CN-122023998-A - Reservoir hydro-fluctuation belt recovery method and system based on multi-source remote sensing data fusion

CN122023998ACN 122023998 ACN122023998 ACN 122023998ACN-122023998-A

Abstract

The invention discloses a reservoir hydro-fluctuation belt recovery method and system based on multi-source remote sensing data fusion. The method comprises the steps of obtaining multi-source remote sensing data of multiple time nodes of a target reservoir hydro-fluctuation belt in a preset time period, generating fusion remote sensing data through a data fusion technology, extracting ecological change characteristics of a target area based on the fusion remote sensing data, carrying out ecological assessment to obtain an ecological assessment result in the preset time period, determining ecological restoration prediction time according to an optimal time interval of ecological restoration predicted according to the assessment result, and formulating an ecological restoration scheme by combining periodic ecological change requirements of the target reservoir hydro-fluctuation belt. The method can comprehensively utilize multi-source remote sensing data, improves the accuracy of ecological assessment, provides scientific basis for ecological restoration of the reservoir hydro-fluctuation belt, and has important practical value and wide application prospect.

Inventors

YANG FENGJUAN
LIU DA
HONG CHANGHONG
LIU SHUAI
Xiao Gengfeng
Mai Yingwen

Assignees

广东省水利水电科学研究院

Dates

Publication Date: 20260512
Application Date: 20260410

Claims (7)

