A phenology based geo-informatics approach to map land use and land cover (2003 - 2013) by spatial segregation of large heterogenic river basins

Abstract

Accurate information on land use and land cover (LULC) is critical for policy decisions especially for management of land and water resources’ activities in large river basins around the world. Phenology based LULC classification is the most promising approach particularly in the areas with diversified cropping patterns. Sometimes in large river basins, local climate and topography provides two different phenological information sets for the same crops in the same season. Based on accurate phenological information of the main crops in spatially segregated units, the remote sensing based classification was used to map the LULC changes for a period of 2003e2013 in the Kabul River Basin (KRB) of Afghanistan. We used remotely sensed Normalized Difference Vegetation Index (NDVI) products of Moderateresolution Imaging Spectroradiometer (MODIS) from Terra (MOD13Q1) and Aqua (MYD13Q1) with 250 m spatial resolution for this study. The overall accuracy (mean) of the LULC classification throughout the study period was around 68.15% ± 9.45while the producer and user accuracies (mean) were 75.9 ± 11.3% and 76.4 ± 11.2%, respectively. Results show that the cropping patterns vary significantly in the spatially disaggregated units. From 2003 till 2013, the ground coverage of wheat, barley and rice was increased by 31%, 7% and 32%, respectively. Overall, there has been only 2% increment in the agricultural area across the KRB between 2003 and 2013. This relatively increased trend of land cover change has taken place as a result of partial improvement in political stability as well as investment in irrigation infrastructure and agricultural development in the region. This study further provides insight to develop new agriculture strategies in order to maintain the ecosystem required to fulfil the rising food demands.