Search

Real-time medium-range weather forecasts (up to 10 days in advance) are prepared at the National Centre for Medium Range Weather Forecasting (NCMRWF), India using the UK Met Office Unified Model (NCUM) system. In this study, the dynamical characteristics of systematic errors of the NCUM model at 17 km horizontal resolution during the Indian summer monsoon (July and August 2017) are examined. It is found that the model has problem in maintaining the monsoon circulation and it weakens the monsoon by weakening the divergent winds over the Indian region. The model has a large spin up in global mean precipitation and evaporation from day-1 to day-3. In addition to the spin up problem, the model loses mass as the forecast length increases from day-1 to day-10. However, the mean surface pressure over India increases in forecasts from day-3 up to day-10. An anticyclonic circulation error over north Arabian Sea and adjoining region reduces moisture transport from the sea to the Indian land areas. The model has systematic tendency to forecast less rain as the forecast length increases. Differences between the ERA-Interim and the NCUM analysis in circulation and moisture parameters indicate uncertainties in the initial conditions and imbalance in moisture and divergence. Relationship between circulation errors and rainfall errors over a river basin (Mahanadi basin in Odisha) was studied. There is no linear relation between divergent or rotational components of wind errors and rainfall errors in forecasts over this basin. Even with a very small error in wind, the rainfall error could be as large as 10 cm/day or more in the basin. [ABSTRACT FROM AUTHOR]