Modelling future weather and water extremes has become critical research endeavors as governments, city planners and even companies try to grapple with the risks facing human communities, infrastructure and assets. Today’s climate models are increasingly sophisticated, with some of them containing enough code to fill 18,000 pages of printed text and requiring tennis court-sized supercomputers to run. But the scientific community is itself still grappling with the complex interactions of factors in the climate system and water cycles.
Professor David Hannah, UNESCO Chair in Water Science at the University of Birmingham, is using observations and model projections to understand the interactions between climate and the hydrological cycle - specifically to project how floods and droughts might change towards the end of the century. Professor Hannah has shown, as part of international collaborations that explore future low (drought) and high (flood) river flows, that “the greatest source of uncertainty is not between what climate model you use, but which hydrological model you run those climate models through, which tells us that we really need to better understand terrestrial hydrological processes to make projections of future extreme environments.”
Notably, the largest uncertainties in hydrological projection are found typically in areas with the most significant water security challenges. One issue Professor Hannah identifies is the impact of changing land use and land cover on the terrestrial water cycle, specifically the changing plant-water interactions as we convert more land to agriculture for food supply. Lower levels of biodiversity and less diverse land cover limit the routing options of terrestrial water, compared to naturally vegetated landscapes, Hannah’s research has found. As a result, monocultures are more vulnerable to climate disturbances and this could weaken the resilience of the planet’s terrestrial water cycle to stressors.
Monocultures are less able to cope with climate extremes, such as droughts, because of their limited ability to respond to these stressors. “As you move from mixed land use like forests and pastures to cropping one thing, this results in the homogenization of the water cycle. The system responds much more similarly. More diverse environments are more resilient and less vulnerable to changing conditions,” Professor Hannah asserts.
For example, conversion of mixed rainforest to agriculture is reducing local evapotranspiration and as a result, less rainfall is predicted to occur by the middle of the century, which then increases the risk of drought, in turn increasing the risk of tree mortality, further reducing evapotranspiration and the hydro-ecological processes that retain water. The vulnerability to desertification through such a feedback loop is increasing due to forest clearing. Professor Hannah was part of a research team that identified South America, along with Western Europe, as future regional drought hotspots with a 20% increase in occurrence by the end of the 21st century.
While desertification and drought are long-term processes, climate models also need to predict the chance of rare extreme events like severe cyclones and windstorms. While their occurrence is set to increase due to climate change, predicting how often or likely they are is extremely difficult because historical data is, by definition, scant. The rarity of these events means there are no persistent long-term meteorological observations which makes it hard to create a robust risk assessment based solely on historical records.
Professor Gregor Leckebusch, professor of meteorology and climatology at the University of Birmingham, is working on ‘multi-model ensemble forecasting’ as one way to better understand severe tropical and extra-tropical cyclone risk. “We only have reliable data for extreme storms in the Atlantic and the Pacific for 50 years maximum with perhaps two or three very severe extra-tropical cyclones and typhoons per year. This is not a large enough sample to produce a robust statistical estimate,” he says.
In ensemble forecasting, predictions are made by combining the outputs of models run with different sets of initial conditions instead of just one, as is the case in deterministic forecasting. By combining outputs, researchers get an ensemble forecast that predicts a range of weather conditions. “If we would like to understand why and how in the future changes may occur, in the context of anthropogenic environmental change, or of natural climate variability, we need to know which factors are steering the appearance and the intensity of these extreme events.”
In Professor Leckebusch’s multi-model approach, the outcomes of each ensemble model are treated as individual “samples of reality”, in which what actually happened historically only represents one sample of what could have happened. “We can take what happened in the real-world in the past as only as one out of a potential 20,000 options. Then we could use the 20,000 options to identify what would be the frequency of occurrence of these events.”
There are far-reaching implications to such modelling, he says, from insurance and financial markets through to urban planning and disaster mitigation.
Top banner image credit: Alamy.
Discover more stories about our work and insights from our leading researchers.