Quality optimization method for offshore multi-source space-time monitoring data

Abstract

The invention relates to a quality optimization method for offshore multi-source space-time monitoring data. The method comprises the following steps: acquiring an offshore multi-source space-time monitoring data set; constructing a multi-level evaluation index system; determining the weight of each index in the evaluation index system; obtaining a quality score of each independent data sample; multi-source data collaborative quality evaluation is carried out through a two-stage fusion strategy; and screening the original monitoring data according to the comprehensive quality evaluation, constructing an optimal data set oriented to a specific application scene, inputting the optimal data set into the application model, and outputting an environmental element inversion result. The method has the advantages that progressive quality optimization from samples to platforms and then to application results is achieved in a unified framework, low-quality samples can be effectively removed, and the utilization efficiency of multi-source data and the accuracy of inversion results in offshore ecological environment monitoring can be improved.

Description

Technical Field

[0001] This invention belongs to the field of data quality optimization technology, and in particular relates to a method for quality optimization of nearshore multi-source spatiotemporal monitoring data. Background Technology

[0002] Nearshore ecological environment monitoring is the foundation for supporting marine resource management and sustainable development. However, multi-source spatiotemporal monitoring data generally suffer from problems such as inconsistent quality, information redundancy or missing information, which can easily introduce systematic biases in target demand monitoring and long-term trend analysis. Therefore, how to scientifically and systematically evaluate data quality and screen high-quality data suitable for monitoring specific ecological factors has become a key issue that marine environmental informatics urgently needs to address.

[0003] Currently, in the field of nearshore ecological environment monitoring, data quality assessment methods still largely rely on single data sources (such as ground observation stations or remote sensing satellites). These methods have significant limitations in terms of spatiotemporal coverage, data resolution, and dynamic response capabilities, making it difficult to effectively address the complex, variable, and highly spatiotemporally heterogeneous monitoring needs of nearshore ecosystems. Furthermore, most methods fail to fully integrate with the specific application scenarios of nearshore ecological environment monitoring, making it difficult to directly guide data fusion and inversion modeling, and failing to fully leverage the synergistic advantages of multi-source observation systems in nearshore ecological environment monitoring. This deficiency often results in assessment results that cannot directly support practical operations such as data screening, fusion optimization, and monitoring and early warning. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for quality optimization of nearshore multi-source spatiotemporal monitoring data.

[0005] Firstly, a method for quality optimization of nearshore multi-source spatiotemporal monitoring data is provided, including:

[0006] Step 1: Acquire multi-source spatiotemporal monitoring data and preprocess it to form a unified nearshore multi-source spatiotemporal monitoring dataset; the multi-source spatiotemporal monitoring data includes multi-source satellite remote sensing data, UAV observation data, and in-situ measured data from land and sea monitoring stations;

[0007] Step 2: Based on the technical characteristics of different data sources, construct a multi-level evaluation index system from multiple preset quality dimensions;

[0008] Step 3: Determine the weight of each indicator in the evaluation index system based on the subjective and objective combined weighting method;

[0009] Step 4: Evaluate the multi-source spatiotemporal monitoring data using the evaluation index system with the weights mentioned above, and obtain the quality score of each independent data sample;

[0010] Step 5: Conduct multi-source data collaborative quality evaluation through a two-level fusion strategy. The two-level fusion includes calculating the data quality score within the platform and performing cross-platform data consistency analysis to obtain a comprehensive quality evaluation of the multi-source data.

[0011] Step 6: Based on the comprehensive quality evaluation, the original monitoring data is screened to construct an optimal dataset for specific application scenarios. The optimal dataset is then input into the application model to output the environmental element inversion results.

[0012] Preferably, in step 1, the preprocessing includes: unifying multi-source satellite remote sensing data, UAV observation data, and in-situ measured data from land and sea monitoring stations to the same time reference, then matching geographical regions according to the data coverage, and eliminating differences in scale and dimension between data through 0-1 standardization.

[0013] Preferably, in step 2, the multi-level evaluation index system includes a type layer, a characteristic layer, and a feature layer.

[0014] The type layer is used to classify data sources according to the observation platform;

[0015] The feature layer includes multiple preset dimensions for evaluating data quality;

[0016] For each preset dimension, the feature layer defines specific feature indicators based on the technical characteristics and monitoring requirements of the corresponding platform.

