Detecting and mapping mature, medium, and young age date palms in the arid lands of Abu Dhabi, using hierarchical integrated approach (HIA)

Abstract

The main objective of the current study was to map dispersed date palm (DP) plantations in the Emirate of Abu Dhabi/United Arab Emirates (UAE), at different age stages: young (<5 years), medium (5–10 years), and mature (≥10 years), using multi-source/multi-sensor data in hierarchical integrated approach (HIA) developed and implemented for the purpose of this study. We initially succeeded in delineating and mapping “mature” DP plantations with acceptable accuracy using Landsat-8 OLI data. However, at this stage, our approach failed to depict “young” and “medium” DP with such level of accuracy, because of inadequate sensor resolution at such level of detail. Consequently, an object-based classification (OBC) approach was implemented using sub-meter WorldView-2 (WV-2) imagery, to depict and map medium and young aged palms in targeted DP plantations. Geographic Information Systems (GIS) was integrated to convert from raster to vector formats, allowing for manual editing of certain young DP polygons hence, increasing the accuracy of their delineation. The outcomes of the implemented approach were the creation of detailed and accurate maps of DP at the three age stages, for the first time in the region. The produced maps were validated using existing ancillary data and field checks. The overall accuracies for young, medium, and mature DP plantations were 86.8%, 88% and 90.7%, respectively; while for mixed-ages DP the value reached up to 94.5%, with an Overall Kappa Statistic estimated at 0.888.