A continuous time series water body remote sensing mapping method

Abstract

The invention discloses a continuous time series water body remote sensing mapping method, proposes a remote sensing image interpretation training sample point automatic update method, time series consistency analysis and false change correction technology, and combines the GEE big data platform to establish a long time series High-precision water system mapping and dynamic monitoring methods solve the problem of time-consuming and labor-intensive continuous-time water body remote sensing mapping and contain a large number of false changes; training samples, which effectively improves the efficiency and accuracy of supervised classification, and at the same time has a high degree of automation and portability; at the same time, the present invention recognizes the time-series consistency of the year-by-year water body classification results, and combines the water body information within the year to establish a water body pseudo Change discrimination and correction rules effectively improve the accuracy of water body dynamic changes and provide accurate long-term water body dynamic information.

Description

technical field [0001] The invention relates to the technical field of water body remote sensing mapping, in particular to a continuous time series water body remote sensing mapping method. Background technique [0002] As an important part of natural productivity, wetlands provide various ecological service values, such as water conservation, flood regulation, migration of nutrients and sediments, and maintenance of biological habitat stability. Due to the impact of climate change and human activities, large areas of wetlands have been destroyed around the world, and the degradation and shrinkage of wetlands has led to extreme climate phenomena (such as floods and droughts) and the reduction of biodiversity. Wetlands are characterized by perennial or seasonal water bodies. Water body mapping based on remote sensing has become an important means of long-term water dynamic monitoring. It can accurately extract the spatial distribution of long-term water bodies and effectively...

AI Technical Summary

Problems solved by technology

[0004] However, in the two types of remote sensing mapping methods, the unsupervised classification method of water body index has problems such as threshold instability, inconsistent classification results of different water body indices, and underestimation of mixed pixels of water body; while the supervised classification method can achieve relatively good results in previous cases. High-precision water system mapping, but its accuracy obviously depends on the training samples. It is time-consuming and labor-intensive to obtain long-term training samples in large-scale areas, and it is difficult to meet the requirements of large-area water body change monitoring.

In addition, when the above two types of methods are applied to the long-term monitoring of water body changes in large areas, the superposition effect of errors generated by remote sensing images in different phases is often ignored, and these errors will be superimposed on the final long-time series mapping results, further affecting Improve the accuracy of water body monitoring changes

At present, there is a lack of methods for fast and accurate long-term water body change monitoring in large areas. The classification error in the image leads to a large number of false changes in the long-term water system map, which makes the water body change monitoring have great uncertainty.

[0005] Therefore, there is an urgent need in the market for a fast remote sensing mapping method for long-term series and dynamic monitoring of water bodies, to solve the time-consuming and laborious difficulties in obtaining training samples for water body remote sensing mapping, and to solve the low recognition accuracy of water body changes caused by classification errors of different phases the puzzle

Images