[0017] Preferably, in step 2, the characteristic layer includes four dimensions: accuracy, completeness, validity, and consistency. The accuracy reflects the degree to which the observed value approximates the true state; the completeness assesses the continuity of the data in the spatiotemporal sequence; the validity measures the applicability of the data to a specific monitoring target; and the consistency emphasizes the synergy of multi-source data when describing the same target.

[0018] Preferably, in step 2, for accuracy, geometric angle parameters related to radiation quality and flare control, as well as the signal-to-noise ratio of sensitive bands, are selected as indicators; for completeness, data missing rate, outlier percentage, and number of bands are selected as indicators; for validity, spatial resolution, information entropy, and spectral resolution of sensitive bands are selected as indicators; and for consistency, format consistency, content consistency, and logical consistency are selected as indicators.

[0019] Preferably, in step 3, the subjective and objective combined weighting method includes:

[0020] The Analytic Hierarchy Process (AHP) is used to construct a judgment matrix based on prior knowledge to obtain subjective weights.

[0021] The entropy weight method is used to calculate objective weights based on the statistical characteristics of the data;

[0022] The subjective weights and objective weights are combined using a linear weighting method to obtain the comprehensive weights of each indicator.

[0023] Preferably, in step 5, the two-level fusion strategy specifically includes:

[0024] Level 1: Calculate the average of the quality scores of multiple samples within the same observation platform to obtain the overall quality score of the platform;

[0025] Level 2: Spatiotemporal matching of data from different platforms, and consistency scores between computing platforms based on successfully matched data;

[0026] By combining the overall quality scores from various platforms with the consistency scores, a comprehensive quality evaluation of the multi-source data is obtained.

[0027] Preferably, step 6 includes:

[0028] Step 6.1: Set a screening threshold and construct an optimal dataset based on the comprehensive quality evaluation screening data;

[0029] Step 6.2: Input the selected dataset into the application model to perform environmental element inversion, and verify the inversion accuracy using measured data as the true value;

[0030] Step 6.3: If the inversion accuracy does not meet expectations, adjust the screening threshold or indicator weights accordingly, and re-execute the evaluation and screening process.

[0031] Secondly, a quality optimization system for nearshore multi-source spatiotemporal monitoring data is provided, for performing any of the methods described in the first aspect, including:

[0032] The acquisition module is used to acquire multi-source spatiotemporal monitoring data and preprocess it to form a unified nearshore multi-source spatiotemporal monitoring dataset; the multi-source spatiotemporal monitoring data includes multi-source satellite remote sensing data, UAV observation data and in-situ measured data from land and sea monitoring stations;

[0033] The module is used to build a multi-level evaluation index system based on multiple preset quality dimensions, taking into account the technical characteristics of different data sources.

[0034] The determination module is used to determine the weight of each indicator in the evaluation index system based on a subjective and objective combined weighting method.

[0035] The first evaluation module is used to evaluate the multi-source spatiotemporal monitoring data using an evaluation index system with the weights mentioned above, and to obtain the quality score of each independent data sample.

[0036] The second evaluation module is used to conduct collaborative quality evaluation of multi-source data through a two-level fusion strategy. The two-level fusion includes the calculation of data quality scores within the platform and cross-platform data consistency analysis to obtain a comprehensive quality evaluation of multi-source data.

[0037] The filtering module is used to filter the original monitoring data according to the comprehensive quality evaluation, construct an optimal dataset for a specific application scenario, input the optimal dataset into the application model, and output the environmental element inversion results.

[0038] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as described in any of the first aspects.

