SIMULATION OF THE MONSOON SEASON IN SOUTH AMERICA

DURING EXTREME PHASES OF ENSO: 1997-1998 AND 1998-1999

Intraseasonal variations

Sensitivity experiments

Alice Grimm - Silvina Solman

Pablo ReyesICTP, Workshop on the Theory and Use of Regional Climate Models, 26 May – 6 June 2003

MOTIVATION

The monsoon season is the peak rainy season in most of South America.

Anomalies that are strong and consistent during shorter periods within the season are smoothed out in a seasonal analysis.

There are significant intraseasonal changes in the ENSO impacts possibly due to regional processes that are best represented in regional models.

The climate forecast for the monsoon season needs improvement and higher temporal resolution.

Domain and topography

Resolution: 100 km

EXPERIMENTSa) Seasonal runs to examine the intraseasonal variations

Cumulus convection scheme: Grell, with Fritsch and Chappell convective closureassumption.

ENC - 16 Oct 1997 to 02 Mar 1998LNC - 16 Oct 1998 to 02 Mar 1999

b) Short runs to test the cumulus convection schemes

Period: 16 Dec 1997 to 02 Feb 1998

ENJ/Grell - FC: Grell scheme, with Fritsch-Chappell convective closure assumption.

ENJ/Grell - AS: Grell scheme, with Arakawa-Schubert convective closure assumption.

ENJ/KUO: Anthes-Kuo scheme.

Seasonal runs & intraseasonal variationsa) El Niňo event 1997-1998 Precip

Seasonal runs & intraseasonal variationsa) El Niňo event 1997-1998 850 hPa winds

Seasonal runs & intraseasonal variationsa) El Niňo event 1997-1998 Temp 2m

Seasonal runs & intraseasonal variationsb) La Niňa event 1998-1999 Precip

Seasonal runs & intraseasonal variationsb) La Niňa event 1998-1999 850 hPa wind

Seasonal runs & intraseasonal variationsb) La Niňa event 1998-1999 Temp 2m

Seasonal runs & intraseasonal variationsc) Differences El Niňo - La Niňa Precip

Seasonal runs & intraseasonal variationsc) Differences El Niňo - La Niňa 850 hPa wind

Seasonal runs & intraseasonal variationsc) Differences El Niňo - La Niňa Temp 2m

Short runs &sensitivity to convection schemes Jan 1998

ConclusionsSEASONAL RUNS & INTRASEASONAL VARIATIONS

Although the rainfall maxima are not always in the right places, a seasonal evolution is evident in the simulation, as is the intraseasonal reversal of the rainfall anomalies in certain regions, mainly between South and Southeast Brazil.

In November of EN and LN year, when remote influences are larger, rainfall is well reproduced. During December and January rainfall is reasonably represented in the southern regions, but the tropical convection in northwestern SA is missing. In February again the rainfall is better represented, although here is underestimation in northern Brazil. Rainfall maxima are generally shifted southward.

The rainfall patterns are consistent with the divergence of the moisture flux. When the precipitation is underestimated, the model shows moisture divergence. There is a tendency for the model to direct moisture flux into southern SA during austral spring and late summer, while in January the moisture flux is directed towards Southeast Brazil.

The regional scale circulation features generally observed in November and January of EN and LN years in Southeast Brazil are visible.

The sign of the anomalies (differences between the rainfall patterns during EN and LN episodes) is consistent with observations. Furthermore, the reversal of the sign of the anomalies between November and January is captured.

The reversal of the temperature anomalies is well represented for November and January, though overestimated in northwestern South America.

ConclusionsSHORT RUNS & SENSITIVITY TO CUMULUS CONVECTION

SCHEMES

It is worth mentioning that there are differences between the CRU data set and the station data from Brazil.

There are not great differences between the rainfall patterns obtained for different starting dates.

The cumulus convection scheme that best represents the entire observed rainfall field is Grell with Arakawa-Schubert closure assumption, because it includes some amount of tropical rainfall, though underestimated. On the other hand, the Kuo scheme better represents the subtropical precipitation, but does not reproduce any tropical precipitation.