Evaluation of a Land-surface Model (Noah) for Semi-arid Environments Terri S. Hogue, Luis Bastidas, Hoshin Gupta, and Soroosh Sorooshian
Study Sites Lucky Hills (Desert Shrub) Kendall (Grassland)
Precipitation Patterns
Precipitation Stats
Testing Schemes Split Sample Proxy Basin
Kendall Objective Functions Normalized Values (to 1997-00)
Kendall Split Sample Results
Proxy-Basin Results – Kendall w/LHills
Differential Split Sample
Differential Split Sample Results
Kendall – 10-day monsoon period (1997-00 pars) OBSERVED CALIBRATED DEFAULT Sensible Heat Latent Heat Ground Heat Ground Temp W/m2 °C
Kendall – 10-day monsoon period (1999 pars) OBSERVED CALIBRATED DEFAULT Sensible Heat Latent Heat Ground Heat Ground Temp W/m2 °C
Preliminary Conclusions The developed methods provide a rigorous framework for comparisons of the behavior of land-surface models across various biomes: Parameter sets from nearly all calibration periods show improved performance over default simulations • Longer calibration-period parameter sets do not capture seasonal /annual variability There is a need for both “wet” and “dry” periods. The increase in latent heat during short climatic events is not modeled well. • Proxy site parameters results in similar performance for most periods
Differential Split Sample Results
Lucky Hills Split Sample Results
Lucky Hills Objective Functions Normalized Values (to 1997-00)