Predictability of the South Asian monsoon in the CFSv2 Operational Forecast Model

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 8 January 2015
Subhadeep Halder, George Mason University, Fairfax, VA; and P. A. Dirmeyer, L. Marx, and J. L. Kinter III

The global Climate Forecast System Version 2 (CFSv2) model of the National Centers for Environmental Prediction (NCEP) is used to quantify and isolate the predictability of the Indian monsoon arising from slowly varying land surface states from that of the ocean, at sub-seasonal to seasonal time scales. Ensemble hindcasts of 4-month duration are initialized in each of 28 years spanning 1982-2009, from the beginning of late spring and early summer months (at 0000UTC on the 1st of April, May, June and July). For each initial date, the baseline ensemble member is initialized from the NCEP CFSR reanalysis data for that date. The other ensemble members are initialized with identical atmosphere, sea ice and ocean states from the baseline simulation, but initial land states from each of the remaining 27 years, thus achieving maximum perturbation. Land states include soil water and ice content, soil and vegetation temperatures, and snow mass. The magnitudes of signal to total ratios of variance are used to quantify predictability of several parameters such as precipitation and surface temperature, and identify global ‘hot spots' for this model. Several coupling metrics are also used to quantify and evaluate the terrestrial and atmospheric components of land-atmosphere feedbacks. Preliminary results reveal biases in the simulation of coupled land-atmosphere feedbacks in the model that appear to be driven by factors such as the parameterization of surface evapotranspiration, boundary layer, radiation and convection as well as the representation of vegetation. Detailed results will be presented.