sensing image includes the following steps: Select reference images: select Landsat 8 remote sensing images as high-resolution images and MODIS remote sensing images as low-resolution images; By constructing a spatiotemporal model, the spectral information of MODIS remote sensing images and the spatial information of Landsat 8 remote sensing images are fused; Generate high temporal and spatial resolution images that retain the high spatial resolution of Landsat 8 remote sensing images while introducing the timeliness of MODIS remote sensing images.

6. The method for inverting chlorophyll-a concentration in an estuary based on multi-model fusion according to claim 1, characterized in that: In step S3, the Pearson correlation coefficient is calculated using the following formula: ， Where, p It represents the Pearson correlation coefficient between chlorophyll a concentration and reflectance of different band combinations of high temporal and spatial resolution remote sensing images; R ( x i ) represents the chlorophyll a concentration i Sample values; R ( y i ) represents the reflectivity of different band combinations of high temporal and spatial resolution remote sensing images. i Sample values; R ( x ) Chlorophyll a concentration is the average value of all samples; R ( y ) represents the average value of all reflectance samples of different band combinations of high temporal and spatial resolution remote sensing images.

7. The method for inverting chlorophyll-a concentration in an estuary based on multi-model fusion according to claim 1, characterized in that: In step S5, the random forest model includes the following construction steps: Data preparation: Collect reflectance data and corresponding chlorophyll a concentration data of different band combinations of high temporal and spatial resolution remote sensing images; Divide the dataset into training and testing sets; Parameter setting: determine the number of decision trees in the random forest, set the maximum depth of the decision tree, and determine the minimum number of sample splits; Model training: The random forest model is trained using the training set data, with the reflectance data of highly correlated band combinations as input features and the chlorophyll-a concentration data as the target variable. During the training process, multiple decision trees are constructed based on the set parameters. Each decision tree grows on a randomly selected sample set and randomly selected features to learn the relationship between reflectance and chlorophyll-a concentration. Model evaluation: The reflectance data in the test set is passed into the model to obtain the chlorophyll a concentration value predicted by the model; evaluation indicators are calculated, including the mean square error, mean absolute error, and determination coefficient.

8. The method for inverting chlorophyll-a concentration in an estuary based on multi-model fusion according to claim 1, characterized in that: In step S5, the GBR model includes the following steps: Data preparation: Collect reflectance data of different band combinations of high temporal and spatial resolution remote sensing images and the corresponding chlorophyll a concentration data; divide the dataset into training and test sets; Parameter initialization: Determine the initial values ​​of model parameters, including learning rate, number of weak learners, maximum depth of decision tree, and minimum number of sample splits; Model training: The GBR model is trained using the training set data, with reflectance data of highly correlated band combinations as input features and chlorophyll a concentration data as the target variable to construct the GBR model. In each iteration, the model calculates the gradient of the loss function based on the current prediction error. Based on the gradient information, a new decision tree is trained as a weak learner to reduce the loss function. The new decision tree is added to the GBR model and works together with the previous weak learner to gradually improve the model's performance. The model training process is repeated until the preset number of iterations is reached or the stopping condition is met. Model evaluation: Use the test set data to evaluate the trained GBR model.

9. The method for inverting chlorophyll-a concentration in an estuary based on multi-model fusion according to claim 1, characterized in that: The hyperparameters include the number of decision trees in the random forest, the maximum depth of the decision tree, the minimum number of samples required for the decision tree to split, the minimum number of samples required for leaf nodes, the learning rate in the GBR model, the number of weak learners, the maximum depth of the decision tree, and the proportion of samples used in each iteration.

Images