Geospatial Technologies for Semiautomated Baseline Database Generation for Large- Scale Land Resource Inventory

Abstract

The goal of land resource inventory is to enable the lab-to-land transfer of agro-technology on a sustainable basis through identification of homogeneous soil management units. The identification of homogeneous landscape ecological unit (LEU) boundaries for soil mapping through conventional methods is timeconsuming and laborious. Hence, it is necessary to develop a semiautomated geospatial framework for delivering reliable soil resource information to the users on time. In the present chapter, the approach for semiautomation in landform delineation using high-resolution IRS Cartosat-1 and LISS-IV data was discussed. Cartosat-1 stereopair data are processed to generate the digital terrain model (DTM) of 10 m spatial resolution. The digital terrain analysis was carried out to generate contour, drainage, slope, and hillshade for landform delineation in two distinct terrain conditions. Object-based slope classification algorithm is developed by following USDA-NRCS slope class thresholds to hasten the process of landform identification. The land use/land cover (LULC) map of the area is generated based on the rabi season data of Cartosat-1 merged LISS-IV (2.5 m) as well as high-resolution (0.5 m) public domain imagery at the backend so as to get the reliable land use boundary at cadastral level through feature optimization algorithm in eCognition software using near-infrared (NIR) and Normalized Difference Vegetation Index (NDVI) data. The integration of three secondary layers, i.e., landform, slope, and LULC, are achieved through the hierarchical object-based segmentation algorithm to develop landscape ecological unit (LEU) map. The logical automation algorithm developed at each stage assists in optimizing sampling intensity, which leads to a considerable saving of man power, labor, cost, and time.