Tropical systematic and random error energetics based on NCEP (MRF) analysis-forecast system - A barotropic approach ― Part Ⅰ: in physical domain

摘要

Deterministic predictability in the perspective of systematic and random error and their growth rates and different components of growth rate budgets like flux, pure generation, mixed generation and conversion in energy/variance form are investigated in physical domain for medium range tropical (30°S-30°N) weather forecast using daily horizontal wind field at 850 hPa up to 5-day forecast for the month of June, 2000 of NCEP (MRF) model. The study reveals the following: 1. The Indian peninsula, the Indonesian region and their adjoining areas over 10°N-20°N latitudinal belt show a large amount of forecast error variance indicating that cumulus parameterization process may play a major role in the generation of tropical systematic error. 2. Sparse observational networks over the tropical region are attributed to the uniform spread of random error over the continental as well as oceanic area. The results suggest that generation of random error in some geographical locations is perhaps due to the inefficient description of sensible heating process in the model. 3. As far as growth rates are concerned, systematic error growth rate increases at initial forecast time and attains maximum value at 2-day forecast then it remains unchanged for rest of the forecast days. Whereas, the growth rate of random error is nearly invariant at 1 and 2-day forecasts and then it increases slowly at subsequent forecast time. 4. Analyzing the flux, pure generation, mixed generation and conversion terms involved with the components of systematic and random error growth rate budget, it is shown that the components have their large variance in those regions where the respective error predominates.