CN-121978032-A - Grassland desertification multispectral remote sensing monitoring system and method

CN121978032ACN 121978032 ACN121978032 ACN 121978032ACN-121978032-A

Abstract

The invention discloses a grassland desertification multispectral remote sensing monitoring system and a grassland desertification remote sensing monitoring method, which relate to the technical field of remote sensing images, wherein the system comprises a sequential image preprocessing registration module, a pixel level object trace construction module, a healthy grassland object baseline modeling module, a trace cooperative comparison module and a desertification dynamic quantification module; the corresponding monitoring method sequentially comprises five steps of time sequence image preprocessing registration, pixel level object candidate track construction, healthy grassland object candidate base line modeling, track cooperative comparison and desertification dynamic quantification. According to the invention, the pixel-by-pixel material track is constructed by improving the monitoring refinement degree, collecting the multispectral time sequence remote sensing image, and the layered structured material weather baseline model is combined to truly restore the original growth rule of different pixels, so that the pixel scale monitoring is further performed, the data support is provided for the desertification feature analysis, the desertification comprehensive judgment and the dynamic quantification are realized, the desertification grade is divided, the technical support is provided for the desertification control work, and the decision efficiency and the implementation effect are improved.

Inventors

WANG LIJUAN
HU DIE
SHA SHA
JIANG XIAOYU

Assignees

中国气象局兰州干旱气象研究所

Dates

Publication Date: 20260505
Application Date: 20260409

Claims (10)

