On the propagation of drought : how climate and catchment characteristics influence hydrological drought development and recovery

Drought is a severe natural disaster resulting in high economic loss and huge ecological and societal impacts. In this thesis drought is defined as a period of below-normal water availability in precipitation (meteorological drought), soil moisture (soil moisture drought), or groundwater and discharge (hydrological drought), caused by natural variability in climate. Drought propagation is the change of the drought signal as it moves from anomalous meteorological conditions to a hydrological drought through the terrestrial part of the hydrological cycle. The objective of this PhD research is to investigate the processes underlying drought propagation and their relation with climate and catchment characteristics, both on the catchment scale and on the global scale. The catchment-scale studies are based on five headwater catchments in Europe with contrasting climate and catchment characteristics. In one of these case study areas, anthropogenic influence on the water system was significant, resulting in severe water scarcity. As I only study natural processes in this thesis, there was a need to separate drought (as defined in this thesis) from human-induced water scarcity in this case study area. I proposed an observation-modelling framework that consists of a hydrological model to simulate the ‘naturalised’ situation and an anomaly analysis method to quantify drought and water scarcity events. Both the time series and the anomaly characteristics of the ‘disturbed’ and ‘naturalised’ situation were compared to quantify human and natural influences on the hydrological system. After simulation of hydrometeorological variables of all case study areas with a conceptual hydrological model and drought identification with the variable threshold level method, time series and characteristics of drought events were analysed. I classified the drought events into six hydrological drought types that are the result of the interplay of temperature, precipitation, evapotranspiration and storage in different seasons. The most common hydrological drought type develops as a result of a rainfall deficit. However, in the development of the most severe hydrological drought events temperature and storage-related processes play an important role, for example through a lack of recovery of the drought. As I aimed to investigate drought propagation also on larger scales, I tested an ensemble mean of a number of large-scale models (both land-surface models and global hydrological models) on their ability to reproduce the drought propagation processes found in the case study areas. The large-scale models did simulate general aspects of drought propagation (e.g. fewer and longer drought events in discharge than in precipitation), but the above-mentioned effects of temperature and storage-related processes were only partly reproduced. In the large-scale model ensemble, daily runoff reacted almost immediately to changes in precipitation, resulting in important deficiencies in drought simulation in cold and semi-arid climates and regions with large storage. For the time being, this limits the use of large-scale models for the study of processes underlying drought propagation on a global scale. Consequently, I used a synthetic conceptual hydrological model to study drought propagation on the global scale. I focused on climate control by isolating forcing effects from effects of catchment properties. The drought characteristics (duration and deficit combined) of both soil moisture and subsurface discharge exhibited strongly non-linear patterns in seasonal climates. The non-linear effects in soil moisture drought were caused by the fact that the development of soil moisture droughts in warm seasonal climates is limited by the wilting point. Hydrological droughts in both warm and cold seasonal climates showed a strong increase of deficit with duration due to a lack of recovery in the dry season or snow season, respectively. This effect was strongest in cold seasonal climates, which indicates that for the development and recovery of within-year hydrological drought temperature is an important factor. The overall conclusion of this research is that, although drought is a complex, nonlinear phenomenon with drought characteristics varying with climate type and catchment characteristics, generic patterns can be derived that reflect the different hydrological processes underlying drought propagation. These processes result in different hydrological drought types that are shown to play a role both on the catchment scale and on the global scale. The non-linear effects of snow and storage-related processes on drought are not incorporated sufficiently in the currently-used large-scale models and drought indices. Possible future steps include more focus on catchment control, in particular the representation of storage, and the role of temperature and evapotranspiration. Additionally, the findings of this research can be applied to hydrological drought forecasting, prediction in ungauged basins, and prediction under global change.