Method and device for optimizing the ecological security pattern of coastal wetlands based on machine learning

Abstract

The invention provides a method and device for optimizing the ecological security pattern of coastal wetlands based on machine learning. The method includes: using a support vector machine and a random forest method to perform pixel-scale supervised classification and extraction on the coastal wetlands of the target coast, so as to obtain coastal wetlands. Wetland extraction results; use the information theory-based patch shape recognition algorithm to identify the shape of each wetland patch in the coastal wetland extraction results; use variance analysis to compare the difference in the degree of ecological disturbance of wetland patches with different shapes, and obtain The optimal coastal wetland landscape type with statistical significance can obtain the landscape ecological security pattern pattern that maximizes ecological resilience. The invention can assist and guide the optimization of the coastal wetland landscape pattern, and provide landscape planning ideas and theoretical basis.

Description

technical field [0001] The invention relates to the field of marine ecology, in particular, to a method and device for optimizing the ecological security pattern of coastal wetlands based on machine learning. Background technique [0002] As a unique ecosystem formed by the interaction of land and sea, coastal wetlands have important ecological functions and values, and can provide a variety of ecosystem services such as carbon sequestration, biodiversity protection, wind and wave protection, leisure and recreation. The ecological security status of coastal wetlands will directly affect the security of the terrestrial and marine ecosystems connected to them, and is an important basis for realizing regional sustainable development. In recent decades, due to the influence of human social and economic activities, the increase of urban population and the rapid expansion of cities have resulted in the destruction or degradation of wetland resources in varying degrees in terms of ...

AI Technical Summary

Problems solved by technology

However, manual extraction of coastlines is time-consuming and laborious, and the extraction accuracy is affected by subjective factors and difficult to unify, so it is difficult to use this method for large-scale coastal wetland extraction.

[0007] 2. It is difficult to avoid human-subjective factors in the process of optimizing the landscape ecological security pattern, and it is impossible to achieve fully automated optimization

For example, in the identification of landscape corridors, existing methods often use the minimum cumulative resistance model, which requires artificially setting the weight of resistance surfaces; in addition, existing methods often use buffer analysis in demarcating research areas or species dispersal areas. The setting of the radius often relies on the qualitative and subjective judgment of the staff, lacking quantitative basis

[0008] 3. Lack of consideration of ecological resilience

When optimizing landscape patterns, the existing methods mostly consider the horizontal migration and diffusion of species, rarely introduce the concept of ecological resilience into the optimization of landscape patterns, and cannot fully understand and exert the resistance and resilience of the ecosystem itself to external disturbances.

Images