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 time-consuming 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.