关键词: 水土流失, 裸土, 遥感, 特征信息提取, 长汀县

Abstract:

Changting County of Fujian Province is one of the most typical reddish soil loss regions in southern China. Serious soil erosion has resulted in environmental and ecosystem degradation in the county. Local people and governments have been struggling for the control of the soil loss for more than two decades. To investigate and assess the changes after the effort, this study used remote sensing technology to analyze spatiotemporal dynamics of the soil exposure degree in the county during the period from 1988 to 2010. The bare soil features were estimated from the Landsat TM/ETM+ images of 1988, 1999, 2004 and 2010 based on two thematic indices-normalized difference soil index (NDSI) and normalized difference impervious surface index (NDISI). The used four images were geometrically corrected and registered to allow spatial overlay analysis between images. After the geometrical correction, the digital number (DN) values of the images were converted to at-satellite reflectance. The NDSI was used to enhance and extract soil features from the four used images with proper threshold values. However, the extracted soil features were found to have been mixed with built-up land features due largely to the spectral confusion between the two categories and the similar result in the NDSI performance. In order to improve the accuracy of the extracted soil feature, the NDISI was further used to filter out the unwanted built-up land features from the extracted soil features using a logical tree algorithm, as the NDISI is the index specially designed for the highlight of impervious surface features represented mainly by built-up lands. This significantly improved the estimate of bare soil and the extraction accuracy can reach up to 95.95%. Based on the extracted bare soil features, the exposure degree of bare soil and spatial and temporal changes of bare soil were analyzed. Results showed that the 22-year fight for treating soil loss has greatly reduced exposed bare soils in Changting County. The area of bare soil has been greatly reduced from 175.5 km² in 1988 to 85.6 km² in 2010, bare-soil exposure intensity dropped down from 0.487 to 0.383, and fragmentation degree of bares soil increased from 0.103 to 0.207. In the two observed periods (1988-1999 and 1999-2010), the decrease in bare soil area was in an accelerate speed, which was -3.74 km² per year during the 1988-1999 period and -4.34 km² per year in the duration between 1999 and 2010. This corresponds closely to the two major policies issued by the Fujiang Provincial Govenment specially for the county’s soil loss treatment. Obviously, government policies accompanied with effective actions were the major driving forces to the reduction of the exposure degree of bare soil in the county during the study period.

Key words: soil loss, bare soil, remote sensing, feature extraction, Changting County