Abstract
The reliability of predictions from climate and weather models is linked to an adequate representation of the land surface processes. To evaluate performance and to improve predictions, land surface models are calibrated against observed data. Despite an extensive literature describing methods of land surface model calibration, few studies have applied a calibration method for semiarid natural vegetation, especially for the semiarid northeast of Brazil, which presents caatinga as its natural vegetation. Caatinga is a highly dynamic ecosystem with the physics at the land surface–atmosphere interface still poorly understood. Therefore, in this study a multiobjective hierarchical method, which provides means to estimate optimal values of the model parameters through calibration, is evaluated. This method is applied to caatinga by using the Integrated Biosphere Simulator (IBIS). Results demonstrated that the calibrated set of vegetation parameters produced a considerably different energy balance from the default parameters. In general, the model was able to simulate the partition of the available energy into sensible and latent heat fluxes when the calibrated parameters were used. The IBIS model was not able to capture short-term, intense changes in latent heat flux from a dry condition to a wetter condition, however, even when the new set of calibrated parameters was used. Therefore, the parameter optimization may not be sufficient if processes are missing or misrepresented. This study is one of the first to understand the physics at the land surface–atmosphere interface in the caatinga ecosystem and to evaluate the ability of the IBIS model to represent the biophysical interactions in this important ecosystem.
