Deterministic Seasonal Quantitative Precipitation Forecasts: Benchmark Skill with a GCM

摘要

Many applications, from agricultural planning (such as crop choice) to estimation of hydro-power and surface water availability require quantitative precipitation forecasts (QPF). While a large number of studies have addressed the problem of forecasting seasonal Indian summer monsoon (ISM) rainfall over the decades, the emphasis generally has been on simulation and forecasting of rainfall anomalies from long-period mean. However, given the trends in monsoon rainfall, the procedure of considering anomalies has inherent errors; thus reconstruction of actual rainfall from the anomalies does not necessarily provide accurate information. Most QPF have been attempted at short-range forecasts, with a variety of techniques like model output statistics. This work represents evaluation of an atmospheric Variable Resolution General Circulation Model (VRGCM) for QPF during monsoon season. The VRGCM, with variable grid resolution provides relatively high (similar to 50 km) resolution over the ISM region, has been validated over all India as well as in different regions like Central and North India, South India, North East India for seasonal forecast of monsoon rainfall. VRGCM simulated QPF appears to provide comparable information as that of anomaly forecasts of monsoon over different regions in India. The forecast skill is appreciable, with significant correlation at all India, South India and North East India regions. The Root Mean Square Error of VRGCM in forecasting quantitative rainfall overall India and Central North India is very low and bit high over South and North East region of India. The performance of VRGCM in forecasting the rainfall in extreme (deficit or excess) monsoon years is also very high with phase synchronisation of 89% in the inter annual variability of rainfall during monsoon at all India scale while the value is about 57%, 76% and 55% over Central North, South and North East India region, respectively.