Reproducibility of Carbon and Water Cycle by an Ecosystem Process Based Model Using a Weather Generator and Effect of Temporal Concentration of Precipitation on Model Outputs

Abstract:

GCM is generally used to produce input weather data for the simulation of carbon and water cycle by ecosystem process based models under climate change however its temporal resolution is sometimes incompatible to requirement. A weather generator (WG) is used for temporal downscaling of input weather data for models, where the effect of WG algorithms on reproducibility of ecosystem model outputs must be assessed. In this study simulated carbon and water cycle by Biome-BGC model using weather data measured and generated by CLIMGEN weather generator were compared.
 The measured weather data (daily precipitation, maximum, minimum air temperature) at a few sites for 30 years was collected from NNDC Online weather data. The generated weather data was produced by CLIMGEN parameterized using the measured weather data. NPP, heterotrophic respiration (HR), NEE and water outflow were simulated by Biome-BGC using measured and generated weather data. In the case of deciduous broad leaf forest in Lushi, Henan Province, China, 30 years average monthly NPP by WG was 10% larger than that by measured weather in the growing season. HR by WG was larger than that by measured weather in all months by 15% in average. NEE by WG was more negative in winter and was close to that by measured weather in summer. These differences in carbon cycle were because the soil water content by WG was larger than that by measured weather. The difference between monthly water outflow by WG and by measured weather was large and variable, and annual outflow by WG was 50% of that by measured weather. The inconsistency in carbon and water cycle by WG and measured weather was suggested be affected by the difference in temporal concentration of precipitation, which was assessed.