1. A grassland desertification multispectral remote sensing monitoring system is characterized by comprising a time sequence image preprocessing registration module, a pixel level object trace construction module, a healthy grassland object baseline modeling module, a trace cooperative comparison module and a desertification dynamic quantification module; The time sequence image preprocessing registration module is used for collecting multispectral time sequence remote sensing images of a monitoring area, executing preprocessing and space registration operation and outputting a registered time sequence image set; The pixel-level weathertrack construction module is used for receiving the registered time sequence image set, calculating a vegetation coverage index by utilizing a remote sensing image pixel-level spectrum analysis algorithm, constructing a weathertrack of each pixel by adopting a time sequence analysis technology, outputting a pixel-level weathertrack set comprising a historical healthy grassland track and a target growth season real-time track, and outputting weatherparameters at the same time; The healthy grassland physical condition baseline modeling module is used for receiving historical healthy grassland tracks, screening healthy grassland pixel tracks, performing fitting optimization and constructing a pixel-level physical condition baseline model; The track collaborative comparison module is used for receiving the target growth season real-time track and the pixel level weathered baseline model, calculating the collaborative similarity distance between the target growth season real-time track and the pixel level weathered baseline model by utilizing a track collaborative similarity algorithm, extracting the offset of the weathered parameters by adopting a dynamic time warping technology, and outputting a collaborative similarity distance data set and a weathered parameter offset data set; The desertification dynamic quantization module is used for receiving the cooperative similarity distance data set and the weathers parameter offset data set, dividing the desertification grade by utilizing the desertification dynamic quantization rule, extracting the desertification occurrence time and the desertification evolution rate, and outputting a desertification monitoring map and a statistical report.
2. The grassland desertification multispectral remote sensing monitoring system according to claim 1 is characterized in that the sequential image preprocessing registration module is used for collecting multispectral sequential remote sensing images of a monitoring area and executing preprocessing and spatial registration operation, and the process of outputting a registered sequential image set is used for collecting multispectral sequential remote sensing images covering the monitoring area according to the geographic position of the monitoring area and the time span of a grassland growing season to form an initial multispectral sequential remote sensing image set, preprocessing each image in the initial multispectral sequential remote sensing image set, eliminating errors generated in the image collecting process and obtaining the preprocessed multispectral sequential remote sensing image set, selecting a geographic reference image, and performing spatial registration on all images in the preprocessed multispectral sequential remote sensing image set and the geographic reference image to ensure that all images are unified to the same geographic coordinate system, so as to obtain the registered sequential image set.
3. The grassland desertification multispectral remote sensing monitoring system according to claim 1, wherein a calculation formula of a remote sensing image pixel-level spectrum analysis algorithm in the pixel-level weathertrack construction module is as follows: , wherein, Is a vegetation coverage index; Normalized adjustment coefficients for pixel-level spectral analysis; the near infrared band reflectivity of a single pixel in the monitoring area is obtained; the red wave band reflectivity of the same single pixel in the monitoring area is obtained; The blue band reflectivity of the same single pixel in the monitoring area; The atmospheric scattering correction coefficient is the red wave band; The atmospheric scattering correction coefficient is blue wave band; And (5) regulating the coefficient for the soil background.
4. The grassland desertification multispectral remote sensing monitoring system of claim 1 is characterized in that a time sequence analysis technology is adopted in a pixel-level weathertrack construction module to construct a weathertrack of each pixel, a pixel-level weathertrack set comprising a historical healthy grassland track and a target growth season real-time track is output, and the process of outputting weatherparameters is that a vegetation coverage index obtained through calculation is integrated into time sequence original data of a pixel-by-pixel vegetation coverage index, the time sequence preprocessing is carried out on the time sequence original data, the preprocessed time sequence original data is fitted by adopting the time sequence analysis technology to construct the weathertrack of each pixel, the historic period track and the target growth season track are extracted from all the constructed weathertracks, the historical healthy grassland track is obtained through health screening, the target growth season real-time track is extracted and integrated into a pixel-level weathertrack set, the weatherparameters representing vegetation growth rules are extracted from each track, and the pixel-level weathertrack set comprising the historic healthy grassland track and the target growth season real-time track is output.
5. The grassland desertification multispectral remote sensing monitoring system is characterized in that a pixel level object-level baseline model in the healthy grassland object-level baseline modeling module is a layered structured model and specifically comprises a basic track layer, a fitting optimization layer and a characteristic mapping layer, wherein the basic track layer is a historical healthy grassland pixel track set which is reserved after health screening and accords with grassland original growth rules and is stored in a corresponding mode one by one according to pixel space positions, the fitting optimization layer is used for fitting and optimizing standardized object-level tracks formed after fitting and optimizing according to track data of the basic track layer, and the characteristic mapping layer is used for establishing a mapping relation between the standardized object-level tracks after fitting and the topography, soil and grassland type characteristics of corresponding pixels.
6. The grassland desertification multispectral remote sensing monitoring system is characterized in that a track collaborative similarity algorithm is utilized in the track collaborative comparison module, the specific steps of calculating the collaborative similarity distance between a target growth season real-time track and a pixel level object-weather base line model are that the target growth season real-time track and a single pixel reference track in the pixel level object-weather base line model are synchronous and regular according to time dimension, track data dimension is unified, grassland pixels in a monitoring area are subjected to pixel-by-pixel extraction of object-weather feature sequences of two groups of tracks based on object-weather parameters, the object-weather feature sequences are feature sequences formed by arranging the object-weather parameters according to time dimension to form track feature data corresponding one by one, similarity quantification calculation is conducted on the two groups of track feature data of each pixel by the track collaborative similarity algorithm to obtain deviation values of the real-time track and the reference track under each pixel, and finally the deviation values of all pixels are integrated according to geographic space position to generate a collaborative similarity distance data set.
7. The grassland desertification multispectral remote sensing monitoring system of claim 6, wherein the calculation formula of the locus synergy similarity algorithm in the locus synergy comparison module is: , wherein, Is the first The tracks of the pixels cooperate with the similarity distance; numbering individual grassland pixels in the monitored area; Is a collaborative weight coefficient; Is the first The Euclidean distance between the real-time track of each pixel and the baseline track; The maximum value of Euclidean distance of all pixels in the monitoring area; Is the first Cosine similarity of the real-time track of each pixel and the baseline track.
8. The grassland desertification multispectral remote sensing monitoring system is characterized in that a dynamic time warping technology is adopted in a track collaborative comparison module to extract offset of the weathers parameters, a collaborative similarity distance data set and a weathers parameter offset data set are output, the process comprises the steps of carrying out time sequence dimension warping on a target growth season real-time track and a pixel-level weathers baseline model track, aligning time sequence nodes of the two tracks pixel by adopting the dynamic time warping technology, extracting deviation conditions of the weathers parameters on the time sequence characteristics according to the aligned tracks, quantifying to form weathers parameter offset data, arranging the calculated collaborative similarity distances according to geographic space, generating the collaborative similarity distance data set, classifying the weathers parameter offset data according to corresponding pixels, forming the weathers parameter offset data set, and synchronously outputting the collaborative similarity distance data set and the weathers parameter offset data set.
9. The grassland desertification multi-spectrum remote sensing monitoring system according to claim 1 is characterized in that the concrete rule of dividing the desertification grade by using the desertification dynamic quantization rule in the desertification dynamic quantization module is that a collaborative similarity distance data set and a weathers parameter offset data set are received, pixel-by-pixel association matching is carried out on the two sets of data sets to obtain a standardized desertification analysis data set, the desertification grade is divided according to the desertification dynamic quantization rule, and the collaborative similarity distance and the weathers parameter offset are used as judging indexes, wherein the desertification grade is specifically no desertification, light desertification, moderate desertification and heavy desertification.
10. A grassland desertification multispectral remote sensing monitoring method, which is suitable for the grassland desertification multispectral remote sensing monitoring system as claimed in any one of claims 1 to 9, and is characterized in that the method comprises the following specific steps: s100, preprocessing and registering the time sequence images, namely acquiring multispectral time sequence remote sensing images according to the monitoring area and the span of the growing season, preprocessing to eliminate errors, registering the multispectral time sequence remote sensing images with a geographic reference image space, and outputting a time sequence image set of a unified coordinate system; S200, constructing pixel-level weathered trajectories, namely calculating vegetation coverage indexes according to a remote sensing image pixel-level spectrum analysis algorithm, constructing a weathered trajectory of each pixel through a time sequence analysis technology, outputting a pixel-level weathered trajectory set comprising a historical healthy grassland trajectory and a target growth season real-time trajectory, and outputting weathered parameters; S300, modeling a healthy grassland physical baseline, namely screening a healthy grassland pixel track and fitting and optimizing the healthy grassland pixel track to construct a layered pixel level physical baseline model comprising a basic track layer, a fitting and optimizing layer and a characteristic mapping layer; S400, track cooperative comparison, namely enabling the target growth season real-time track to be time-ordered with the baseline model track, calculating cooperative similarity distance through a track cooperative similarity algorithm, and outputting two types of data sets through dynamic ordered extract candidate parameter offset; and S500, performing dynamic desertification quantification, namely associating and matching two types of data sets, dividing four types of levels of no desertification, slight desertification, moderate desertification and severe desertification by taking the cooperative similarity distance and the weathers parameter offset as indexes, extracting the desertification occurrence time and the desertification evolution rate, and outputting a desertification monitoring map and a statistical report.

