The Role of Land-Atmosphere Interactions in the Improvement of Heatwave Prediction

Abstract

Increased heatwave frequency across the United States has led to the need for improved predictability of heatwave events. A detailed understanding of land-atmosphere interactions and the relationship between soil moisture and temperature extremes could provide useful information for prediction. This study identifies, for many locations, a threshold of soil moisture below which there is an increase in the sensitivity of atmospheric temperature to declining soil moisture. This shift to a hypersensitive regime causes the atmosphere to be more susceptible to atmospherically driven heatwave conditions. The soil moisture breakpoint where the regime shift occurs is estimated using segmented regression applied to observations and reanalysis data. It is shown that as the soil gets drier, there is a concomitant change in the rate of decrease in latent heat flux and increase in sensible heat flux leading to a strong positive feedback of increased air temperature near the surface, which further dries out the soil. Central, southwestern, and southeastern parts of the US seem to have regions of clear regime shifts, while the eastern part of the US generally does not get dry enough to reveal significant breakpoints. Sensible heat flux is seen to be a primary driver of this increased temperature sensitivity aided by the drop in latent heat flux. An investigation of flux tower sites verifies the breakpoint-flux relationships found in reanalysis data. Accurate estimation of these breakpoints can contribute to improved heatwave prediction. This study also investigates forecast model dataset from the Subseasonal Experiment project (SubX) and the Unified Forecast Systems Subseasonal to Seasonal prototype 5, 6 and 7 (UFS p5, p6 and p7) to draw comparisons to the results found in the study of observations and reanalysis data. Model skill in accurately representing the regime shifts and underlying wetness of the region of study is determined using various skill metrices. The connections between the subseasonal forecast models’ heatwave prediction and their ability to accurately represent the aforementioned land-atmosphere coupling relationships are also quantified. The models are found to be better at extreme heat prediction than at representing soil moisture values within the right regime of sensitivity. They also perform better at extreme heat day forecasts when they are in drier conditions. In addition to diagnosing dynamical models, this study seeks to provide a new statistical approach to predicting heatwaves exacerbated by soil moisture conditions.