[0039] The beneficial effects of this invention are as follows: Addressing the heterogeneous characteristics of nearshore multi-source spatiotemporal monitoring data, this invention constructs a three-tiered indicator system of "type-characteristic-feature," determines weights through a weighting method combining subjective and objective approaches, and employs a two-level fusion strategy to achieve collaborative quality evaluation of multi-source data. Furthermore, this invention introduces a quality optimization step tailored to specific monitoring / inversion tasks. By setting quality thresholds to construct an optimized dataset, it inputs this dataset into a nearshore ecological environment application model for inversion and verification. The model accuracy index is used to provide feedback correction to the quality threshold and indicator weights. Compared to existing methods, this invention achieves progressive quality optimization from samples to platform and then to application results within a unified framework. This effectively eliminates low-quality samples and improves the utilization efficiency of multi-source data and the accuracy of inversion results in nearshore ecological environment monitoring. Attached Figure Description

[0040] Figure 1 This invention provides a flowchart of an implementation method for quality optimization of nearshore multi-source spatiotemporal monitoring data;

[0041] Figure 2 This is a schematic diagram of the evaluation index system for space-based multi-source remote sensing data provided by the present invention;

[0042] Figure 3 This is a schematic diagram of the airborne UAV data evaluation index system provided by the present invention;

[0043] Figure 4 This is a schematic diagram of the data evaluation index system for land-based and sea-based monitoring stations provided by the present invention;

[0044] Figure 5a This is a schematic diagram comparing the effectiveness of the evaluation method of this invention with traditional data evaluation methods;

[0045] Figures 5b-5d for Figure 5a A magnified view of a portion of the image;

[0046] Figure 6 This diagram illustrates the comparison of the accuracy of the inversion results between the preferred dataset and the comparison dataset of this invention. Detailed Implementation

[0047] The present invention will be further described below with reference to embodiments. The description of the embodiments below is only for the purpose of helping to understand the present invention. It should be noted that those skilled in the art can make several modifications to the present invention without departing from the principle of the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

[0048] Example 1:

[0049] To address the problems of existing technologies, Embodiment 1 of this application provides a quality optimization method for nearshore multi-source spatiotemporal monitoring data. This method introduces multi-platform collaborative evaluation, cross-source consistency analysis, and application results to effectively improve the efficiency of comprehensive utilization of multi-source data and the reliability of inversion results.

[0050] Specifically, such as Figure 1 As shown, the method provided in this application embodiment includes:

[0051] Step 1: To meet the target needs of multi-source spatiotemporal monitoring of nearshore ecological resources and environment (such as red tide early warning and water quality inversion), acquire multi-source spatiotemporal monitoring data and perform time synchronization, spatial registration and standardization preprocessing to form a unified nearshore multi-source spatiotemporal monitoring dataset; the multi-source spatiotemporal monitoring data includes multi-source satellite remote sensing data, UAV observation data and in-situ measured data from land and sea monitoring stations.

[0052] Specifically, the preprocessing includes: unifying multi-source satellite remote sensing data, UAV observation data, and in-situ measured data from land and sea monitoring stations to the same time reference, matching geographical regions according to the data coverage, and eliminating differences in scale and dimension between data through 0-1 standardization.

[0053] Step 2: Based on the technical characteristics of different data sources, construct a multi-level evaluation index system from multiple preset quality dimensions.

[0054] In step 2, the multi-level evaluation index system includes a type layer, a characteristic layer, and a feature layer. The type layer macroscopically distinguishes the observation scale, coverage, and applicable scenarios of different data sources, and classifies them into space-based, air-based, and land-based / sea-based data observation platforms. The characteristic layer is developed from four dimensions: data accuracy, completeness, validity, and consistency. Accuracy reflects the degree of approximation between observed values ​​and the objective reality. Completeness assesses the continuity and degree of missing data in time or space sequences. Validity measures the applicability and contribution of data information to a specific monitoring target. Consistency emphasizes the synergy and correlation of results from multiple data sources when describing the same target. The feature layer further refines and defines representative feature indicators based on the technical characteristics of each platform and specific monitoring needs.

[0055] In step 2, for accuracy, geometric angle parameters related to radiation quality and flare control, as well as the signal-to-noise ratio of sensitive bands, are selected as indicators; for completeness, data missing rate, outlier percentage, and number of bands are selected as indicators; for validity, spatial resolution, information entropy, and spectral resolution of sensitive bands are selected as indicators; and for consistency, format consistency, content consistency, and logical consistency are selected as indicators.

[0056] Step 3: Determine the weight of each indicator in the evaluation index system based on the subjective and objective combined weighting method.

[0057] Step 4: Use the evaluation index system with the weights to evaluate the multi-source spatiotemporal monitoring data and obtain the quality score of each independent data sample.

