Assessing the impact of climate variability on seasonal streamflow forecasting in the Iberian Peninsula

Abstract

ABSTRACT The water shortage represents for some regions, in particular for the Iberian Peninsula (IP), one of the most severe limiting factors to maintain a sustainable development, since water resources play a crucial role in various socio-economic and environmental needs, such as agriculture, industry, hydropower industry, and tourism sector. The problem of water scarcity is likely to become more severe according to the projected decrease/increase in water availability/demand. Then, the motivation of this dissertation relies on the necessity to improve the understanding of the large-scale climate variability that drives the streamflow variability on the IP, which becomes the basis for developing streamflow forecast at scales that are of paramount importance for reservoir operations and irrigation management decisions, protection of the environment or in the reduction of expenses in flood and drought mitigation. During the first part of this work, an analysis of the streamflow database was performed. A process to identified stations with inhomogeneities in data series, mainly derived from the regulation processes, was carried out using a combined methodology based on Pettitt test and Common Area Index over original database of 1380 gauging stations. As result, a total 382 stations were selected for this study, covering the period from October 1975 to September 2008. Also, the main spatial and temporal characteristics of streamflow variability in the IP were described. The second part of this Thesis consisted on identifying the main climate factors that have a noteworthy influence on near future (lagging from one to four seasons) seasonal streamflow variability of the IP Rivers and provides an insight into the possible mechanisms and physical processes behind these relationships. Firstly, teleconnection indices, which represent most of the dominant sources of climate variability, were evaluated as potential predictors. Secondly, Singular Value Decomposition (SVD) technique was employed to identify and isolate the main modes of covariability between seasonal streamflow and the climate variables (sea surface temperature, geopotential height at 500 hPa in Northern Hemisphere and global temperature and precipitation) that precede it from one to four seasons. Once the main climatic predictors were identified, predictions based on them were conducted. A leave-one-out cross-validation approach based on a multiple linear regression approach that combining Variance Inflation Factor and Stepwise Backward selection was used to avoid multicollinearity and select the best subset of predictors. The forecasting methodology was developed for four forecasting scenarios, related to the number of seasons prior to which the forecasting is made, from one year (4S scenario) until one season (1S scenario) in advance, updating and improving the predictions seasonally. The correlation coefficient (RHO), Root Mean Square Error Skill Score (RMSESS) and the Gerrity Skill Score (GSS) were used to evaluate the forecasting skill. For the predictions made based on teleconnection indices, in case of autumn streamflow, good forecasting skill (RHO > 0.5, RMSESS > 20%, GSS > 0.4) was found for a third of the stations located in the Mediterranean Andalusian Basin, the North Atlantic Oscillation of the previous winter being the main predictor. Also, fair forecasting skill (RHO > 0.44, RMSESS > 10%, GSS > 0.2) was found in stations in the northwestern IP (16 of these located in the Douro and Tagus Basins) with two seasons in advance. For winter streamflow, fair forecasting skill was found for one season in advance in 168 stations, with the Snow Advance Index as the main predictor. Finally, forecasting was poorer for spring streamflow than for autumn and winter, since only 16 stations showed fair forecasting skill in with one season in advance, particularly in the northwestern of IP. The use of SVD improved the forecasting skills of the autumn streamflow, in particular relevant for 3S scenario. In this case, up to 42 stations present fair forecasting skills (RHO > 0.44, RMSESS > 10%, and GSS > 0.2), particularly in the Mediterranean Andalusian Basin, with Zdjf2 and Adjf2 (related to winter NAO) as predictors, but also in the north-northwestern IP (Douro, Miño-Sil, Cantabrian and upper Ebro Basins), being Rson1, Adjf1 and Rdjf1 (linked to ENSO) the main predictors. Also, a refinement in spring streamflow forecasting is observed, specially in 3S scenario, when values of RHO > 0.44, RMSESS > 10%, and GSS > 0.2 are obtained in 21 stations, mainly locate in the northeastern quadrant of IP. In those cases Pjja1 (linked to summer ENSO phenomenon), Zjja1 and Tjja1 (which could be associated with the summer Northern Annular Mode and with criosphere variability) were used as predictors. Conversely, winter streamflow was not forecasted in almost any stations. In this case, SAI appeared to be the only reliable predictor. In conclusion, this Thesis presents a valuable contribution to the studies regarding seasonal streamflow forecasting of the IP Rivers. Some distinguishing features are that it relies on a long, complete, reliable and spatially well-distributed streamflow database, which enables to describe with a high spatial resolution the potential use of different climate signals as predictors of seasonal streamflow in different areas of the IP, which become very useful for making local decision in water resources management. Also, it explores the links between climate signal and streamflow variability of the IP Rivers, not only evaluating the most commonly used climate indices but also exploring further relationships between climate variability and streamflow in the following seasons. Finally, this study can provide a more comprehensive view of relationship of climate variability and streamflow on seasonal timescales in a way that can significantly contribute to streamflow forecasting purposes (with various forecasting schemes, according to the time in advance the predictions are made), providing the option of developing water-management policies some seasons in advance and with the possibility of modifying or adjusting these strategies as the predictions are updated.