Spatio-temporal modelling of groundwater resources to improve drought risk assessments : proposal of indicators for Siliana catchment in Tunisia

Details

Supervisors

Vanclooster, Marnik

Faculty

Faculté des bioingénieurs

Degree label

Master [120] : bioingénieur en sciences et technologies de l'environnement, à finalité spécialisée

Abstract

This master’s thesis proposes a monitoring system of hydrogeological droughts located in the catchment area of the Siliana river, a tributary of the Medjerda river located in northern Tunisia and dominated by farming. Three methods are adopted and propose indicators integrating groundwater resources in the diagnosis of drought over 1997-2013 period. The first method is a meteorological based drought assessment approach using CHIRPS database providing rainfall estimates from satellite observations and in-situ stations. These data are used to calculate the Standardized Precipitation Index (SPI) and monitor monthly rainfall anomaly. Precipitation deficit at different time scales ranging from one to twenty-four months are tested to detect a direct link between rainfall deficit and groundwater storage variation. The second method is a hydrogeological oriented approach based on in situ available piezometric data. Piezometric levels available during the low (summer) and high (winter) water seasons are normalized and provide a piezometric indicator of groundwater dynamics for each selected well. This normalized indicator is divided into water content classes. For the last method, a hydrological model is implemented using the Soil and Water Assessment Tool (SWAT) software coupled with QGIS software and allows assessing groundwater recharge based on water infiltration. This recharge indicator is divided into volume recharge classes. Spearman’s rank correlation between SPI and piezometric indicator showed highest coefficient scores for the twelve (SPI-12) and twenty four (SPI-24) month time scales, the depth of the groundwater body apparently influencing the correlation. The confusion matrices tend to confirm the correlation analysis. In contrast, the comparisons between piezometry or meteorology drought indicators and the modelled recharge indicator showed very low result, suggesting that a hydrological model hardly represents hydrogeological dynamics.