[0058] Step 5: Conduct multi-source data collaborative quality evaluation through a two-level fusion strategy. The two-level fusion includes calculating the data quality score within the platform and performing cross-platform data consistency analysis to obtain a comprehensive quality evaluation of the multi-source data.

[0059] Step 6: Based on the comprehensive quality evaluation, the original monitoring data is screened to construct an optimal dataset for specific application scenarios. The optimal dataset is then input into the application model to output the environmental element inversion results.

[0060] Example 2:

[0061] Based on Example 1, Example 2 of this application provides a more specific method for quality optimization of nearshore multi-source spatiotemporal monitoring data, including:

[0062] Step 1: Acquire multi-source spatiotemporal monitoring data and preprocess it to form a unified nearshore multi-source spatiotemporal monitoring dataset; the multi-source spatiotemporal monitoring data includes multi-source satellite remote sensing data, UAV observation data and in-situ measured data from land and sea monitoring stations.

[0063] Step 2: Based on the technical characteristics of different data sources, construct a multi-level evaluation index system from multiple preset quality dimensions.

[0064] Step 3: Determine the weight of each indicator in the evaluation index system based on the subjective and objective combined weighting method.

[0065] In step 3, the subjective and objective combined weighting method includes:

[0066] The Analytic Hierarchy Process (AHP) is used to construct a judgment matrix based on prior knowledge to obtain subjective weights.

[0067] The entropy weight method is used to calculate objective weights based on the statistical characteristics of the data;

[0068] The subjective weights and objective weights are combined using a linear weighting method to obtain the comprehensive weights of each indicator.

[0069] Specifically, the determination of subjective weights includes: constructing a two-layer weighting system using the analytic hierarchy process (AHP); firstly, establishing judgment matrices for the three types of data sources—space-based, air-based, and land-based / sea-based—and comparing each other pairwise across the four dimensions of accuracy, completeness, effectiveness, and consistency at the characteristic layer; secondly, constructing judgment matrices again for the specific indicators under each characteristic dimension at the feature layer; quantifying relative importance using the 1-9 scaling method, and obtaining the characteristic layer weights and feature layer weights by solving for the maximum eigenvalue and corresponding eigenvector of each judgment matrix; all judgment matrices must pass a consistency check to ensure the rationality of the evaluation logic, thus forming a subjective weighting system that integrates prior knowledge and application needs.

[0070] The determination of objective weights includes: using the entropy weight method to directly calculate the objective weights of all feature indicators, and dividing the indicators into positive and negative categories according to the monitoring objectives; quantifying the dispersion of data distribution by calculating the information entropy of each indicator, calculating the difference coefficient to quantify the distinguishing ability of different indicators, and normalizing them to obtain the objective weights; this method is entirely based on the statistical characteristics of the data and avoids the influence of human subjective preferences.

[0071] The determination of comprehensive weights includes: firstly, calculating the effective subjective weights, and converting the two-layer weight structure obtained by the analytic hierarchy process (AHP) into a single-layer structure; for the j-th feature index under the i-th feature dimension, where the weight of the i-th dimension is... The weight of this indicator is Its effective subjective weight The calculation is as follows:

[0072] =

[0073] A linear weighting strategy is used to fuse subjective and objective weights:

[0074]

[0075] in, Let j be the final weight of the j-th indicator. , The weights are calculated using the subjective hierarchical analysis method and the objective entropy weight method, respectively; the fusion coefficient α can be flexibly adjusted according to the monitoring scenario to achieve dynamic optimization of the weight configuration.

[0076] Step 4: Use the evaluation index system with the weights to evaluate the multi-source spatiotemporal monitoring data and obtain the quality score of each independent data sample.

[0077] Specifically, the preprocessed data is input into the evaluation index system, using the comprehensive weight obtained in step 3. The overall quality score of the samples is calculated by weighted summation. The specific calculation process is as follows:

[0078] =

[0079] Among them, Q sample For the quality score of a single sample, For the first The weights of each feature indicator, The dimensionless value of the j-th feature index of the sample after standardization (range [0,1]). This represents the total number of characteristic indicators for this type of data source.

[0080] Step 5: Conduct multi-source data collaborative quality evaluation through a two-level fusion strategy. The two-level fusion includes calculating the data quality score within the platform and performing cross-platform data consistency analysis to obtain a comprehensive quality evaluation of the multi-source data.

