Abstract
A drought is a slowly developing natural phenomenon that can occur in
all climatic zones and can be defined as a temporary but significant
decrease in water availability. Over the past three decades, the cost of
droughts in Europe has amounted to over 100 billion euros and it is
expected to considerably increase as future droughts are projected to be
more severe and long-lasting. Although drought monitoring and management
are largely studied in the literature, they often fail at yielding
precise information on critical events occurrence and associated impacts
such as reduction of electricity production or crop failures. This is
due to the difficulty of capturing the complexity of drought dynamics,
which evolve over diverse temporal (and spatial) scales, including
short-term meteorological droughts, medium-term agricultural droughts,
and long-term hydrological droughts, as well as the non-physical aspects
related to droughts (water management, irrigation, etc.). In this work,
we contribute a Machine Learning based framework named FRIDA (FRamework
for Index-based Drought Analysis) for the identification of
impact-based, site-specific drought indexes. FRIDA is a fully automated
data-driven approach that relies on advanced feature extraction
algorithms to identify relevant drought drivers from a pool of candidate
hydro-meteorological variables. Selected predictors are combined into an
index representing a surrogate of the drought conditions in the
considered area, including either observed or simulated water deficits
or remotely sensed information about the state of the crops. FRIDA is
portable across different contexts for supporting the formulation of
basin-specific indexes to better inform drought management strategies.
Several real-world examples will be used to provide a synthesis of
recent applications of FRIDA in case studies featuring diverse
hydroclimatic conditions and variable levels of data availability.