Open AccessExploring the relationship between complexity and performance in a land surface model using the multicriteria method
Author(s) -
Leplastrier M.,
Pitman A. J.,
Gupta H.,
Xia Y.
Publication year - 2002
Publication title -
journal of geophysical research: atmospheres
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.67
H-Index - 298
eISSN - 2156-2202
pISSN - 0148-0227
DOI - 10.1029/2001jd000931
Subject(s) - calibration , energy balance , mode (computer interface) , representation (politics) , surface runoff , evaporation , environmental science , energy (signal processing) , surface (topology) , set (abstract data type) , water balance , computer science , mathematics , meteorology , statistics , physics , geology , thermodynamics , geometry , ecology , political science , law , programming language , operating system , geotechnical engineering , politics , biology
The performance of five modes of a land surface model, the Chameleon Surface Model (CHASM), was investigated after calibration via the multicriteria method to monthly totals of evaporation and runoff from the Valdai data set. The use of CHASM allows for an exploration into the relationship between surface energy balance complexity and optimal performance by isolating the impacts of different parameterizations of the surface energy balance. When compared to quantities used within the calibration process, CHASM's performance was significantly increased with calibration over default simulations regardless of calibration length or mode complexity. Within the calibration period, CHASM's performance increased with increasing complexity in the representation of the surface energy balance. Outside the calibration period there was little improvement to simulations from additional complexity in the surface energy balance representation above the simplest mode. Calibration is shown to reduce the scatter between modes suggesting that some of the differences between models in PILPS Phase 2d may be explained by the specification of parameter values. For simulations of quantities not used in calibration, performance can be reduced as a result of calibration. This implies that evaporation and runoff may not be the best quantities for calibration in order to improve model performance. It is suggested that the best quantities to calibrate may be mode and model specific.