Mapping smallholder maize farm distribution using multi-temporal Sentinel-1 data integrated with Sentinel-2, DEM and CHIRPS precipitation data in Google Earth Engine

Colette de Villiers, Cilence Munghemezulu, Solomon G. Tesfamichael, Zinhle Mashaba-Munghemezulu, George J Chirima

Abstract


Mapping smallholder maize farms in complex and uneven rural terrains is a major challenge to ensure accurate spatial representation of the farming units. Remote sensing technologies rely on various satellite products for identifying maize cropland cover among other land cover types. Multi-temporal Sentinel-1 Synthetic Aperture Radar (SAR), Sentinel-2, Digital Elevation Model (DEM) and Climate Hazards Group InfraRed Precipitation with Station (CHIRPS) data were investigated for mapping maize crop distributions for the growing seasons over 2015 – 2021 in the Sekhukhune municipal area of the Limpopo Province, South Africa. Sentinel-1 SAR variables, including monthly VH, VV, VV+VH and Principal Component Analysis VH polarisation data were integrated with Sentinel-2 Normalized Difference Vegetation Index (NDVI), terrain (DEM), and precipitation (CHIRPS) data. The data were used in the Random Forest (RF) classification to distinguish maize crops from four other land cover types. The findings indicated that the models that used only Sentinel-1 as input data had overall accuracies below 71%, while a combination of Sentinel-1 and Sentinel-2 NDVI data produced the most accurate maize crop distribution map. The overall best performance models, producing overall accuracies above 83% were those where Sentinel-1 (VV+VH) data were integrated with all the ancillary data. In all, 24 models were compared. Overall, the McNemar test indicated enhanced performance for models where all the ancillary input data had been included. The results of our study showed considerable temporal variation in maize area estimates with 59240.84 ha in the 2018/2019 growing season compared to 18462.51 ha in 2020/2021. The spatial information gathered through these models proved to be valuable in that it is essential for addressing food security, one of the objectives of the Sustainable Development Goals (SDGs).


Full Text: PDF