Assessing whether soil moisture content (SMC) can be estimated for wetlands in the grassland biome of South Africa using freely available space-borne sensors
Date
2019
Authors
Gangat, Ridhwannah
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Soil moisture content (SMC) takes on an important role in the hydrological functioning of wetlands. Temperature increases associated with climate change is expected to impact the hydrological regime of wetlands. Therefore, regional monitoring of SMC is essential for improved understanding of potential changes to the hydrological regime of wetlands while supporting decision making and interventions. Conventional methods of measuring SMC are costly and have a limited view of processes occurring at regional to global scales. In contrast, remote sensing can potentially offer a regular, regional overview of the hydrological function of wetlands and is therefore more cost-affordable compared to conventional methods. In the past, estimations of SMC with remote sensing lacked a sufficient spatial resolution for palustrine inland wetland ecosystem types, particularly in semi-arid countries. However, the use of recently launched and freely available high spatial resolution sensors, such as the Sentinel series, may overcome these limitations. In this study, the use of European Space Agency’s Sentinel-1A and 1B (S1A, S1B; Synthetic Aperture Radar) and Sentinel-2A and 2B (S2A, S2B; optical) sensors were evaluated for their ability to predict SMC for wetlands and drylands in the grassland biome of South Africa. The percentage Volumetric Water Content (%VWC) for 200 points was measured in the Colbyn Nature Valley which is dominated by a palustrine wetland. The %VWC in the wetlands and terrestrial area of the study area were measured using a hand-held SMT-100 soil moisture and temperature meter at a 5 cm soil depth during March and May 2018 (the peak of the hydroperiod) and regressed against the Synthetic Aperture Radar (SAR) and optical data using a parametric and non-parametric models. The results showed that Sentinel images can predict the percentage SMC, with both the S1B and S2B images achieving the highest coefficient of determinations (R² > 0.8; R² > 0.9) and relatively low Root Mean Square Errors (RMSE = 10 %; 12 %) respectively. Predicted maps showed significantly lower ranges of SMC below 50 % (p ≤ 0.05) in the terrestrial area compared to the higher ranges of SMC (≥ 50 %) in wetlands for both sensors. Although the SAR C-band is limited to the upper 5 cm of the soil depth, it shows potential to measure ranges of SMC for palustrine wetlands and terrestrial areas in the grassland biome of South Africa which will be beneficial for wetland inventorying.
Description
For a full Master’s thesis:
A thesis submitted in partial fulfillment of the requirements for the degree of Master Science in the Faculty of Science, University of Witwatersrand
may 2019