Inconsistent Mesoscale Spectra in Models and Reanalyses

Analyses and 10-day forecasts were conducted every 12 hours for the period of January-March, 2016, using NOAA NCEP’s Global Forecast System (GFS) at high resolution (T1534; hereafter “high-res”) and low resolution (T254; hereafter “low-res”). The 200-hPa kinetic energy (KE) spectra of the GFS high-res and low-res analyses, ERA-Interim (European Reanalysis Interim; resolution T254), and JRA-55 (Japanese 55-year Reanalysis; resolution T319) were found to be inconsistent with each other at mesoscales and smaller scales. The discrepancies among the analyses exceed an order of magnitude near the smallest representative scales of no aliasing risk, and none of them follows the Nastrom-Gage spectra found in wind observations along aircraft routes. The spatial patterns of the mesoscale band-passed KE, as well as their associated local spectrum slopes, clearly indicate the effect of terrain and convection on the KE spectrum.

The GFS high-res forecasts, relative to the low-res forecasts, have larger KE in the small scales, but the KE increases with forecast lead time. Interestingly, applying stochastic perturbations in the GFS low-res forecasts does not change the global KE spectra, which implies that stochastic perturbations only redistribute KE in space and/or horizontal diffusion strongly dissipates energy anomalies in the model. Consistent with seasonal migration, the spatial patterns of mesoscale band-passed KE show southward displacement in January-March, 2016. We thus conclude that (a) using isotropic, homogeneous, or time-invariant framework to study the KE spectrum behavior needs caution; (b) the Nastrom-Gage spectra were not re-produced in widely-used reanalysis products of different resolutions (i.e., ERA-Interim and JRA-55), and can be easily manipulated in forecast models; (c) pursuing Nastrom-Gage spectra in GCMs may be misleading, and more observations for finding true global KE spectrum are needed.