Uncertainty analysis of hydrological model results and assessment regarding water risks in the Mediterranean Basin – Case study: Chiba, Tunisia

Kurzfassung

Mediterranean countries are very vulnerable to possible impacts of climate induced changes on hydrological budgets and extreme events. Mediterranean regions are already affected by a variety of natural and man-made threats regarding water security like severe droughts, extreme flooding, salinization of coastal aquifers, degradation of fertile soils and increasing desertification due to unsustainable management practices. With climate change amplifying these problems in the future there would arise increased potential for tensions and conflicts in this area. Despite the scientific knowledge that these changes may occur it is difficult to quantify these changes due to a variety of inherent sources of uncertainty. That was enough reason for the European Commission (EC) to initialize a series of projects within the Seventh Framework Programme Collaborative Research Project (FP7). Therein the project CLIMB (Climate Induced Changes on the Hydrology of Mediterranean Basins - Reducing Uncertainty and Quantifying Risk through an Integrated Monitoring and Modeling System) was started which should investigate present and future climate induced changes in the hydrological budgets and extremes for Mediterranean and neighboring countries. This master thesis applies techniques to quantify uncertainties and associated risks in Chiba basin for a reference (1971 – 2000) and scenario period (2041 – 2070) respectively. The Chiba catchment is a test site of the CLIMB project and situated on Cap-Bon peninsula in the north east of Tunisia. Two studies form a series of 108 hydrological model simulations that are being performed with the hydrological model WaSiM-ETH. One is a parameter uncertainty study (PUS) that investigates the effects of applying a varying set of important model parameters on a constant climate data set. The other is a climate signal uncertainty study (CUS) to quantify the impact of using different climate data sets on one basic model parameterization. Evaluation of CUS and PUS enables to quantify uncertainties of key outputs of interest as well as to investigate how these values may change in the future. This is complemented by a sensitivity analysis (SA) to determine most important input factors for ’relative soil water content in the root zone’ and ’ground water recharge’. SA also enables an investigation whether uncertainties of those values can be further reduced. Besides that a feasibility study examines whether tomatoes can be cultivated in the watershed under future climate conditions (rainfed --> without irrigation). At last selected climate change and hydrological indices analyze developments regarding droughts and extreme events as well as water scarcity in the study area. Nowadays water scarcity already forces anthropogenic actions in Chiba basin like irrigation of agricultural land and overexploitation of Korba ground water aquifer. This facilitates a drop of ground water levels near the coast and thus sea water intrusion with deteriorating effects on water quality. Therefore quantifying climate induced changes on hydrological water budgets and extremes (e.g. droughts) is of great interest in this region. Findings of this study estimate annual mean values of surface air temperature in Chiba basin to rise by 1.8C +/- 0.1C between the reference and scenario period. Annual sums of rainfall indicate a decrease of 59.3 mm +/- 17.2 mm (-16.9 % +/- 4.9 %). This has further negative effects on relative soil water content in the root zone [-0.013 +/- 2.5E-03 (-14 % +/- 2.7 %)] as well as actual evapotranspiration [-40.4 mm +/- 7.9 mm (-12.5 % +/- 2.4 %)]. Climate induced changes in the water balance of Chiba basin amount to 3.33 million m3 (-66.1 %) of less available water (best estimate) on average and for every year. Precipitation constitutes the key output to worry within this study. High variability (high values of COV [coefficient of variation]) and a wide band of projected rainfall in the wet season (September to March) result in a high degree of uncertainty for this variable. Rainfall also influences characteristics of relative soil water content in the root zone and thus actual evapotranspiration rates. These variables feature elevated COV and more distinct margins of simulated values in times of varying rainfall availability. Higher confidence exists that rainfall will be scarce during summer (JJA) which also leads to a small band of very low values for relative soil water content in the root zone and actual evapotranspiration in all model runs. Temperature and potential evapotranspiration exhibit only small values of COV and all model simulations project a rise with statistical