Premium
Downscaling large‐scale circulation to local winter rainfall in north‒eastern Mexico
Author(s) -
Cavazos Tereza
Publication year - 1997
Publication title -
international journal of climatology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.58
H-Index - 166
eISSN - 1097-0088
pISSN - 0899-8418
DOI - 10.1002/(sici)1097-0088(199708)17:10<1069::aid-joc183>3.0.co;2-i
Subject(s) - climatology , downscaling , environmental science , precipitation , atmospheric circulation , geography , geology , meteorology
The large‐scale atmospheric controls on winter rainfall variability for north‒eastern Mexico and south‒eastern Texas are diagnosed based on 8 years (1985–1993) of daily data from the Goddard Space Flight Center four‒dimensional data assimilation scheme. Downscaling techniques based on artificial neural nets are used to derive transfer functions from the large‒scale circulation to local precipitation. Daily rainfall amounts are predicted from synoptic sea‒level pressure, 500 hPa heights, and the 1000–500 hPa thickness, and summed over the cool season (NDJF). The monthly rainfall totals are also predicted independently from other indices of the large‒scale circulation, such as 1000–500 hPa thickness, the Pacific/North American (PNA) pattern, and a standardized Southern Oscillation Index (SOI) derived from the Tahiti minus Darwin sea‒level pressure difference. Time‒lagged component scores from a rotated principal component analysis of sea‒level pressure, 500 hPa heights, and 1000–500 hPa thickness serve as input to a neural net that produces time‒series of daily rainfall amounts for 20 grid‒points in the study area. The correlation between the observed and predicted rainfall is greater than 0·7 in the coastal plains of the Gulf of Mexico and less than 0·7 over the Sierra Madre and the Gulf, suggesting an increase in the importance of local rainfall processes in the last two regions. The analysis shows a systematic relationship between the performance of the net and physiography, which is confirmed by the consistency of the patterns of spatial correlation, mean absolute error, and root‒mean‒square error at different time‒scales. The net captures the phase and amplitude of most of the rainfall events, reflecting the influence of the large‒scale circulation. However, interannual fluctuations in rainfall associated with persistent El Niño conditions are partly bypassed by the net, suggesting that more information is needed to predict extreme events. The PNA pattern does not seem to be associated with local rainfall in north‒eastern Mexico and south‒eastern Texas during the period analysed. © 1997 The Royal Meteorological Society.