Abstract
Forests have significant impacts on the global carbon cycle, hydrological processes, and biodiversity. Driven by socioeconomic developments, forests experienced drastic changes since the mid-20th century in China. Although declassified spy satellite and other Earth observation satellite data offer remote sensing technologies for mapping these long-term changes, challenges remain unsolved for applications of large volumes of historical data. This study uses long-term satellite observations, including declassified satellite data in the 1960s and Landsat products since the 1970s to monitor the decadal changes. A semi-automated method was developed for the rapid registration of declassified images with reference to Landsat data. The method was applied to quantify the forest cover (FC) in Sichuan Province (excluding Chongqing), China. Combined with a Landsat-based FC change product, it revealed that the FC in Sichuan declined rapidly by 38% from the 1960s to 2005. The FC was estimated to be 45.19 ± 1.62% in the 1960s and 38.98 ± 2.06% in 1975, but it rapidly decreased to 28.91 ± 2.07% in 1990 and 27.87 ± 2.14% by 2005. Supplemented with the official statistics, the FC in Sichuan was reported to increase to 38.03% by 2018. Although differences between the remote sensing-based estimates and the statistics were observed, they highlight the challenges in reconstructing historical land use changes for carbon and other studies. The drastic loss of forests before 1990 and the stabilizing afterward reflects the changes in forest policies, which transitioned from serving timber products to forest conservations.
Highlights
China has undergone frequent social and economic changes that has led to rapid economic development and increased consumption of natural resources since the 1940s
In contrast to the feature points usually selected by experts through visual interpretation, SIFT did not tend to select features such as river turning points, mountain ridges, or valleys; instead, the algorithm selected features at multiple scales and finalized the location at the center of a multiscale feature
The SIFT features are not as “obvious” as manually selected features, the accuracy is usually high because SIFT reports the location of the feature at the subpixel scale, and the number of feature points is ensured to satisfy the requirements
Summary
China has undergone frequent social and economic changes that has led to rapid economic development and increased consumption of natural resources since the 1940s. Alarmed by the devastating floods in 1998, China launched the logging ban and two major conservation programs, such as the Natural Forest Conservation Program (NFCP) and the Grain to Green Program (GTGP), aimed at mitigating environmental degradation and conserving natural resources through afforestation activities [3]. Forest cover (FC) in China increased from 12.7% in 1976 to 20.36% in 2008, according to official forest surveys [4], [5]. Such efforts to increase forest and other vegetation types have contributed to a greening trend over the global vegetated areas since the 2000s [6]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
