Can we foresee drought and its impacts?

Contributed by Samuel J. Sutanto and Henny A.J. Van Lanen; Hydrology and Quantitative Water Management Group, Wageningen University, the Netherlands.

The 2018 and 2019 droughts are still fresh in our memory, when extended areas in Europe suffered, once more, from extreme drought. Climate projection studies show that drought very likely will become more frequent. Similar to the 2015 drought (Van Lanen et al., 2016), impacts of the 2018 drought in Europe were tremendous and affected many sectors. For example, farmers in multiple European regions experienced the worst harvest in generations with yields of some crops down up to 50%. Livestock farming was also hampered because additional fodder had to be purchased for the cattle. In Germany, the price of gasoline went up due to the increase in transportation costs because ships from Rotterdam port could not carry the normal cargo. Last but not least, electricity prices in Norway also went up because of the country’s high dependence on hydropower, which was affected by low streamflow. As scientists working on drought, we got simple but straight to the point questions from news and media, several stakeholders, water managers and colleagues, such as, can you predict drought well in advance, can you forecast the impacts of drought?

Can we predict the drought hazard in advance?

With the development of the pre-operational ANYWHERE Drought Early Warning System (ADEWS) under the EU-funded H2020 ANYWHERE project (http://anywhere-h2020.eu/), we can confidently say “yes” that the drought hazard can be predicted a few months in advance. Please note that the ADEWS forecasts drought hazard indices up to 7 months ahead rather than just weather (e.g. precipitation) or hydrology (e.g. soil water content, river flow). The drought that started in the summer of 2018 in some parts of the west and central Europe was foreseen in the ADEWS since spring 2018 (Sutanto et al., 2019a). The seasonal forecast issued at the beginning of May 2018 with lead-time of 3 months clearly showed the evolution of drought from month to months in these regions. However, we must admit that we later found that the drought forecasts underestimated drought severity compared to observed, especially in northeast European regions. Nevertheless, the forecasts clearly showed the upcoming drought. The drought hazard was better predicted with seasonal forecasts issued in June 2018. This was expected since the shorter the lead-time, the better the forecast skill. Our results support the few previous studies on drought forecasting that the drought hazard can be predicted more or less up to 2-3 months in advance.

Can we forecast the impacts of drought?

There was no positive answer to this question until a few months ago. All drought early warning systems across the world only provide drought hazard forecasts. Drought impact may not necessarily occur in a region that is predicted to suffer from drought hazard, if there is no exposure and vulnerability (see Figure below). Drought hazard, exposure, and vulnerability are three major components that commonly are used to indicate drought risk and impact (conventional bottom-up approach). However, it is still questionable if an impact actually will occur even there is a high risk in a certain region. Frankly speaking, we say that drought impacts most likely will occur, but not necessarily do.

To investigate whether drought impacts,such as impacts on water supply, crops and energy demand among others, can be predicted a few months ahead (not risk), or not, we have developed drought impact forecasting functions that relate specific drought impacts in a certain European region to drought hazard indices. We used machine learning, Random Forest (RF), the ADEWS and an impact database taken from the European Drought Impact Inventory (EDII, Stahl et al., 2016) to build the functions (top-down approach, Sutanto et al., 2019b). This study shows that drought impacts can be predicted a few months ahead for certain impacted sectors and pan-European regions. Thus, based on the study, if someone now asks us again: can we forecast the impacts of drought? We will gladly say: “yes” up to three months ahead and in some cases even longer. However, keep in mind that the skill of the functions strongly depends on the quality and quantity of the impact database, the forecasted hazards, and the method i.e. machine learning. Clearly, it appeared that we could not develop a skillful drought impact forecasting model in a region that has limited number of reported impacts.

What are the next steps?

Well, drought monitoring and early warning systems are already operating in some countries/regions, such as the US, Switzerland, Germany, Europe, and Africa. Using these systems, drought hazard can be forecasted some months ahead. Still missing in these systems are drought impact functions to forecast impacts. Therefore, we encourage the operators of these systems to initiate development of drought impact functions, starting with developing of a comprehensive drought impact database that includes all relevant impacted sectors and has adequate spatial and temporal resolutions.

References

Birkmann, J., Cardona, O.D., Carreno, M.L., Barbat, A.H., Pelling, M., Schneiderbauer, S., Kienberger, S., M.Keiler, Alexander, D., Zeil, P., and Welle, T.: Framing vulnerability, risk and societal responses: the MOVE framework, Nat Hazards, 67, 193–211, doi: 10.1007/s11069-013-0558-5, 2013.

Stahl, K., Kohn, I., Blauhut, V., Urquijo, J., De Stefano, L., Acacio, V., Dias, S., Stagge, J.H., Tallaksen, L.M., Kampragou, E., Van Loon, A.F., Barker, L.J., Melsen, L.A., Bifulco, C., Musolino, D., de Carli, A., Massarutto, A., Assimacopoulos, D., and Van Lanen, H.A.J.: Impacts of European drought events: insights from an international database of text-based reports, Nat. Hazards Earth Syst. Sci., 16: 801–819, doi:10.5194/nhess-16-801-2016, 2016.

Sutanto, S.J., Van Lanen, H.A.J., Wetterhall, F., and Llort, X.: Potential of pan-European seasonal hydro-meteorological drought forecasts obtained from a Multi-Hazard Early Warning System, BAMS, doi:10.1175/BAMS-D-18-0196.1, 2019.

Sutanto, S.J., van der Weert, M., Wanders, N., Blauhut, V., and Van Lanen, H.A.J.: Moving from drought hazard to impact forecasts, Nature Communications, doi:10.1038/s41467-019-12840-z, 2019.

Van Lanen, H.A.J., Laaha, G., Kingston, D.G., Gauster, T., Ionita, M., Vidal, J-P, Vlnas, R., Tallaksen, L.M., Stahl, K., Hannaford, J., Delus, C., Fendekova, M., Mediero, L., Prudhomme, C., Rets, E., Romanowicz, R.J., Gailliez, S., Wong, W.K., Adler, M-J., Blauhut, V., Caillouet, L., Chelcea, S., Frolova, N., Gudmundsson, L., Hanel, M., Haslinger, K., Kireeva, M., Osuch, M., Sauquet, E., Stagge, J.H. and Van Loon, A.F.: Hydrology needed to manage droughts: the 2015 European case, Hydrological Processes, 30: 3097–3104, doi: 10.1002/hyp.10838, 2016.