Impact of Vegetation Treatment and the Role of Land-Atmosphere Coupling on Drought Prediction in the NCEP CFSv2

The physical parameterization of the land surface processes and the representation of surface characteristics play a crucial role in the partitioning of heat and moisture fluxes that contribute to the growth and attributes of the atmospheric boundary layer that produce convection and, in turn, can provide a feedback mechanism to the land surface. Consequently, a better representation of land model physics and the coupling processes is important to the improvement of forecast models.

Study has shown that the hindcast climatological Coupled Drought Index (CDI) in the NCEP Coupled Forecast System Version 2 (CFSv2) deviates quickly from the reanalysis-based CDI into a wetter state. In addition, land-atmospheric coupling breaks down in CFSv2 during drought conditions (dry coupling) leading to the weakening and termination of the drought conditions. The loss of seasonal forecast drought skill is attributed to the failure of CFSv2 to hold drought conditions.

Preliminary analyses indicate that increased (anomalous) terrestrial evapotranspiration in CFSv2 is leading to its inability to hold drought conditions. There are two hypotheses for the cause of this erroneous increase. One is that it is caused by parameter changes that increase the uptake rate of deep soil waterin CFSv2’s Noah land model n an attempt to correct a warm bias by increasing surface evaporation.An alternative hypothesis is that the increase in evapotranspiration is due to a lack of dynamic vegetation in the model, which allows for continued transpiration during a drought event (due to the use of a vegetation phenology based on climatology).

We conduct CFSv2 ensemble reforecast experiments for selected 11 years and use the summer drought of 2011 as a case study to examine the two hypotheses set above and answer the scientific question “To what degree can better representation of land-atmospheric coupling processes improve the drought prediction skill in CFSv2?”. This is mainly accomplished by a) using realtime vegetation fraction observations in the CFSv2 runs to identify the moisture source in CFSv2, and b) coupling the CFSv2 with an advanced Noah land model with Multiple Parameterization (Noah-MP) choices where the parameterization for subsurface hydrologic processes and dynamic vegetation options are available. The CFS skills in predicting Sea Surface Temperature (SST), precipitation, and 2-meter temperature (T2m) anomalies are compared and a brief discussion of the differences is presented.