Seasonal prediction and predictability of Eurasian spring snow water equivalent in NCEP Climate Forecast System version 2 reforecasts

Abstract

The spring snow water equivalent (SWE) over Eurasia plays an important role in East Asian and Indian monsoon rainfall. This study evaluates the seasonal prediction capability of NCEP Climate Forecast System version 2 (CFSv2) retrospective forecasts (1983–2010) for the Eurasian spring SWE. The results demonstrate that CFSv2 is able to represent the climatological distribution of the observed Eurasian spring SWE with a lead time of 1–3 months, with the maximum SWE occurring over western Siberia and Northeastern Europe. For a longer lead time, the SWE is exaggerated in CFSv2 because the start of snow ablation in CFSv2 lags behind that of the observation, and the simulated snowmelt rate is less than that in the observation. Generally, CFSv2 can simulate the interannual variations of the Eurasian spring SWE 1–5 months ahead of time but with an exaggerated magnitude. Additionally, the downtrend in CFSv2 is also overestimated. Because the initial conditions (ICs) are related to the corresponding simulation results significantly, the robust interannual variability and the downtrend in the ICs are most likely the causes for these biases. Moreover, CFSv2 exhibits a high potential predictability for the Eurasian spring SWE, especially the spring SWE over Siberia, with a lead time of 1–5 months. For forecasts with lead times longer than 5 months, the model predictability gradually decreases mainly due to the rapid decrease in the model signal.