Predictability of weather and climate

Abstract:

Posit numbers, a recently proposed alternative to floating-point numbers, claim to have smaller arithmetic rounding errors in many applications. By studying weather and climate models of low and medium complexity (the Lorenz system and a shallow water model) we present benefits of posits compared to floats at 16 bit. As a standardised posit processor does not exist yet, we emulate posit arithmetic on a conventional CPU. Using a shallow water model, forecasts based on 16-bit posits with 1 or 2 exponent bits are clearly more accurate than half precision floats. We therefore propose 16 bit with 2 exponent bits as a standard posit format, as its wide dynamic range of 32 orders of magnitude provides a great potential for many weather and climate models. Although the focus is on geophysical fluid simulations, the results are also meaningful and promising for reduced precision posit arithmetic in the wider field of computational fluid dynamics.

Abstract:

Abstract Storm Eunice was a severe windstorm that impacted Central Europe in February 2022. The meteorology and synoptic dynamics of Eunice have been studied in depth in several studies examining features of the storm such as its sting jet. The contribution of climate change to the storm dynamics and severity was examined in previous work, which found that in counterfactual weather forecasts - given an identical initial synoptic setup - climate change had measurably increased the severity of the storm. Here we move beyond meteorological attribution and quantify the role of climate change in the insured losses incurred during Eunice in, to the best of our knowledge, the first impact attribution of its kind for a European windstorm event. We combine the same counterfactual weather forecasts with three loss models, including two state-of-the-art commercial models, finding that the increases in meteorological severity do translate through to significant increases in estimated loss. We estimate a conditional increase in insured loss of nearly €2 bn between pre-industrial and present-day climates. Of particular note is the existence of several members within the forecast ensembles whose losses are far greater than what unfolded in reality. This includes one realisation, simulated in a warmer “future” climate, in which the estimated loss could reach over 10x the realised loss during Eunice. The plausible existence of such a catastrophic loss is of considerable relevance to a wide variety of stakeholders across adaptation planning and the financial sector. We suggest that our results practically demonstrate not only the utility of counterfactual weather forecasts in quantifying impacts attributable to climate change, but also the value of academic - private partnerships in which the two sectors are able to bring different areas of expertise.

Abstract:

Abstract:

More frequent heatwaves in Europe are posing considerable risks to human health, infrastructure, and ecosystems. However, the contributions of external forcing factors such as well-mixed greenhouse gases (GHGs) and aerosols remain to be better quantified. Here, using model outputs from the Large Ensemble Single Forcing Model Intercomparison Project (LESFMIP), a recent atmospheric reanalysis and a machine learning method—self-organising maps (SOMs), we attribute European heatwave trends during 1940–2020 to various external forcings. The Europe-averaged heatwave trend during 1940–2020 (0.87 days per decade) is well captured by the multi-model mean (MMM) response with GHGs dominating the trend. The positive heatwave trend in GHGs and ozone is offset by the effects of aerosols during 1940–1979, leading to weak negative heatwave trends. In contrast, the increase in GHGs has driven about half (53 ± 17%; MMM and model-spread) of the strong heatwave trends in 1980–2020 (2.5 days per decade), amplified by the reduction in aerosols (23 ± 15%). This highlights the increasing risk of more frequent heatwaves in Europe if GHG emissions continue to rise without significant mitigation measures. Analysis of atmospheric circulation by SOMs reveals that four major atmospheric circulation patterns, dominated by a blocking high anomaly, are linked to the most spatially-intense European summer heatwaves. A relatively large increase in the occurrence of blocking-like atmospheric circulation has likely exacerbated heatwave trends in Southern and Eastern Europe in 1980–2020. However, this atmospheric circulation trend is much weaker in the model response, and also seems to be outside the internal variability in most of the models. This may partly explain the underestimated heatwave trends in Southern and Eastern Europe. Constraining and further understanding of the thermodynamic and dynamic response in the LESFMIP models is important for attributing and predicting the multi-annual and decadal variability of climate and weather extremes.