Developing an integrated GIS-remote sensing methodology for estimating groundwater trends in the Upper Molopo River Catchment, South Africa
Abstract
Groundwater is an essential part of the hydrologic functions and an important element for socio-economic development. Groundwater plays significant role in arid and semi-arid regions worldwide by providing for both human consumption and the preservation of the ecology. The semi-arid upper Molopo River catchment (UMRC) is no exception. Observations from this catchment are that there is a water shortage which is attributed to groundwater level decline caused by poor water groundwater monitoring done only in one part of the study area at the expense of the other parts. Hence, an alternative groundwater monitoring system should be implemented in cases where such a system is not in place.
The estimation of trends in groundwater resource is important for sustainable utilization and management. Groundwater potential zones in the study area were delineated using Geographical Information Systems (GIS) and Remote sensing (RS) techniques. Various thematic layers, such as soil type, geology, elevation, slope, lineament, land use type and drainage density, were created; and integrated in ArcMap 10.2 on a scale of 1 to 5 according to their relative significance to groundwater potential. The integrated map generated five categories, which are, very low, low, moderate, high and very high. The groundwater potential sites were related to groundwater level data obtained from the Department of Water and Sanitation.
A direct relationship existing between groundwater and phreatophytic vegetation condition as surrogate for groundwater regime in the absence of rainfall has been noted. Phreatophyte vegetation densities in areas, such as the upper Molopo River catchment, are controlled by the existence of groundwater during dry season and exhibit the behaviour of groundwater level. Remote Sensing techniques were utilized to map phreatophyte vegetation density from 1995 to 2015 using supervised maximum likelihood algorithm. The multi-temporal Soil-Adjusted Vegetation Index (SAVI) images were created and used as a base from which phreatophytic vegetation density classes were extracted. Vegetation cover change detection was performed to determine the spatial-temporal trends in phreatophyte vegetation condition.
It was noted that evapotranspiration (ET) demand, in the dry seasons of arid and semi-arid environments, is met by groundwater discharged by phreatophytes, and through seepage into the shallow groundwater sites. This relationship is substantial for groundwater modelling. As a result, a surface energy balance algorithm for land (SEBAL) was applied in the upper Molopo River catchment to estimate potential ET, which provided indication of groundwater
depths condition. The input variables for the SEBAL model were derived from Landsat data. The variables included land surface temperature, surface emissivity, surface albedo, Normalized Difference Vegetation Index, vegetation proportion, leaf area index, and elevation map. Other input variables were meteorological data - specific humidity, wind speed, air temperature and air pressure. A simple regression analysis was then used to establish the relationship between groundwater level and potential ET for the year 2015 and groundwater level.
Phreatophytic vegetation water potential was also investigated in order to model groundwater availability. The assumption was that trends in groundwater depth conditions could be explained by phreatophytic vegetation water potential. An integrated remote sensing-field survey methodology was utilized to map the spatio-temporal trends in phreatophytic vegetation water potential between 1995 and 2015. The Normalized Difference Water Index (NDWI) was derived from each specified satellite image to retrieve spatio-temporal information about vegetation water potential. A decline in NDWI values was then noted from 1995 to 2015. Relative water content (RWC) measurements were also collected from three surveyed phreatophytic vegetation species, namely Ziziphus Mucronata, Euclea Undulata and Rhus Lancea, in order to determine leaf water potential. A simple regression analysis was also used to establish the relationship between vegetation water potential and groundwater depth.
The multinomial logistic regression analysis revealed a significant relationship between groundwater potential sites and groundwater level. The simple regression analysis revealed a linear trend between groundwater level and the 2015 phreatophytic vegetation density. Shallow groundwater was observed in areas of high phreatophyte vegetation density, in contrast to low phreatophyte vegetation density where deep groundwater levels were observed. The linear regression analysis revealed a significant relationship between groundwater level and ET intensity. It was concluded that trends in potential ET, during the study area’s dry seasons, exhibited groundwater level. The linear regression analysis revealed a significant relationship between leaf water potential and NDWI values, and between groundwater level and phreatophytic vegetation density. Hence, the phreatophytic vegetation water potential can be utilized as a diagnostic tool for groundwater behaviour and for the establishment of informed decisions.
Implications emerging from the obtained results suggest that GIS and RS can efficiently and reliably provide valuable information regarding potential sites for groundwater in the UMRC and for groundwater monitoring and management purposes. These results also indicate the importance of remote sensing techniques in unpacking trends in phreatophyte vegetation condition, potential ET, and phreatophytic vegetation water potential as surrogate for groundwater levels for informed decisions on groundwater monitoring and management.