1. A reservoir hydro-fluctuation belt recovery method based on multi-source remote sensing data fusion is characterized by comprising the following steps: acquiring multi-source remote sensing data of a preset time node in a preset time period of a target reservoir hydro-fluctuation belt, wherein the multi-source remote sensing data comprises radar remote sensing data and satellite remote sensing data, and performing fusion operation on the multi-source remote sensing data to obtain fused remote sensing data; Extracting ecological change characteristics of the target reservoir hydro-fluctuation belt within a preset time period according to the fused remote sensing data, and carrying out ecological assessment on the target reservoir hydro-fluctuation belt according to the ecological change characteristics to obtain an ecological assessment result within the preset time period; predicting an optimal time interval for carrying out ecological restoration on the hydro-fluctuation belt of the target reservoir according to the ecological assessment result to obtain ecological restoration prediction time; and acquiring periodic ecological change demand data of the target reservoir hydro-fluctuation belt, and determining an ecological restoration scheme of the target reservoir hydro-fluctuation belt according to the periodic ecological change demand data and the ecological restoration prediction time.
2. The reservoir hydro-fluctuation belt recovery method based on multi-source remote sensing data fusion according to claim 1, wherein the multi-source remote sensing data of a preset time node in a preset time period of a target reservoir hydro-fluctuation belt is obtained, the multi-source remote sensing data comprises radar remote sensing data and satellite remote sensing data, and fusion operation is performed on the multi-source remote sensing data to obtain fusion remote sensing data, specifically: Acquiring multi-source remote sensing data of a preset time node of a target reservoir hydro-fluctuation belt in a preset time period, wherein the multi-source remote sensing data comprise satellite remote sensing data and radar remote sensing data; Introducing a SIFT algorithm, and performing convolution operation based on a Gaussian kernel function on the satellite remote sensing data and the radar remote sensing data according to the SIFT algorithm to construct a scale space of the satellite remote sensing data and the radar remote sensing data; Identifying initial image extreme points of the satellite remote sensing data and the radar remote sensing data according to the scale space, calculating the contrast of the initial image extreme points, and removing noise points according to the contrast to obtain characteristic points of the satellite remote sensing data and the radar remote sensing data; Taking the characteristic points as center points, calculating gradient amplitude directions of pixel points in a preset range of the center points, and constructing a gradient direction histogram; Determining the main direction of each feature point according to the gradient direction histogram, and drawing the image features of the satellite remote sensing data and the radar remote sensing data according to the feature points and the main direction to obtain the satellite remote sensing features and the radar remote sensing features of the target reservoir hydro-fluctuation belt; The Euclidean distance between each satellite remote sensing feature and the radar remote sensing feature is calculated, the similarity between each satellite remote sensing feature and the radar remote sensing feature is determined according to the Euclidean distance, and a similarity matrix is constructed; Feature matching is carried out according to the similarity matrix, feature pairs with similarity larger than a preset similarity threshold value are identified, and a feature matching set is constructed; And carrying out spatial registration on the satellite remote sensing data and the radar remote sensing data according to the feature matching geometry, and carrying out fusion operation on the spatially registered satellite remote sensing data and the radar remote sensing data to obtain fused remote sensing data.
3. The reservoir hydro-fluctuation belt recovery method based on multi-source remote sensing data fusion according to claim 2, wherein the fusion operation is performed on spatially registered satellite remote sensing data and radar remote sensing data to obtain fused remote sensing data, specifically: obtaining a standard ground object remote sensing image dataset, and labeling the ground object name of the standard ground object remote sensing image dataset to obtain labeling data; Constructing a ground object recognition model based on a convolutional neural network, constructing an input layer, a convolutional layer, a pooling layer and a full-connection layer of the ground object recognition model, determining a loss function and an optimizer of the model, and importing the labeling data into the ground object recognition model for training; Dividing the satellite remote sensing data and the radar remote sensing data subjected to spatial registration into a plurality of grid areas according to preset sizes, and importing the satellite remote sensing data and the radar remote sensing data of each grid area into a ground object recognition model after training is completed to perform ground object recognition to obtain a satellite remote sensing-ground object recognition result and a radar remote sensing-ground object recognition result of each grid area; Comparing the satellite remote sensing-ground object recognition result and the radar remote sensing-ground object recognition result of each grid area, and judging the ground object recognition consistency of the satellite remote sensing data and the radar remote sensing data in the same grid area to obtain ground object recognition consistency data; Dividing each grid region into a ground feature consistency region and a ground feature non-consistency region according to the ground feature identification consistency data, carrying out normalization processing on multispectral wave bands of satellite remote sensing data based on Brovey transformation fusion method when the grid region is the ground feature consistency region, multiplying the multispectral wave bands after normalization processing by pixel values of corresponding positions of image data of radar remote sensing data to obtain a pixel product matrix, and carrying out stretching change on the pixel product matrix to obtain a first remote sensing data fusion scheme; When the grid area is a ground object non-consistency area, identifying pixel noise points of satellite remote sensing data and radar remote sensing data based on a mean value filtering algorithm, using remote sensing data with fewer pixel noise points as reference data, calibrating the other remote sensing data as supplementary data, acquiring pixel noise point positions of the reference data, and replacing pixels of the supplementary data with pixels of the pixel noise point positions to obtain a second remote sensing data fusion scheme; and carrying out fusion operation on the satellite remote sensing data and the radar remote sensing data according to the first remote sensing data fusion scheme and the second remote sensing data fusion scheme to obtain fused remote sensing data.
