A Long-term Numerical Study of the Potential Predictability of Seasonal Mean Fields of Water Resource Variables using MRI/JMA-AGCM

Abstract

Using the Meteorological Research Institute of Japan Meteorological Agency, atmospheric global model (MJ98), the potential predictability is investigated (maximum possible predictability when sea surface temperature (SST) is perfectly predicted) of the seasonal mean fields of four water resource variables, i.e. precipitation minus evaporation (P - E), the terrestrial water storage, snow water equivalence, and runoff. 50-year (1949-1998) ensemble integrations are performed from six different initial conditions, forced with the same SST and sea ice cover.
The seasonal mean field variance ratios of the SST-forced variability, to the total variability, are computed to measure potential predictability of the four variables. The variance ratios of P - E are generally high in the tropics, but low in the extratropics. The geographical pattern of potential predictability of the total terrestrial water storage is similar to that of P - E, since it is the net water flux into the ground surface. The variance ratios of snow water equivalence were as low in DJF as those of precipitation at high latitudes. However, the values in the coastal area of the Gulf of Alaska is high in MAM, although the other regions remained low. The geographical distribution of the variance ratio of runoff, has similar features to that of the total terrestrial water storage, but the values of the former are slightly lower than those of the latter. High variance ratios are found in some areas of the extratropics. Singular value decomposition (SVD) analysis suggests that the ratios of P - E, and snow water equivalence, are due to the teleconnection forced by the tropical SST, while the higher variance ratio of the total terrestrial water storage is due to its persistence.