Impact of Climate Change on Forecast Perturbation Growth

The influence of climate change on weather forecast predictability is examined through the analysis of ensemble perturbation growth.  Information from the National Center for Atmospheric Research Community Climate Modeling System version 4 (NCAR CCMS4) Representative Concentration Pathway 8.5 (RCP8.5) for the current climate and the predicted climate of 2100 is assimilated into the Navy Global Environmental Model (NAVGEM) using the operational NRL Atmospheric Variational Data Assimilation System – Accelerated Representer (NAVDAS-AR) four-dimensional data assimilation system.  Random initial perturbations are then introduced and evolved in NAVGEM to produce 15-day ensembles.  In the study in which this technique is introduced and described, (McLay et al., 2016, Q.J.R. Meteorol. Soc.. doi: 10.1002/qj.2806), statistically significant changes in ensemble perturbation variances and growth rates are found in the future scenario.  These changes include notable increases in ensemble uncertainty in the polar regions and at jet level in the winter, and increases in ensemble uncertainty in the summer for moist metrics. In that first paper, these changes are related to changes in both baroclinic and convective instability.  Here we show new results that expand upon the first study by examining the differences in perturbation growth between the current and future scenarios from several new perspectives. We find that much of the change in ensemble variance is closely related to changes in the temporal variance. That is, while ensemble perturbation growth itself may look distinctly different in the current and future climates, ensemble perturbation growth normalized by temporal variance looks very similar. This is particularly noticeable when considering specific humidity. We also find that while ensemble variance itself changes in the future climate, the fraction of forecast error explained by the ensemble remains similar.  One particularly striking result is the substantial increase in ensemble variance (and inferred decrease in predictability) in the lower stratosphere.   Examination of power spectra shows that while the temporal variance in the stratospheric fields increases primarily on the small scales, perturbation growth increases on all spatial scales. Changes in stratospheric temporal variability and ensemble variance are related to changes in the stratospheric jet and tropospheric baroclinic instability.   The change in the linearity of the perturbation growth will also be examined and implications for weather predictability in future climates will be discussed.