[0081] In step 5, the two-level fusion strategy specifically includes:

[0082] Level 1: Calculate the average of the quality scores of multiple samples within the same observation platform to obtain the overall quality score of the platform.

[0083] Specifically, the first level is platform-based fusion, which involves aggregating and calculating the quality scores of multiple samples within each of the three types of observation platforms: space-based, air-based, and land-based / sea-based.

[0084] =

[0085] in, Let n be the quality score calculated for the i-th sample in step 4, and n be the total number of samples on the platform. The platform is given a comprehensive quality score; through this fusion method, the overall quality score of each observation platform within the target area is obtained.

[0086] Level 2: Spatiotemporal matching of data from different platforms, and consistency scores between computing platforms based on successfully matched data.

[0087] Specifically, the second level is cross-platform consistency analysis. Based on the spatiotemporal correlation and observation consistency of multi-source data, a collaborative evaluation strategy for different platforms is constructed. First, data from different platforms are matched according to the principle of spatiotemporal proximity, and valid data pairs with a spatial distance less than a preset threshold and an observation time difference within an allowable window are selected to establish the spatiotemporal correspondence of multi-platform data. Then, the consistency of data between platforms is quantified. Observations from each platform are extracted from the successfully matched data, and the correlation coefficient between each pair of platforms is calculated.

[0088]

[0089] in, Let be the average consistency coefficient between platform k and platform l. The number of valid data pairs for spatiotemporal matching between the two platforms. Let be the Pearson correlation coefficient for the Mth pair of data.

[0090] By combining the overall quality scores from various platforms with the consistency scores, a comprehensive quality evaluation of the multi-source data is obtained.

[0091] Specifically, generate an overall consistency score across multiple platforms:

[0092]

[0093] Where K is the total number of platforms participating in the evaluation (K=3). The value range is [0,1], and the higher the value, the better the synergy of multi-platform observations;

[0094] Finally, integrating the scores of individual platforms with the scores of consistency between platforms, and considering the functional complementarity of space-based, air-based, and ground-based / sea-based platforms in nearshore monitoring according to the specific application requirements of monitoring, the platform weights are determined based on experience: weighting of space-based satellite remote sensing data. Data empowerment for airborne drones Data empowerment for land-based and sea-based monitoring stations ,make sure

[0095] = 1; In practical applications, data can be assigned values ​​according to specific target requirements, and the platform quality score and consistency index can be combined:

[0096] = + + +

[0097] in, To provide a comprehensive quality score for multi-source data, , , These correspond to the quality scores of space-based platforms, air-based platforms, and land-based / sea-based platforms, respectively. It is the overall consistency score across multiple platforms.

[0098] Step 6: Based on the comprehensive quality evaluation, the original monitoring data is screened to construct an optimal dataset for specific application scenarios. The optimal dataset is then input into the application model to output environmental element inversion results that are superior to those obtained from single evaluation data screening, thereby achieving data quality optimization for application scenarios.

[0099] Step 6 includes:

[0100] Step 6.1: Set a screening threshold and construct an optimal dataset based on the comprehensive quality evaluation screening data;

[0101] Step 6.2: Input the selected dataset into the application model to perform environmental element inversion, and verify the inversion accuracy using measured data as the true value;

[0102] Step 6.3: If the inversion accuracy does not meet expectations, adjust the screening threshold or indicator weights accordingly, and re-execute the evaluation and screening process.

[0103] For example, a selection threshold is set based on the comprehensive quality score to construct an optimal dataset, while a comparison dataset for single quality evaluation screening is established. The optimal dataset and the comparison dataset are respectively input into the application model, and the accuracy index of the two sets of inversion results is calculated and compared with the in-situ measured data as the true value. If the accuracy index of the optimal dataset is better than that of the comparison dataset, the high-precision inversion result is output. If it is not better, the selection threshold or the index weight in step 3 is adjusted according to the accuracy difference feedback, and the scoring and screening process is re-executed until the accuracy index of the optimal dataset is better than that of the comparison dataset.

