Modelo de regresión lineal múltiple para determinar influencias del índice Niño 1+2 y la MJO sobre las precipitaciones en Guayaquil durante enero-febrero-marzo y abril
There were elaborated multiple linear regression models to determine the adjustment degree of niño 1+2 and Madden-Julian Oscillation indexes with the accumulated precipitation in Guayaquil, for january, february, march, and april. The correlation models were built separately for each month with time series from 1978 to 2004 of monthly accumulated precipitation, niño 1+2 anomaly, and Madden-Julian Oscillation index for 120W, with a significance level of 95%. It was found that for january, march, and april the linear correlation adjust with coefficients R2 of 0.80, 0.77, and 0.82 respectively. The precipitation levels for february does not showed significant correlation with the indexes, it seems to respond to other factors, presumably atmospheric. There were made comparisons between the values calculated with the models and the observed precipitation levels of the years 2005, 2006, and 2007 in order to determine the hit score of the forecast. To perform comparisons there were established three ranges based on the tercicles of the observed data time series. The ranges were defined as below normal (lower tercil), normal (middle tercil), and above normal (upper tercil). It was found that for march the skill score is 100% however, forecasted values for january and march, showed a low skill score. The regression model found may serve as a tool to find relations between precipitations, sea temperature and atmospheric convergence.
| Main Author: | |
|---|---|
| Format: | Journal Contribution biblioteca |
| Language: | Spanish / Castilian |
| Published: |
Instituto Oceanográfico de la Armada, Guayaquil, Ecuador
2007
|
| Subjects: | Atmospheric circulation, Mathematical analysis, Atmospheric precipitations, Seasonal variations, Surface temperature, |
| Online Access: | http://hdl.handle.net/1834/2380 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | There were elaborated multiple linear regression models to determine the adjustment degree of niño 1+2 and Madden-Julian Oscillation indexes with the accumulated precipitation in Guayaquil, for january, february, march, and april. The correlation models were built separately for each month with time series from 1978 to 2004 of monthly accumulated precipitation, niño 1+2 anomaly, and Madden-Julian Oscillation index for 120W, with a significance level of 95%. It was found that for january, march, and april the linear correlation adjust with coefficients R2 of 0.80, 0.77, and 0.82 respectively. The precipitation levels for february does not showed significant correlation with the indexes, it seems to respond to other factors, presumably atmospheric. There were made comparisons between the values calculated with the models and the observed precipitation levels of the years 2005, 2006, and 2007 in order to determine the hit score of the forecast. To perform comparisons there were established three ranges based on the tercicles of the observed data time series. The ranges were defined as below normal (lower tercil), normal (middle tercil), and above normal (upper tercil). It was found that for march the skill score is 100% however, forecasted values for january and march, showed a low skill score. The regression model found may serve as a tool to find relations between precipitations, sea temperature and atmospheric convergence. |
|---|