Quantifying the co-occurrence of hydrological, meteorological, and agricultural droughts on a global scale

Droughts affect more people around the world than any other natural hazard and are projected to intensify due to climate change. Droughts have traditionally been divided into four broad categories: socio-economic, meteorological, agricultural, and hydrological. While the common cause for drought is abnormally low precipitation, different drought types may be caused by other factors, such as increased evaporation or anthropogenic influence.

The different drought types are often studied separately, but they are not independent of each other and they often co-occur. We quantify the co-occurrence of meteorological, agricultural, and hydrological droughts at global scale, to our knowledge for the first time. We use the 3-month Standardised Precipitation and Evaporation Index (SPEI), 1-month Soil Moisture Anomaly (SMA) and 3-month Standardised Streamflow Index (SSI) as proxies for meteorological, agricultural, and hydrological droughts, respectively. We compute the drought indices for globally at sub-basin scale for years 1981-2010, using the newly published HydroATLAS level 8 catchments (average size of approximately 750 km²) as spatial units. Each unique meteorological, agricultural, and hydrological drought event is characterised by the commonly used duration, intensity and severity metrics. The co-occurrence of different severe drought events is defined by using the spatial and temporal intersection of the identified events. We analyse the co-occurrence of severe drought events using Association Rules data mining method in order to quantify the relationship between the drought types, and their co-occurrence.

Our results indicate that the global average probability of co-occurrence of all three drought types in a single drought event is 30%. The probability of the occurrence of a single drought type is 61% (SMA), 64% (SPEI) and 69% (SSI) of all unique drought events. However, these figures vary considerably between continents. Interestingly, we find that SMA and SPEI are poor predictors to SSI, which might be attributed to the different nature of the processes, as streamflow is affected by upstream conditions. Precipitation and soil moisture are more local processes with weaker links to anthropogenic influence, irrigated areas being an obvious exception. We also detect an increasing global trend in severe drought events and the co-occurrence of drought types. Our results, however, are likely sensitive to the chosen indices, thresholds and the definition of co-occurrence, thus further studies are needed.

We argue that quantifying the co-occurrence of different drought types provides important information for early warning systems and drought management planning. It may also be useful for the development of comprehensive composite drought indicators. Understanding the linkages between drought types may support longer-term, proactive drought management planning that is better tailored to regional climates.