Land cover change dynamics mapping and predictions using EO data and a GIS-cellular automata model: the case of Al-Baha region, Kingdom of Saudi Arabia

摘要

The present study designed to monitor and predict land cover change (LCC) in addition to characterizing LCC and its dynamics over Al-Baha region, Kingdom of Saudi Arabia, by utilizing remote sensing and GIS-cellular automata model (Markov-CA). Moreover, to determine the effect of rainwater storage reservoirs as a driver to the expansion of irrigated cropland. Eight Landsat 5/7 TM/ETM images from 1975 to 2010 were analyzed and ultimately utilized in categorizing LC. The LC maps classified into four main classes: bare soil, sparsely vegetated, forest and shrub land, and irrigated cropland. The quantification of LCC for the analyzed categories showed that bare soil and sparsely vegetated was the largest classes throughout the study period, followed by forest, shrubland, and irrigated cropland. The processes of LCC in the study area were not constant, and varied from one class to another. There were two stages in bare soil change, an increase stage (1975-1995) and decline stage (1995-2010), and the construction of 25 rainwater-harvesting dams in the region was the turning point in bare soil change. The greatest increase was observed in irrigated cropland after 1995 in the expense of the other three categories as an effect of extensive rainwater harvesting practices. Losses were evident in forest and shrubland and sparsely vegetated land during the first stage (19751995) with 5.4 and 25.6 % of total area in 1995, while in 1975, they covered more than 13.8 and 32.7 % of total area. During the second stage (1995-2010), forest and shrubland witnessed a significant increase from 1569.17 km(2) in 1975 to 1840.87 km(2) in 2010. Irrigated cropland underwent the greatest growth (from 422.766 km(2) in 1975 to 1819.931 km(2) in 2010) during the entire study period, and this agriculture expansion reached its zenith in the 2000s. Markov-CA simulation in 2050 predicts a continuing upward trend in irrigated cropland and forest and shrubland areas, as well as a downward trend in bare soil and sparsely vegetated areas; the spatial distribution prediction indicates that irrigated cropland will expand around reservoirs and the mountain areas. The validation result showed that the model successfully identified the state of land cover in 2010 with 97 % agreement between the actual and projected cover. The output of this study would be useful for decision makers and LC/land use planners in Saudi Arabia and similar arid regions.