Using hyperspectral reflectance to detect different soil erosion status in the Subtropical Hilly Region of Southern China: a case study of Changting, Fujian Province
2013
Lin, Chen | Zhou, Sheng-Lu | Wu, Shaohua
Hyperspectral reflectance is widely used for determining important properties of soil erosion. However, there have been few studies which focus on the influence of soil erosion intensity on the characteristics of hyperspectral reflectance, and such information would provide a new tool to improve quantitative understanding of soil erosion. In this study, 35 soil samples were collected from three regions with different erosion intensities in Changting County, a typical severely eroded county in the ferralic cambisol region of southern China, and classified into three groups according to different erosion controlling status. All the samples were scanned at wavelengths from 400 to 2,498 nm by an ASD Field Spec Portable Spectrometer, and the erosion intensity of each sample was calculated using the Revised Universal Soil Loss Equation. Multivariate stepwise linear regression was then employed to model the soil erosion intensity based on reflectance. The results suggested that the absorption peaks of each sample were in a similar wavelength range, while the absorption depth varied with different erosion status, and the reflectance of extremely eroded soil samples were the highest. During modelling of erosion intensity, the result was poor when all the samples were combined, but improved greatly at certain wavelength ranges when samples were classified into three groups based on different erosion controlling status. The extreme erosion group markedly outperformed the other two groups, in which the R²values between the actual and predicted erosion intensity were 0.67, 0.85 and 0.80 for each spectral type. The results indicated that hyperspectral reflectance is a promising method for accurately monitoring erosion intensity.
Показать больше [+] Меньше [-]Ключевые слова АГРОВОК
Библиографическая информация
Эту запись предоставил National Agricultural Library