[0104] The effectiveness of the present invention will be further analyzed below through specific experimental results:

[0105] This embodiment focuses on the nearshore ecological environment monitoring scenario of chlorophyll concentration monitoring in Hangzhou Bay: Data from September 5 to September 20, 2023, were collected in the Hangzhou Bay area, including remote sensing data from GOCI-II L2, MODIS Aqua L2, and Sentinel-3 L2, field measurement data from DJI Mavic 3 Enterprise UAVs, and data from marine monitoring stations in the harbor area; spatiotemporal filtering and preprocessing were performed according to the aforementioned method of this invention to construct a multi-source spatiotemporal detection data quality evaluation dataset.

[0106] According to the aforementioned method of the present invention, a data evaluation index system is constructed for the three platforms, each with its own characteristics, as follows: Figure 2 , Figure 3 , Figure 4 The data quality scores were used to determine the scoring levels, as shown in Table 1. Python was used to standardize the data from 0 to 1 and extract or calculate corresponding feature values. Subjective weights of the evaluation index system were determined by expert scoring; all p-values ​​were less than 0.01, passing the consistency test. Objective weights were calculated using the entropy weight method to determine the fusion coefficient. The comprehensive weights of individual indicators were obtained, and the specific weight distributions are shown in Tables 2, 3, and 4 below. The standardized values ​​of the corresponding features were multiplied by their respective weights to obtain the quality score of a single sample. A weighted average within the same platform was then used to obtain the platform's overall quality score. The quality scores for the space-based platform, air-based platform, and ground-based / sea-based platform were 0.65, 0.87, and 0.73, respectively, with quality levels of Good, Excellent, and Good. A time threshold of 24 hours and a spatial distance of 500 meters were set. Multi-source remote sensing and UAV data were matched based on the ground station location. Only successfully matched samples were retained to calculate changes in chlorophyll concentration. For multi-source remote sensing and UAV data, a three-band reflectance model was used as the estimate, resulting in a consistency score of 0.56 among the multi-source data. Based on empirical information, in nearshore environmental monitoring, remote sensing data is considered equally important due to its large-scale synchronous observation characteristics, while UAV data and ground monitoring data are considered equally important. =0.4、 = =0.3, resulting in an overall data quality score of 1.30 for this batch of data, classifying the data quality as medium. Through practical application, this invention demonstrates its ability to clearly identify low-quality samples on a single platform and to obtain quantitative results of data consistency across datasets through consistency analysis, providing reliable data quality assurance for subsequent applications.

[0107] Table 1 Data Quality Rating Levels

[0108]

[0109] Table 2. Weight distribution of space-based data in the example

[0110]

[0111] Table 3 shows the weight distribution of the empty basis data in the example.

[0112]

[0113] Table 4. Weight distribution of foundation data in the example

[0114]

[0115] Traditional methods typically evaluate data from different platforms (such as space-based, air-based, and land-based / sea-based) independently. In contrast, this invention expands the evaluation dimensions and overcomes the limitations of single weights by constructing a multi-level evaluation system and employing subjective and objective weighting methods. It innovatively quantifies the consistency between data from multiple platforms and adds a quality optimization step tailored to application scenarios. Figure 5a As shown ( Figures 5b-5d for Figure 5a (Enlarged view of China's Earth-Sea-Based Data). Even if the quality of the selected sample data (both air-based and Earth-Sea-based) is rated as good or excellent, this invention can still identify inherent contradictions between the data through consistency analysis, ultimately resulting in a poor overall evaluation. This demonstrates that this invention avoids the problem of traditional methods relying on individual high-scoring data and ignoring conflicts between multiple data sources, leading to overall misjudgment. It provides a more reliable basis for evaluating the quality of multi-source data.

[0116] In the data quality optimization stage, the "excellent" level was selected as the screening threshold in the example, and Earth-based data was used as the ground truth. A commonly used international three-band semi-analysis model was employed for chlorophyll inversion modeling. Traditional single-data-quality assessment methods selected 23 remote sensing images to form a comparative dataset; using the multi-source quality assessment and consistency constraint method proposed in this invention, 16 remote sensing images passed the quality screening, forming a preferred dataset. Chlorophyll inversion was performed based on both the comparative and preferred datasets, and the accuracy was verified against Earth-based measured data. The resulting scatter plot comparison results are shown below. Figure 6 As shown, the preferred dataset of this invention shows a closer fit between the inversion results and the measured values, with a higher coefficient of determination (R²) and a smaller root mean square error, indicating that the inversion accuracy is improved even with a reduced sample size. This embodiment demonstrates that the data quality optimization method proposed in this invention can effectively eliminate low-quality or contradictory data samples and reasonably retain high-quality samples, thereby achieving substantial optimization of the quality of nearshore chlorophyll inversion data.