significance. Therefore a higher degree of confidence exists in this case. Ground water recharge and depth to ground water table do not seem to be severely affected by climate induced changes throughout the modeling. Nevertheless high values of COV lead to appreciable uncertainty regarding ground water recharge rates in individual years. Climate change indices show an increase of summer days (125.8 to 163.2 --> +30.1 %) as well as tropical nights (18.5 to 52.5 --> +184 %) with statistical significance. Annual sums of wet day precipitation diminish with statistical significance (325.9 mm to 268.9 mm --> -17.5 %). Days with precipitation above 10 mm (6.07 to 6 --> -1.1 %), 20 mm (1.63 to 1.37 --> -16.3 %), 25 mm (1.07 to 0.67 --> -37.5 %) and 38 mm (0.33 to 0.17 ! -50 %) exhibit a decrease (statistical significance for 25 mm). 65 mm and 75 mm rainfall events do not occur in Chiba basin. 38 mm, 65 mm and 75 mm were investigated for they are supposed to cause different types of erosion in semi-arid Tunisia. Consecutive dry days increase (79.9 to 94.3 --> +18 %) but without statistical significance. Mean duration of ’WDSI’ (5.63 to 3.47 --> -38.5 %) as well as values of ’TX90p’ (8.73 to 8.7 --> -0.4 %) dwindle in the future. All climate change indices feature trends of rising temperature and less rainfall. Together with more consecutive dry days this facilitates the occurence of more pronounced droughts in the future. Nevertheless climate data does neither indicate longer warm spell duration (’WSDI’) nor a higher percentage of very hot days (’TX90p’). Besides that critical single precipitation events (which could amplify soil erosion phenomena in Chiba basin) do not feature gains in intensity. Several hydrological indices and additional sources indicate that Chiba basin is highly affected by water scarcity phenomena and overexploitation of natural resources. Besides that water footprints of wheat, olives and tomatoes account for additional water demands of approximately 4.9 billion m3 (about 90 % of that additonal demand results from olives, tomatoes even need less water than on global average) for the whole of Tunisia. These crops are also predominantly cultivated in Chiba basin indicating that considerable amounts of water could be saved in the study area if these crops were cultivated under global average conditions. Climate (’Prec_Ref’, ’Prec_Fut’) as well as soil related parameters (’Theta_res’, ’N’, ’Alpha’) constitute most sensitive components for relative soil water content in the root zone in 1971 – 2000 as well as 2041 – 2070. Selected representatives of land use (’Crop_LAI’) and ground water (’Gwst’) related input factors appreciably affect outcomes of this key output as well. Selected representatives of land use (’Crop_LAI’, ’Forest_Z0’, ’Forest_VCF’) and soil (’N’) related input factors constitute most sensitive components for ground water recharge rate in the reference as well as the scenario period. Climate (’Temp_Ref’, ’Prec_Ref’) only has small effects. Ground water related parameters are not at all affecting ground water recharge. SA identified a series of important input factors related to parameters of Korba ground water aquifer, soils and plants. Further data collection and research in this field may build a better data base to reduce uncertainties related with input factor parameterization. Climate data does not profit from such approaches but establishing a monitoring system of climate stations in Chiba basin results in several benefits. An enhanced data availability of observed climate data enables to gain independence from climate models when past and present climate characteristics are investigated. Besides that it facilitates choosing general circulation models (GCMs) that are able to better reproduce observed climate conditions and using those to project future developments. This master thesis contributes to the CLIMB project by quantifying uncertainties and associated risks. Furthermore hints are given on how uncertainties can be further reduced. Quantification and reduction of uncertainty also facilitates a more efficient use of financial resources. Besides that determining the robustness\credibility or uncertainty of individual findings is of great value for political decision makers as well as stakeholders. In addition outcomes support CLIMB to create a new scientific knowledge that enables the development of more appropriate adaptation strategies in the future (e.g. shifts in current agricultural practices to crops with a more sustainable water footprint, more effective irrigation systems or the establishment of a monitoring framework of climate stations to enhance data availability in this region). All these efforts are important to better cope with the effects of projected climate induced changes on the hydrology of Chiba basin.