Brief note on using geomatics to study land-cover change in the Tarai since the 1950s

Abstract

This note addresses the use of geomatics for studying the changes in land-cover in the Tarai since the 1950s. The author explains the main principles of geomatics (for social scientists) and shows how certain geomatic methods contribute to addressing this theme. After a presentation of the inventory of sources – digitized maps and Landsat and Spot satellite images for the most part – the methodology is explained. The latter is based on the integration into a small geographic information system (GIS), of various finely reworked, georeferenced maps, both raster and vector, that ultimately show land-cover in the Tarai on various scales and at different periods. These maps can be the result of satellite-image classifications using various remote-sensing techniques, and in particular pixel-supervised classifications used here and which identify spatial objects based on their known spectral signatures. However, while our land-cover classifications are fairly accurate at district level, they are less accurate at local level. Indeed, it is difficult to individualise and map small objects by pixel classification, such as rural dwellings in the Tarai that can be mistaken for bare or harvested fields, even on images of a very high spatial resolution. The use of manual vectorisation of these small space objects partly compensates for these inaccuracies.

Keywords

• Tarai
• Geomatics
• GIS
• supervised classification
• Landsat
• SPOT 5