University of Twente Student Theses


Crop discrimination using time series Sentinel-1 SAR data

Mutasha, Brian Mulenga (2022) Crop discrimination using time series Sentinel-1 SAR data.

[img] PDF
Abstract:Despite efforts to end world hunger and undernourishment, food security is a serious concern in many countries. Accurate and timely information about crop types is important for proper food production management and monitoring This study examined whether phenological information obtained from Sentinel-1 time series and Support Vector Machine algorithm would allow to discriminate rice and maize. The research utilized secondary dataset made available by the International Rice Research Institute. The Sentinel-1 time series data were used to extract the temporal mean backscatter for each field at different growth stages. Statistical tests to determine whether there were significant differences between rice and maize growth stages were done using the Mann-Whitney U Test. The results show that differences exist during the crop development phases that could be utilized to discriminate rice from maize. A significant difference was observed at flowering and harvest stages in VV polarization and VV/VH ratio. The backscatter difference was also significant in the VV/VH ratio polarization. It was observed that crop establishment phase had highest overall accuracy (O.A = 80.6%, Kappa = 0.58), while harvest stage had lowest overall accuracy (O.A. = 67.7%, Kappa = 0.28). The classification results show comparable crop type distributions with field observation.
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Subject:43 environmental science, 48 agricultural science
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page