[0117] It should be noted that the parts in this embodiment that are the same as or similar to those in Embodiment 1 can be referred to each other, and will not be repeated in this application.

[0118] Example 3:

[0119] Based on Example 2, Example 3 of this application provides a quality optimization system for nearshore multi-source spatiotemporal monitoring data, including:

[0120] The acquisition module is used to acquire multi-source spatiotemporal monitoring data and preprocess it to form a unified nearshore multi-source spatiotemporal monitoring dataset; the multi-source spatiotemporal monitoring data includes multi-source satellite remote sensing data, UAV observation data and in-situ measured data from land and sea monitoring stations;

[0121] The module is used to build a multi-level evaluation index system based on multiple preset quality dimensions, taking into account the technical characteristics of different data sources.

[0122] The determination module is used to determine the weight of each indicator in the evaluation index system based on a subjective and objective combined weighting method.

[0123] The first evaluation module is used to evaluate the multi-source spatiotemporal monitoring data using an evaluation index system with the weights mentioned above, and to obtain the quality score of each independent data sample.

[0124] The second evaluation module is used to conduct collaborative quality evaluation of multi-source data through a two-level fusion strategy. The two-level fusion includes the calculation of data quality scores within the platform and cross-platform data consistency analysis to obtain a comprehensive quality evaluation of multi-source data.

[0125] The filtering module is used to filter the original monitoring data according to the comprehensive quality evaluation, construct an optimal dataset for a specific application scenario, input the optimal dataset into the application model, and output the environmental element inversion results.

[0126] It should be noted that the system provided in this embodiment is the corresponding system of the method provided in embodiment 2. Therefore, the parts that are the same as or similar to those in embodiment 2 in this embodiment can be referred to each other, and will not be described again in this application.

[0127] In summary, this invention provides a quality optimization method for nearshore multi-source spatiotemporal monitoring data, which can achieve a comprehensive evaluation of the internal quality and cross-platform consistency of multi-source data, and use the evaluation results to construct an optimal dataset for specific application scenarios, significantly improving the reliability of nearshore ecological environment monitoring and inversion products.

Claims

1. A method for quality optimization of nearshore multi-source spatiotemporal monitoring data, characterized in that, include: Step 1: Acquire multi-source spatiotemporal monitoring data and preprocess it to form a unified nearshore multi-source spatiotemporal monitoring dataset; the multi-source spatiotemporal monitoring data includes multi-source satellite remote sensing data, UAV observation data, and in-situ measured data from land and sea monitoring stations; Step 2: Based on the technical characteristics of different data sources, construct a multi-level evaluation index system from multiple preset quality dimensions; Step 3: Determine the weight of each indicator in the evaluation index system based on the subjective and objective combined weighting method; Step 4: Evaluate the multi-source spatiotemporal monitoring data using the evaluation index system with the weights mentioned above, and obtain the quality score of each independent data sample; Step 5: Conduct multi-source data collaborative quality evaluation through a two-level fusion strategy. The two-level fusion includes calculating the data quality score within the platform and performing cross-platform data consistency analysis to obtain a comprehensive quality evaluation of the multi-source data. Step 6: Based on the comprehensive quality evaluation, the original monitoring data is screened to construct an optimal dataset for specific application scenarios. The optimal dataset is then input into the application model to output the environmental element inversion results.

2. The method for quality optimization of nearshore multi-source spatiotemporal monitoring data according to claim 1, characterized in that, In step 1, the preprocessing includes: unifying multi-source satellite remote sensing data, UAV observation data, and in-situ measured data from land and sea monitoring stations to the same time reference, matching geographical regions according to the data coverage, and eliminating differences in scale and dimension between data through 0-1 standardization.