4. The method for recovering the water-level-fluctuating zone of the reservoir based on the multi-source remote sensing data fusion according to claim 1, wherein the method is characterized in that the ecological change characteristics of the target water-level-fluctuating zone of the reservoir in a preset time period are extracted according to the fused remote sensing data, and the ecological evaluation is carried out on the target water-level-fluctuating zone of the reservoir according to the ecological change characteristics to obtain an ecological evaluation result in the preset time period, and specifically comprises the following steps: Sequencing the fused remote sensing data according to the acquired time node sequence, calculating the spectral feature vector of each pixel point in the fused remote sensing data of each time node, and constructing a spectral feature vector set; calibrating the fused remote sensing data of the first time node as a reference time phase, and calculating the spectrum angle difference between the spectrum feature vector of each spectrum feature vector set and the spectrum feature vector of the pixel point of the reference time phase; calculating a change vector between a spectral feature vector of each pixel point in the fused remote sensing data of each time node and a spectral feature vector of a corresponding pixel point in a reference phase based on a change vector analysis method, and calculating the amplitude and the direction of the change vector according to the change vector; Identifying ground feature data fused with remote sensing data of each time node based on a ground feature identification model, determining ground feature change characteristics of the ground feature data in a preset time period according to the spectrum angle difference, the change vector amplitude and the direction, and determining ecological change characteristics of a target reservoir hydro-fluctuation belt according to the ground feature change characteristics; And evaluating the ecological change condition of the target reservoir hydro-fluctuation belt within a preset time period according to the ecological change characteristics to obtain an ecological evaluation result.
5. The method for recovering the water-level-fluctuating zone of the reservoir based on multi-source remote sensing data fusion according to claim 1, wherein the optimal time interval for predicting the ecological recovery of the target water-level-fluctuating zone of the reservoir according to the ecological assessment result is obtained, and the ecological recovery prediction time is specifically: sequencing the ecological assessment results according to a time sequence, and constructing ecological time sequence data of a target reservoir hydro-fluctuation belt; constructing an ecological prediction model based on an ARIMA algorithm, drawing an ACF graph and a PACF graph according to the ecological time sequence data, and determining parameter settings of the ecological prediction model according to the ACF graph and the PACF graph, wherein the parameters comprise a partial autocorrelation coefficient, an autoregressive order and a moving average order; Importing the ecological time sequence data into the ecological prediction model for model fitting and training, acquiring real-time ecological assessment data of the current preset time length of the target reservoir hydro-fluctuation belt, importing the real-time ecological assessment data into the ecological prediction model, and predicting ecological changes of the future preset time length of the target reservoir hydro-fluctuation belt to obtain an ecological change prediction result; And marking a plurality of time intervals showing ecological forward succession of the target reservoir hydro-fluctuation belt according to the ecological change prediction result, and selecting the time interval with the best ecological condition of the target reservoir hydro-fluctuation belt as the optimal time interval for carrying out ecological restoration on the target reservoir hydro-fluctuation belt to obtain ecological restoration prediction time.
6. The method for recovering the water-level-fluctuating zone of the reservoir based on multi-source remote sensing data fusion according to claim 1, wherein the method for obtaining the periodic ecological variation demand data of the water-level-fluctuating zone of the target reservoir and determining the ecological recovery scheme of the water-level-fluctuating zone of the target reservoir according to the periodic ecological variation demand data and the ecological recovery prediction time is specifically as follows: Acquiring periodic ecological change demand data of a water-level-fluctuating zone of a target reservoir; Determining vegetation flooding conditions of the target reservoir hydro-fluctuation belt according to the periodical ecological variation demand data, selecting vegetation species according to the flooding conditions, and determining plant planting types of ecological restoration of the target reservoir hydro-fluctuation belt; And constructing an ecological restoration scheme of the target reservoir hydro-fluctuation belt according to the plant planting types and the ecological restoration prediction time period.
7. The reservoir hydro-fluctuation belt recovery system based on the multi-source remote sensing data fusion is characterized by comprising a storage and a processor, wherein the storage comprises a reservoir hydro-fluctuation belt recovery method program based on the multi-source remote sensing data fusion, and when the reservoir hydro-fluctuation belt recovery method program based on the multi-source remote sensing data fusion is executed by the processor, the following steps are realized: acquiring multi-source remote sensing data of a preset time node in a preset time period of a target reservoir hydro-fluctuation belt, wherein the multi-source remote sensing data comprises radar remote sensing data and satellite remote sensing data, and performing fusion operation on the multi-source remote sensing data to obtain fused remote sensing data; Extracting ecological change characteristics of the target reservoir hydro-fluctuation belt within a preset time period according to the fused remote sensing data, and carrying out ecological assessment on the target reservoir hydro-fluctuation belt according to the ecological change characteristics to obtain an ecological assessment result within the preset time period; predicting an optimal time interval for carrying out ecological restoration on the hydro-fluctuation belt of the target reservoir according to the ecological assessment result to obtain ecological restoration prediction time; and acquiring periodic ecological change demand data of the target reservoir hydro-fluctuation belt, and determining an ecological restoration scheme of the target reservoir hydro-fluctuation belt according to the periodic ecological change demand data and the ecological restoration prediction time.

