B41D-0470
Multi-temporal UAV based data for mapping crop type and structure in smallholder dominated Tanzanian agricultural landscape
Thursday, 17 December 2015
Poster Hall (Moscone South)
Jyoteshwar R Nagol1, Caspar Chung2, Jan Dempewolf3, Sixbert Maurice4, Winfred Mbungu5 and Siza Tumbo4, (1)University of Maryland College Park, College Park, MD, United States, (2)Northeastern University, Boston, MA, United States, (3)University of Maryland, College Park, MD, United States, (4)Sokoine University of Agriculture, Morogoro, Tanzania, (5)Virginia Polytech State University, Blacksburg, VA, United States
Abstract:
Timely mapping and monitoring of crops like Maize, an important food security crop in Tanzania, can facilitate timely response by government and non-government organizations to food shortage or surplus conditions. Small UAVs can play an important role in linking the spaceborne remote sensing data and ground based measurement to improve the calibration and validation of satellite based estimates of in-season crop metrics. In Tanzania most of the growing season is often obscured by clouds. UAV data, if collected within a stratified statistical sampling framework, can also be used to directly in lieu of spaceborne data to infer mid-season yield estimates at regional scales.Here we present an object based approach to estimate crop metrics like crop type, area, and height using multi-temporal UAV based imagery. The methods were tested at three 1km2 plots in Kilosa, Njombe, and Same districts in Tanzania. At these sites both ground based and UAV based data were collected on a monthly time-step during the year 2015 growing season. SenseFly eBee drone with RGB and NIR-R-G camera was used to collect data. Crop type classification accuracies of above 85% were easily achieved.
