Thursday, 10 January 2013: 9:00 AM
Room 15 (Austin Convention Center)
The southeastern United States is no stranger to periods of dry and wet precipitation patterns, and has been the subject of study, partially due to the recent 2006-2007 drought. Given that the southeastern United States can and does experience periods of intense precipitation deficit and excess, prediction of precipitation in the region is important, not only from an agricultural standpoint, but also due to strain on water supply and financial resources. Authors have argued that a multi-model approach to prediction may surpass prediction skill in individual models, and we aim to examine predictability of southeastern US precipitation in the National Multi Model Ensemble (NMME) system, a newly funded multi-agency experiment including models from 7 institutional partners. There is also a potential linkage between observed southeastern US precipitation and tropical Pacific sea surface temperature anomalies (SSTA) or ENSO phase. If this is true, these strong teleconnections can lead to potential predictability of precipitation in the southeastern US, and given the above-mentioned reasons, we have chosen the southeastern US as an area of interest to test predictive skill in the NMME system. Authors have shown a relationship between El Nino Southern Oscillation (ENSO) and precipitation patterns in the southeastern US in winter, where warm (cold) ENSO events favor rain (drying) over the southeastern US. Recent contrasting arguments have shown that the ENSO-southeastern US precipitation link is weak during the winter half-year (November through April), and correlated only to internal atmospheric variability in the summer half-year (May through October). A stronger link between SSTA in the tropical Pacific and the southeastern US would lead to stronger predictability of precipitation. Though some papers point to a strong link between SSTA in the tropical Pacific and precipitation over the southeastern US, especially in winter, other results show that there may be limited prediction skill overall. We consider rank probability skill score and anomaly correlation as measure of predictive skill of southeastern US precipitation, and find weak positive RPSS in winter months and weak negative values in summer months. Positive values in winter indicate skill better than climatology. We also consider anomaly correlation of southeastern US model precipitation with observed, and find the highest positive correlation in winter months. This leads to the conclusion that if there is in fact predictive skill in the southeastern US, it is highest in during the winter. We also show a case study of the 2006-2007 drought, in which we find that though observations showed a strong period of precipitation deficit in FMA2007, the modeled drought did not persist through FMA2007, and instead showed a positive precipitation anomaly (fig. 1a, shown is forecast initialized in December2006, verifying FMA2007). We also find that the NMME system did not show the correct SST anomaly sign in the tropical Pacific during FMA2007 (fig. 1b). Notice that the observations show a thin, cold SST anomaly structure in the tropical Pacific, whereas the NMME system shows overall warm anomalies. This result is confirmed in seasonal lead times 2.5 through 8.5. We then compare the correlation and regression coefficients of Southeastern US precipitation vs. SSTs, and find that in observations as well as the NMME system, and find that the NMME system shows overall larger correlation as compared to observations. The results point to the conclusion that the discrepancies seen in FMA2007 could be explained by the absence of cold anomalies in the tropical pacific in the models, and that the focused area of variability seen in observations could be linked to precipitation anomalies in the southeast US. The spatial and intensity differences in 2006 could also be due to the reduced cold anomaly or possibly the structure of the anomaly itself. Also possible is that the models do not accurately reproduce the observed link to tropical Pacific SSTs. We aim to answer the question of the amount of prediction skill of southeastern US precipitation in a multi-model ensemble given the possibility of larger predictive skill using this approach, and discuss the southeastern US precipitation/tropical Pacific SST link within the NMME system. We use the 2006-2007 drought as an example of an extreme dry event in the southeastern US.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner