Mapping crop types in Levubu area of Limpopo using Landsat 8 and Sentinel-2 sensing data
Date
2021
Authors
Tshigoli, Thivhafuni Portia
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Of late food security has become one of the serious global challenges; hence crop production
and monitoring plays a crucial role. One of the solutions to food security challenges is through
developing proper management systems of agricultural resources. Remote sensing offers
unique opportunities for frequent monitoring and management of crop types. This is done
through mapping the distribution of crops which provides farmers, policy makers, decision
makers and researchers an opportunity to develop better strategies for crop management and
production. The study used Landsat 8 OLI and Sentinel-2 MSI, which were acquired for a
single date, in mapping the spatial distribution of different crop types in Levubu area in the
Limpopo province, South Africa. Two main objectives were followed in this study: firstly, to
assess the potential of Landsat 8 and Sentinel-2 data in understanding the distribution pattern
of crop, secondly, to determine the potential of the random forest algorithm in classifying
different crops using different remote sensing datasets. Eight classes (Avocado, Banana, Built
up, Macadamia nut, Maize, Mango, Vegetation and Water body) were classified in this study
using random forest (RF) machine learning classifier. Other classes such as Built up,
Vegetation and Water body were classified to assess the ability of RF to distinguish between
crop classes and non-crop classes. The results show overall accuracies of 77.08% and 87.04 %
with kappa coefficients of 70.72% and 83.56% for Landsat 8 and Sentinel-2 using RF,
respectively. The study found that the overall accuracy, user’s and producer’s accuracy results
of Sentinel-2 MSI were better than Landsat 8 OLI. Therefore, Sentinel-2 MSI is suitable for
more accuracy and increased details in mapping crops in Levubu area.
Description
A research report submitted in fulfilment of the requirements for the degree of Master of Science in GIS and Remote Sensing to the Faculty of Science, School of Geography, Archaeology and Environmental Studies, University of the Witwatersrand, Johannesburg, 2021