Please use this identifier to cite or link to this item:
標題: Model structure and land parameter identification: An inverse problem approach
作者: Tsuang, B.J.
Tu, C.Y.
關鍵字: inverse problem;parameter identification;Taipei;Hualien;evapotranspiration;land surface processes;simple biosphere model;general-circulation models;tropical forest;soil-moisture;circuit model;surface;climate;evapotranspiration;vegetation;sib
Project: Journal of Geophysical Research-Atmospheres
期刊/報告no:: Journal of Geophysical Research-Atmospheres, Volume 107, Issue D10.
[1] Model structures for the parameterization of energy and water exchanges between the atmosphere and the land are explored, and the parameters are identified. Minimizing the root-mean-square error (RMSE) of the calculated air temperature is used as the objective function. Eight parameters are determined within 14 iterations over two urban sites in a subtropical island. The identified values of these parameters can be compared with those listed in the literature. The results show that (1) a proper choice of canopy resistance is important in determining the mean temperature; (2) using a scheme to calculate stomatal resistance from vegetated areas shows improvements in simulating the diurnal variations of the temperature, decreasing the RMSE by 0.4 K; (3) using a scheme to accommodate the effect of evaporation from skin reservoirs during rain events shows improvements in determining the temperature immediately after the rain, decreasing the RMSE by 0.1 K; (4) while incorporating a soil-moisture availability function does not show improvement in the model performance unless an extra water table variable is added, which decreases the RMSE by 0.03 K. Incorporating all of the above model structures, the calculated hourly air temperature has a correlation coefficient of as high as 0.97 with a RMSE of 1.4 K.
ISSN: 0747-7309
DOI: 10.1029/2001jd000711
Appears in Collections:環境工程學系所

Show full item record

Google ScholarTM




Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.