3. The method for quality optimization of nearshore multi-source spatiotemporal monitoring data according to claim 2, characterized in that, In step 2, the multi-level evaluation index system includes a type layer, a characteristic layer, and a feature layer; The type layer is used to classify data sources according to the observation platform; The feature layer includes multiple preset dimensions for evaluating data quality; For each preset dimension, the feature layer defines specific feature indicators based on the technical characteristics and monitoring requirements of the corresponding platform.

4. The method for quality optimization of nearshore multi-source spatiotemporal monitoring data according to claim 3, characterized in that, In step 2, the characteristic layer includes four dimensions: accuracy, completeness, validity, and consistency. Accuracy reflects the degree to which the observed value approximates the true state; completeness assesses the continuity of the data in the spatiotemporal sequence; validity measures the applicability of the data to a specific monitoring target; and consistency emphasizes the synergy of multi-source data when describing the same target.

5. The method for quality optimization of nearshore multi-source spatiotemporal monitoring data according to claim 4, characterized in that, In step 2, in terms of accuracy, geometric angle parameters related to radiation quality and flare control, as well as the signal-to-noise ratio of sensitive bands, are selected as indicators; in terms of completeness, data missing rate, outlier ratio, and number of bands are selected as indicators; in terms of validity, spatial resolution, information entropy, and spectral resolution of sensitive bands are selected as indicators; and in terms of consistency, format consistency, content consistency, and logical consistency are selected as indicators.

6. The method for quality optimization of nearshore multi-source spatiotemporal monitoring data according to claim 5, characterized in that, In step 3, the subjective and objective combined weighting method includes: The Analytic Hierarchy Process (AHP) is used to construct a judgment matrix based on prior knowledge to obtain subjective weights. The entropy weight method is used to calculate objective weights based on the statistical characteristics of the data; The subjective weights and objective weights are combined using a linear weighting method to obtain the comprehensive weights of each indicator.

7. The method for quality optimization of nearshore multi-source spatiotemporal monitoring data according to claim 6, characterized in that, In step 5, the two-level fusion strategy specifically includes: Level 1: Calculate the average of the quality scores of multiple samples within the same observation platform to obtain the overall quality score of the platform; Level 2: Spatiotemporal matching of data from different platforms, and consistency scores between computing platforms based on successfully matched data; By combining the overall quality scores from various platforms with the consistency scores, a comprehensive quality evaluation of the multi-source data is obtained.

8. The method for quality optimization of nearshore multi-source spatiotemporal monitoring data according to claim 7, characterized in that, Step 6 includes: Step 6.1: Set a screening threshold and construct an optimal dataset based on the comprehensive quality evaluation screening data; Step 6.2: Input the selected dataset into the application model to perform environmental element inversion, and verify the inversion accuracy using measured data as the true value; Step 6.3: If the inversion accuracy does not meet expectations, adjust the screening threshold or indicator weights accordingly, and re-execute the evaluation and screening process.

9. A quality optimization system for nearshore multi-source spatiotemporal monitoring data, characterized in that, For performing the method according to any one of claims 1 to 8, comprising: The acquisition module is used to acquire multi-source spatiotemporal monitoring data and preprocess it to form a unified nearshore multi-source spatiotemporal monitoring dataset; the multi-source spatiotemporal monitoring data includes multi-source satellite remote sensing data, UAV observation data and in-situ measured data from land and sea monitoring stations; The module is used to build a multi-level evaluation index system based on multiple preset quality dimensions, taking into account the technical characteristics of different data sources. The determination module is used to determine the weight of each indicator in the evaluation index system based on a subjective and objective combined weighting method. The first evaluation module is used to evaluate the multi-source spatiotemporal monitoring data using an evaluation index system with the weights mentioned above, and to obtain the quality score of each independent data sample. The second evaluation module is used to conduct collaborative quality evaluation of multi-source data through a two-level fusion strategy. The two-level fusion includes the calculation of data quality scores within the platform and cross-platform data consistency analysis to obtain a comprehensive quality evaluation of multi-source data. The filtering module is used to filter the original monitoring data according to the comprehensive quality evaluation, construct an optimal dataset for a specific application scenario, input the optimal dataset into the application model, and output the environmental element inversion results.

10. A computer device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as claimed in any one of claims 1 to 8.

Images