Antananarivo - 2017 Land cover map

Abstract

The map proposed here is a key data to understand urban agriculture’s dynamics in relation to urbanization in Antananarivo agglomeration and agriculture’s contribution to the city food system while debates are ongoing to update Antananarivo city master plan. This land cover map was obtained using the Moringa treatment chain, a prototype chain developed by CIRAD (TETIS Joint Research Unit). It is based on the combined use of a very high spatial resolution Pleiades image acquired on 8 January 2017 by the ISIS system of the CNES (French Space Agency) and high spatial image time series Sentinel-2 and Landsat-8 for a period from October 2016 to September 2017, in a classification approach combining OBIA (object based image analysis) and data mining (Random Forest) trained by a reference database collected in situ. The processing is done with free software (Orfeo Toolbox and Python). The field database consists of field surveys conducted with GPS device. The surveys were conducted in 2017 during the raining season. Outline of the plots were drawn on the Pleiades image. This database of 3068 plots was then divided in two: 80% of the plots per class were randomly selected to form the learning database and the remaining 20% was selected to form the validation database. This map was developed within the frame of a research project (“Legend project”) supported by the metaprogramme GloFoodS (Transitions to global food security) jointly conducted by INRA and Cirad, the two leading French research institutions for agriculture (www.glofoods.inra.fr). The objective of our study was to test an innovative mapping method in an area of great complexity where urban and agriculture are intertwined : the Antananarivo’s agglomeration.