Description

Grassland desertification multispectral remote sensing monitoring system and method Technical Field The invention relates to the technical field of remote sensing images, in particular to a grassland desertification multispectral remote sensing monitoring system and a grassland desertification multispectral remote sensing monitoring method. Background The grassland is used as a core component of the land ecological system, has multiple functions of ecological protection, soil and water conservation, biological diversity maintenance and the like, and is a key carrier of an ecological safety barrier. However, the global grassland desertification problem is increasingly severe under the influence of factors such as climate change, excessive grazing and the like, and a series of chain reactions such as land productivity decline, ecological system degradation and the like are caused, so that serious threat is formed to regional ecological safety and sustainable development. Traditional grassland desertification monitoring relies on ground investigation, and has the defects of time consumption, labor consumption, limited coverage range, poor timeliness and the like, while multispectral remote sensing technology has become the mainstream technology of desertification monitoring by virtue of the advantages of large-scale, rapid and non-contact monitoring, but the existing monitoring scheme based on remote sensing still has difficulty in meeting the requirements of fine and dynamic monitoring. The current main-stream grassland desertification remote sensing monitoring technology has many defects that the analysis is carried out based on image data of single or few time nodes, continuous tracking of the vegetation growth full-period weathered track is lacked, dynamic evolution rules of the desertification process are difficult to reflect, the construction of a healthy grassland foundation line model is simpler, the integral averaging treatment is not adopted, the influence of differential features such as terrain, soil, grassland types and the like under the pixel scale is not considered, the baseline referential is insufficient, the track similarity calculation is dependent on a single distance index, the cooperative quantification of the Euclidean distance and the cosine similarity is not realized, the extraction of the weathered parameter offset is lack of effective time alignment technical support, the deviation is large, the desertification grade division is judged by adopting a single index, the comprehensive evaluation of the similarity distance and the weathered parameter offset is not integrated, the desertification occurrence time and the evolution rate cannot be accurately extracted, the monitoring precision is low, the dynamic quantification capability is weak, and the accurate decision of the desertification treatment is difficult to support. The defects in the prior art lead to the problems of low precision, insufficient dynamic property, insufficient comprehensive information and the like of the grassland desertification monitoring result, can not provide timely and accurate technical support for desertification control, and restricts the effect of ecological protection work. Therefore, development of a multispectral remote sensing monitoring system and a multispectral remote sensing monitoring method capable of realizing pixel-level weathertrack construction, layered baseline model optimization, multidimensional cooperative comparison and desertification dynamic quantification is urgently needed, and the refinement degree and dynamic tracking capability of desertification monitoring are improved by integrating key technologies such as time sequence image processing, time sequence analysis and dynamic normalization, so that scientific basis is provided for early warning, treatment and ecological restoration of grassland desertification. Disclosure of Invention The invention aims to make up the defects of the prior art, provides a grassland desertification multispectral remote sensing monitoring system and a grassland desertification remote sensing monitoring method, which can construct pixel-by-pixel object candidate tracks and screen healthy tracks through collecting multispectral time sequence remote sensing images and preprocessing consistent with registration guarantee data, realize the mapping of baselines and pixel characteristics by combining layered object candidate baseline models, divide desertification grades through track cooperative comparison and desertification dynamic quantization, extract desertification information and output reports, and provide scientific support for grassland desertification control. In order to solve the technical problems, the invention provides the technical scheme that on one hand, the grassland desertification multispectral remote sensing monitoring system comprises a time sequence image preprocessing registration module, a pixel level object and weather track