Intermediate time error growth and predictability: tropics versus mid-latitudes

Abstract

The evolution of identical twin errors from an atmospheric general circulation model is studied in the linear range (small errors) through intermediate times and the approach to saturation. Between forecast day 1 and 7, the normalized error variance in the tropics is similar to that at higher latitudes. After that, tropical errors grow more slowly. The predictability time τ taken for tropical errors to reach half their saturation values is larger than that for mid-latitudes, especially for the planetary waves, thus implying greater potential predictability in the tropics. The discrepancy between mid-latitude and tropical τ is more pronounced at 850 hPa than at 200 hPa, is largest for the planetary waves, and is more pronounced for errors arising from wave phase differences (than from wave amplitude differences). The spectra of the error in 200 hPa zonal wind show that for forecast times up to about 5 d, the tropical error peaks at much shorter scales than the mid-latitude errors, but that subsequently tropical and mid-latitude error spectra look increasingly similar. The difference between upper and lower level tropical τ may be due to the greater influence of mid-latitudes at the upper levels.