Towards improving GFS Forecasts over India: Sensitivity of GFS convection scheme to moisture

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 8 January 2015
Suvarchal Kumar Cheedela, Univ. of Miami, Miami, FL; and A. Sukumarapillai, I. K. Hu, and B. E. Mapes

Monsoon mission aims to develop selected weather models for a better forecast of Indian Monsoon through an international collaboration. One of the two key models adapted for the mission is Climate Forecast System (CFS), a coupled model developed at the National Center for Environmental Prediction (NCEP). Current work concerns the convective parametrization of Global Forecast System (GFS) model, the atmospheric component of CFS. Many biases have been documented for monsoon forecasts using CFS, these biases can often be traced to biases in the atmospheric component, GFS. One such bias is limited precipitation over north of Bay of Bengal, and excess precipitation over the equatorial Indian ocean, at all lead times. Our primary objective is to reduce this bias in the model, without spoiling the overall climate and weather simulation. Our hypothesis about this bias is that it may stem from insufficient sensitivity of the existing GFS convective parametrization to environment moisture, a problem common in many models. Since GFS currently uses a convective parametrization with entrainment based on relative humidity of the environment, optimization efforts might achieve gains without extensive new coding requirements. We will modify this formulation in a series of experiments, attempting to design the modifications to be global-climate-neutral so that the possible gains can be seen without requiring whole-model retunings. Initialized hindcats, and studies of model tendency budgets, will be utilized to diagnose and evaluate our modifications without requiring long runs. By closing the diagnostic loop more tightly with such short runs, we hope to be able to try many more experiments.