A semi-automatic cropland mapping approach using GEOBIA and random forests on black-and-white aerial photography

Abstract

For decades land-use and land-cover (LULC) conversions have had an important impact on land- and ecosystem degradation, accordingly (historical) LULC information is important for the assessment of such impacts. This information can be derived from black-and-white (B&W) aerial photography. Such photography is often visually interpreted, which is a very time-consuming approach. This study shows that machine learning can be applied on only brightness to derive LULC information. Cropland acreage is semi-automatically mapped by means of Geographic Object-Based Image Analysis (GEOBIA) and Random Forest classification in two study sites in Ethiopia and in The Netherlands. The result is a thematic map with two classes: 1) agricultural cropland and 2) other types of land cover. Overall mapping accuracies attained are 90 % and 96 % for the two study areas respectively. This mapping method increases the timeline at which historical cropland expansion can be mapped purely from brightness information in B&W photography up to the 1930s.