Description

Reservoir hydro-fluctuation belt recovery method and system based on multi-source remote sensing data fusion Technical Field The invention relates to the technical field of data fusion, in particular to a reservoir hydro-fluctuation belt recovery method and system based on multi-source remote sensing data fusion. Background The reservoir hydro-fluctuation belt refers to a special area formed by reservoir due to quaternary water level fluctuation, has periodic water level change characteristics, causes extremely complex ecological environment in the area, and often has the problems of vegetation degradation, soil erosion, biodiversity reduction and the like. The ecological restoration of the reservoir hydro-fluctuation belt has important significance for maintaining the ecological balance of the area, improving the quality of water and promoting the biodiversity. However, due to the dynamics and complexity of the reservoir hydro-fluctuation belt ecosystem, its ecological restoration work faces many challenges. Along with the development of remote sensing technology, multi-source remote sensing data (such as satellite remote sensing data and radar remote sensing data) provides a new technical means for ecologically monitoring and recovering the hydro-fluctuation belt of the reservoir. The multi-source remote sensing data has the characteristics of wide observation range, high acquisition frequency, rich information dimension and the like, and can provide refined and dynamic ecological information for the reservoir hydro-fluctuation belt. However, it is difficult for a single type of remote sensing data to comprehensively reflect ecological characteristics of a hydro-fluctuation belt, for example, satellite remote sensing data has weak observation capability in a cloud coverage area, and radar remote sensing data has limitations in terms of ground object classification accuracy. Therefore, the multisource remote sensing data are fused to improve the space-time resolution and the information accuracy of the data, and the multisource remote sensing data become an important research direction for solving the ecological restoration problem of the hydro-fluctuation belt of the reservoir. In the prior art, the fusion method of the multisource remote sensing data still has the defects in the aspects of information extraction and analysis, such as low accuracy in recognition and registration of the features of the ground object, single extraction method of the ecological change features, lack of an effective ecological restoration time prediction model and the like. These deficiencies restrict the application effect of the remote sensing technology in ecological restoration of the hydro-fluctuation belt of the reservoir. Therefore, a comprehensive method based on multi-source remote sensing data fusion is needed, which can accurately extract the ecological change characteristics of the hydro-fluctuation belt of the reservoir, carry out scientific ecological assessment and time prediction, and formulate a reasonable recovery scheme in combination with ecological recovery requirements so as to promote the ecological restoration work of the hydro-fluctuation belt of the reservoir. Disclosure of Invention In order to solve at least one technical problem, the invention provides a reservoir hydro-fluctuation belt recovery method and system based on multi-source remote sensing data fusion. The first aspect of the invention provides a reservoir hydro-fluctuation belt recovery method based on multi-source remote sensing data fusion, which comprises the following steps: acquiring multi-source remote sensing data of a preset time node in a preset time period of a target reservoir hydro-fluctuation belt, wherein the multi-source remote sensing data comprises radar remote sensing data and satellite remote sensing data, and performing fusion operation on the multi-source remote sensing data to obtain fused remote sensing data; Extracting ecological change characteristics of the target reservoir hydro-fluctuation belt within a preset time period according to the fused remote sensing data, and carrying out ecological assessment on the target reservoir hydro-fluctuation belt according to the ecological change characteristics to obtain an ecological assessment result within the preset time period; predicting an optimal time interval for carrying out ecological restoration on the hydro-fluctuation belt of the target reservoir according to the ecological assessment result to obtain ecological restoration prediction time; and acquiring periodic ecological change demand data of the target reservoir hydro-fluctuation belt, and determining an ecological restoration scheme of the target reservoir hydro-fluctuation belt according to the periodic ecological change demand data and the ecological restoration prediction time. In this scheme, obtain the multisource remote sensing data of target reservoir hydro-fluctuation belt preset time node in t