2019 Volume 35 Issue 4 Pages 61-68
Establishment of an early monitoring system is essential to minimize drought damage to agricultural products. Spatiotemporal analysis of both meteorological and agricultural land use conditions can be used for this purpose by collecting time series remotely sensed images over multiple-seasons. The vulnerability of agricultural farmlands to drought varies over agricultural areas according to the different types of cropping systems. In the present study, the cropping patterns of various agricultural areas on Lombok Island, in the eastern part of Indonesia were first classified, and then the monthly precipitation changes over fifteen years were analyzed. Eight types of cropping systems were identified in the agricultural areas based on an image classification using synthetic Landsat8 OLI (Operational Land Imager) and MODIS (MODerate resolution Imaging Spectroradiometer) images. An accuracy assessment was carried out using training samples data collected in the field surveys, together with interpretation of high-resolution satellite images. The total accuracy was 74.53%, with a Kappa coefficient of 0.67. Next, the above-normal, normal, and subnormal rainfall per year were selected based on the fifteen years monthly rainfall data of TRMM (Tropical Rainfall Measuring Mission), and then the Vegetation Health Index (VHI) derived from the MODIS images was then analyzed for the eight classified types of cropping systems in the agricultural areas for each case of years. As a result, a high possibility to drought was identified in the areas where double cropping paddy fields and single cropping dry fields in rainy season. By contrast, areas with triple cropping paddy fields and dry fields throughout a year had a relatively low possibility to drought in this study area.