70 Major river basins are defined using TRIP. Each of these areas is the catchment of a river gauge station where discharge data are available. "Observed" terrestrial water storage is calculated by basin-atmosphere water balance (Masuda, Oki and Yatagai, in preparation; similar to Masuda et al. 2001, GRL) making use of discharge data (GRDC and others combined) and atmospheric reanalysis data (NCEP/NCAR, NCEP R2 and ERA15). Only the relative value of storage is obtained from this method. On the other hand, "modelled" terrestrial water storage is calculated as the sum of soil water (including frozen part), snow water equivalent and river water storage. The climatological seasonal cycle of terrestrial water storage obtained by the two method agree each other generally well in many basins in the cold region and the temperate zone. In the Lena river basin, eastern Siberia, it is found that river water storage is overestimated probably because the flow velocity assumed in our version of TRIP is too low for this river. In some part of the cold region, mainly European Russia, model water storage in winter is much larger than observed, and it is probably overestimation of snow water equivalent due to over-correction of input precipitation. In the tropics, the magnitude of moisture convergence is different from one reanalysis to another, and some "observational" results agree with the model and seem realistic, but some do not.
Based on satellite-derived Surface Radiation Budget (SRB) release 2 data set produced by NASA Langley Research Center, specifically its 3-hourly monthly longwave data set, "observed" diurnal range of clear-sky upward longwave radiation at the surface is calculated. It should be noted that the quality of the data is good only in the coverage of geostationary satellites. This observed variable is compared with the amount of soil water of the uppermost layer of the model (0 -- 5 cm). As anticipated, a large diurnal range corresponds to a small value of soil moisture, and this out-of-phase relationship between these variables holds well in many areas. In some areas such as China, however, the correspondence is not good. We guess that some factors not represented either in the forcing of GSWP or the MATSIRO model may be at work in these areas.
We have calculated potential evaporation following the formula of Kondo and Xu (1997, Tenki; also discussed by Yamazaki et al. 2004 J. Hydromet.) with the forcing data of GSWP2. The ratio between precipitation and potential evaporation is considered a climatological index of surface wetness, and its distribution has general correspondence with that of vegetation. We expect that the ratio between actual and potential evaporation is also a function of the climate of the place, but it varies due to the activity of plants and also due to human activity such as irrigation. We plan to analyze this ratio in the results of MATSIRO in comparison with those of a bucket-type model and to discuss the effect of plants, though we have not finished the work by the deadline of the abstract.
Supplementary URL: