Understanding the prediction and structure of Indian monsoon low-pressure systems in the Subseasonal-to-Seasonal prediction project

Abstract:

Indian monsoon low-pressure systems (LPSs) are synoptic-scale cyclonic vortices which produce abundant precipitation over the Indian subcontinent. Despite their capability to cause catastrophic flooding, the potential for the predictability of these weather systems on the timescales of numerical weather prediction and extended range models remains less explored. In this work, we examine in detail the prediction and structure of LPSs in the Subseasonal-to-Seasonal (S2S) prediction project. Using a feature tracking algorithm, LPSs are identified in eleven S2S models during a common re-forecast period of June-September 1999-2010 and then compared with results from the ERA-Interim (ERA-I) reanalysis dataset. Forecast verification statistics are produced for the frequency, position and intensity of LPSs. In addition, composite three-dimensional structures of LPSs in all S2S models are compared with ERA-I.

The results show that all S2S models under-simulate the frequency of LPSs. Their transits are represented by all models, but large biases are observed in the Australian Bureau of Meteorology, China Meteorological Administration (CMA), Environment and Climate Change Canada and Hydrometeorological Centre of Russia (HMCR) models. The CMA model exhibits the highest position error and the intensity of LPSs is overestimated by most models. The CMA, National Centers for Environmental Prediction and UK Met Office models have the best ensemble spread-error relationship for the position and intensity, whereas the HMCR model has the worst. In general, all models simulate large-scale features of LPSs. Except the HMCR model, all models simulate the warm-over-cold thermal structure of LPSs. In all models, the vertical structure of relative vorticity is shallower and weaker than in ERA-I. In most models, LPS composites feature drier middle and upper troposphere than in ERA-I.