Abstracts for the 5th International GAME Conf.

3-5 October 2001

Aichi Trade Center

Nagoya Japan

Land Data Assimilation in the GAME-HUBEX

Kenji TANAKA (1), Osamu KOZAN (2), Tadanori NAKAMURA (3), Michiharu SHIIBA (2), Shuichi IKEBUCHI (1)

In this study, a dataset that is homogeneous in time and space was created by using meteorological and hydrological data obtained during HUBEX-IFO (1998/5/1-8/31). The domain of this dataset is from E110 to E122, and from N31 to N36 with 5min (about 10km) resolution. This dataset was produced to be used as forcing data for Land Data Assimilation by land surface scheme (SiBUC) and to be utilized to validate regional 4DDA by JSM-SiBUC. Although our goal is to create an hourly dataset, the majority of meteorological data in the HUBEX Project are at 3-hourly or 6-hourly intervals. Then, hourly data for 3-hourly or 6-hourly stations should be produced by being interpolated in space or time. Using hourly observed and hourly interpolated data at 146 stations, mesh dataset was created using spatial interpolation. This dataset had much difficulty in the description of diurnal variation of radiation and precipitation. Thus, satelite remote sensing data from GMS-5 was introduced. Through the comparison with field measurement, it was shown that only simple application of GMS VIS channel data could improve the diurnal variation of solar radiation. Although we still have difficulty in hourly precipitation, energy and water budget of the Huaihe River Basin was calculated in hourly increments by using SiBUC and the produced dataset to start the discussion of GAME-WEBS (Water & Energy budget Studies). Simulated runoff was smaller than observed one. Threfore, model parameters should be adjusted according to long-term water budget. Furthermore, the problem of initialization and updating of soil moisture is also handled. In the Huaihe river basin, most part of the basin is covered by agricultural cropland (paddy field and farmland). Therefore, significant amount of water is induced to the field by human activity. In that case, error in soil moisture is not only a problem of initialzation. To reduce the error in initial state and error caused by human activity (irrigation), data assimilation algorithms was designed to run in an operational mode. Temperature and soil moisture are closely related each other through evapotranspiration term which appears in both prognostic equations. Therefore, Extended Kalman Filter can be used to connect these variables in this highly non-linear system. State variables are three temperatures (Tc,Tg,Td) and three soil moistures (W1,W2,W3). System equations are prognostic equations of these variables. Measurement vector is surface temperatures (Tc,Tg) which are provided from GMS IR channel data. The system has been tested by using surface measurement data and GMS data. The difficulties, we are facing now, in applying GMS data is to estimate (divide) canopy and ground temperatures from IR data and other variables in the model.

Submittal Information

Name : Date :
    Kenji TANAKA
    30-May-01-20:50:20
Organization : Theme :
    Disaster Prevention Research Institute, Kyoto University
    Theme 2
Address : Presentation :
    Water Resources Research Center, D.P.R.I., Gokasho, Uji 611-0011
    Poster or oral
Country : Abstract ID :
    JAPAN
    T2KT30May01205020
Phone : Fax :
    +81-774-38-4246
    +81-774-32-3093
E-mail :
    tanaka@wrcs.dpri.kyoto-u.ac.jp