New progress has been made in intelligent remote sensing of complex forests in karst areas

Under the background of large-scale ecological protection and restoration, China's southwest karst region has become a hot spot for

Deep learning is considered a revolutionary technology in the field of machine learning, which has shown great potential

The results show that the model can classify and segment complex forests in high-resolution remote sensing images, and the overall accuracy of the study area includes Yunnan, Guangxi and Guizhou provinces, and the overall accuracy reaches 89.

The study further found that monoculture plantations can quickly improve regional ecosystem productivity, but also affect regional water balance, especially eucalyptus plantations, which are mainly distributed in southern Guangxi and northern

Studies have shown that eucalyptus plantations may have higher productivity and economic value, but our study also found that eucalyptus trees consume more water than other forest types, and the intelligent extraction and accurate identification of complex forests at the regional scale can help to better quantify the economic and ecological value of forests

The study shows how to combine the latest high-resolution satellite remote sensing images with the most advanced methods in the field of artificial intelligence and successfully apply them to the accurate identification of highly heterogeneous forest landscapes in southwest China, which is crucial

The research was supported by the National Key R&D Program, the National Natural Science Foundation of China, and the Pilot A of the Chinese Academy of Sciences

Links to papers

Spatial identification and mapping of complex forest types in southwest China

Temporal and spatial differentiation characteristics of different forest productivity and evapotranspiration in southwest China