May 6th, 2025
Key Findings
- The aim of this study was to determine whether vapor pressure deficit (VPD) variability is predictable over the Western U.S. and to understand the main source of predictability.
- Seasonal forecasts using GFDL’s SPEAR outperform persistence forecasts across the Western U.S.
- Weighted model-analog forecasts show comparable forecast skill to the VPD predictions generated by the SPEAR seasonal forecast system, offering a simple yet complementary tool for VPD forecasting.
- In general, weighted model-analog forecasts outperform their unweighted counterparts. Nevertheless, the unweighted model-analog forecasts demonstrate better skill than the VPD-only forecasts, indicating the presence of additional sources of predictability beyond VPD alone.
- By incorporating different predictors into the forecasting process, the authors found that sea surface temperature, particularly in the tropical Pacific, serves as the primary source of predictability for VPD variability in the region.
Jiale Lou, Youngji Joh, Thomas L. Delworth, Liwei Jia. npj Climate and Atmospheric Science. DOI: 10.1038/s41612-025-01028-6
Fire hazards can have catastrophic effects on human society and ecosystems, posing a pronounced threat to public health, food security, infrastructure, assets, and natural resources. Over the past few decades, fire events have been increasing in size and severity worldwide. Atmospheric vapor pressure deficit (VPD) measures the difference between saturation vapor pressure and actual vapor pressure, and its variability is closely related to fire activity in the Western U.S.
This study employed both observations and state-of-the-art models to compare different forecasting methods to understand the forecast skill and predictability of VPD variability over the Western U.S. and to understand the main source of predictability. It showed that dynamical forecasts using GFDL’s state-of-the-art dynamical forecast system, SPEAR, demonstrated skillful predictions of VPD variability in the Western U.S., exceeding the forecast skill of persistence forecasts. Further, their findings suggest that sea surface temperature is a critical source of VPD predictability over the Western U.S.
The authors compared the forecast skill of monthly VPD variability, using SPEAR, to two benchmarks: persistence and model-analog forecasts. Seasonal forecasts, large ensemble simulations, and pre-industrial control simulations from SPEAR were used. Dynamical forecasts (SPEAR) demonstrated skillful predictions of VPD variability in the Western U.S., exceeding the persistence forecast skill, which indicates additional sources of VPD predictability within the climate system.
In the model-analog framework, the authors selected analog states resembling the observed initial conditions from the model space, and the subsequent evolution of those initial model-analogs yields forecast ensembles. To quantify the contribution of different climate variables to VPD prediction, they developed a weighted model-analog forecast and evaluated its skill in comparison to VPD-only and unweighted forecasts. The optimally weighted model-analog exhibits forecast skill for VPD variability comparable to that of the dynamical forecast system.
It is urgent to better understand the causes and consequences of fire events and the extent to which such high-impact weather/climate events can be predicted on seasonal-to-interannual time scales. By gaining a better understanding of fire weather conditions and improving predictions, communities could take preparatory actions, and decision-makers could better allocate resources and develop fire management plans.
