Large area mapping of land-cover change in Rondônia using multitemporal spectral mixture analysis and decision tree classifiers

TLDR: In this paper, a multistage process was used to map primary forest, pasture, second growth, urban, rock/savanna, and water using 33 Landsat scenes acquired over three contiguous areas between 1975 and 1999.

Abstract: [1] We describe spatiotemporal variation in land cover over 80,000 km2 in central Rondonia. We use a multistage process to map primary forest, pasture, second growth, urban, rock/savanna, and water using 33 Landsat scenes acquired over three contiguous areas between 1975 and 1999. Accuracy of the 1999 classified maps was assessed as exceeding 85% based on digital airborne videography. Rondonia is highly fragmented, in which forests outside of restricted areas consist of numerous, small irregular patches. Pastures in Rondonia persist over many years and are not typically abandoned to second growth, which when present rarely remains unchanged longer than 8 years. Within the state, annual deforestation rates, pasture area, and ratio of second growth to cleared area varied spatially. Highest initial deforestation rates occurred in the southeast (Luiza), at over 2%, increasing to 3% by the late 1990s. In this area, the percentage of cleared land in second growth averaged 18% and few pastures were abandoned. In central Rondonia (Ji-Parana), deforestation rates rose from 1.2% between 1978 and 1986 to a high of 4.2% in 1999. In the northwest (Ariquemes), initial deforestation rates were lowest at 0.5% but rose substantially in the late 1990s, peaking at 3% in 1998. The ratio of second growth to cleared area was more than double the ratio in Luiza and few pastures remained unchanged beyond 8 years. Land clearing was most intense close to the major highway, BR364, except in Ariquemes. Intense forest clearing extended at least 50 km along the margins of BR364 in Ji-Parana and Luiza. Spatial differences in land use are hypothesized to result from a combination of economic factors and soil fertility.