Mapping the maize growth period using multi-temporal sentinel 1 and 2 imagery: a case study in Kasisi area of Chongwe district.

Effective agricultural monitoring is essential for ensuring food security and efficient resource management. This study aimed to use Synthetic Aperture Radar (SAR) from Sentinel 1 and Optical imagery from Sentinel 2 multi spectral instrument (MSI) for mapping and monitoring Maize fields in the Kasisi area of Chongwe District, Zambia, from November, 2019 to April 2020. This was done by capturing the temporal variations in maize growth and mapping its coverage from November 2019 to April 2020. The analysis focused on tracking maize phenological stages— sowing, emergence, vegetative growth, and maturity—through biweekly observations of SAR backscatter and NDVI. Dual-polarized SAR data (VV and VH) were analyzed to detect structural and moisture changes in maize, while NDVI and NDWI indices from Sentinel-2 provided complementary vegetation and water condition metrics. These indices also enhanced a Random Forest classifier used for land cover classification. Field-validated training data supported the classification, which achieved an overall accuracy of 96.97% and a Kappa coefficient of 0.95. Sowing was identified between 1st–15th November 2019, with emergence occurring by mid December. Maturity was reached by mid-January 2020, followed by a post-maturity decline in backscatter from March to April, marking the harvesting phase. The results demonstrate the effectiveness of SAR and optical data fusion for identifying maize growth stages and mapping crop extent, particularly in cloud-prone tropical regions. This approach offers a scalable, weather independent solution for precision agriculture and vital input for yield forecasting in Sub-Saharan Africa.

Keywords: SAR, Crop Monitoring, Random Forest, Sentinel 1, Sentinel 2, NDVI, NDWI