Rainfall Prediction for Northeast Brazil during MAM and AMJ 2003

Abstract

The forecasts for MAM and AMJ 2003 show that the Northeast Brazil region will experience rainfall above the normal, due to a southward migration of the ITCZ and he intensification of the meridional SST gradient in the tropical Atlantic. The fact that El Niño has weakened and almost dissapeared has also contributed to this situation. We analyze the fact that the updated forecasts for MAM and AMJ 2003 run with SST anomalies of February show above the normal rainfall, while the same seasons showed defficient rainfall when the model was run when SST anomalies of December 2003 or January 2003. CPTEC, IRI, ECMWF, UKMet Office, and Hadley Centre have been issuing at the beginning of 2003 forecassts that called for drought conditions on the region, and the changes in the oceanic and atmopsheric fields from January 2003 to February 2003 have motivated this reversal on the rainfall forecast for the region.

Introduction:

Monthly to seasonal dynamical atmospheric prediction at CPTEC have been performed since January 1995. The model used for these predictions is the CPTEC/COLA Atmospheric General Circulation Model AGCM (Cavalcanti et al. 2002, Marengo et al. 2003), which was derived from the NCEP model (Kinter et al. 1988) and includes the SSIB module (Xue et al. 1991). The horizontal resolution of the CPTEC/COLA AGCM is T62 and there are 28 levels in the vertical. The seasonal predictions at CPTEC, during the first 3 years, were results of an ensemble of 4 members, using 4 consecutive days as initial conditions. In 1998 there was an increase in the number of members, with the prediction based on 25 integrations. The application of boundary conditions also changed, from the use of persisted Sea Surface Temperature (SST) anomalies, to predicted SST. In the tropical Pacific the SST is given by NCEP coupled model forecast, and in the tropical Atlantic the SST is predicted using a statistical model (Pezzi et al. 1998). Outside these regions persisted SST anomalies are used.

Predictions for the rainy season of Nordeste (Northeast of Brazil) have been discussed in several papers, using various dynamical models and all them have shown the high predictability and model skill in depicting rainfall anomalies, especially during extreme years associated with El Niño (Nobre et al. 1995, Nobre and Cavalcanti 1996, Folland et al. 2002, Marengo et al. 2003). Similarly, statistical ad stochastic methods have been used in predicting rainfall anomalies since the late 1980s: A. Colman-UK Hadley Centre; S. Hastenrath-University of Wisconsin-Madison, USA; T. Xavier-FUNCEME/Universidade de Ceara-Brasil; SIMOC- CPTEC, Brazil. Precipitation anomalies in Northeast Brazil (area averaged 2⁰S-12⁰S; 45⁰W-35⁰W) from a simulation of 11 years (1986 to 1996), using T42L18 reproduces much of the interannual variability in this region, (Cavalcanti et al. 1998). The forecast of the March-May 1998, 1999 and 2000 rainy season in Northeast Brazil verified well. The maximum of rainfall in northern Northeast Brazil occurs during the March-June season, with the pre-rainy season detected in December-January. Thus, we discuss the MAM and AMJ seasonal rainfall forecasts.

The year 2002 featured the beginning of an El Niño year that wa considered as moderate, and that has reached its peak during December 2002 and January 2003. Seasonal rainfall forecasts run using December 2002 and January 2003 boundary conditions have produced extreme negative rainfall deperatures and even drought, amd CPTEC as well as other major climate centers issued forecasts in January and February 2003 showing drought risk in the region. At that time, during December 2002 and January 2003, the tropical Pacific was still warm and the tropical Atlantic was not favorable for above normal rainfall in Northeast Brazil. By February 2003, theSST field in the Equatorial Pacific started to show normality or even colder waters (Suggesting the end of El Nino), while the SST fields in the tropical Atlantic were favorable for rainy conditions in Northeast Brazil, contrasting with the de dry forecasts issued the previous month.

One of the most important factors that have determined the rate of decay of the current El Nino is the wind along the equator. While some anomalous westerly winds will probably continue near the dateline as a response to the above normal SST near and to the east of the dateline, the magnitude and location of zonal wind anomalies across the whole basin will determine the speed at which the event will dissipate. By now, (mid-March), there is an area of anomalously shallow thermocline perturbations in the western Pacific that have been moving eastward and working to change the weaken the El Niño.

Thus, we will show on this report the MAM 2003 forecasts (forced with January and February 2003 SST anomalies), that produced contrasting results: rainfall well below the normal in most of Northeast Brazil (also reported by dynamics models from IRI, ECMWF, UK Met Office and the statistical model from Hadley Centre), and above the normal when forced with February SST anomalies. On this report, we analyze both forecasts and discuss the differences.

Methodology:

The seasonal prediction at CPTEC is performed in two ways, one using persisted SST anomalies in all oceans and another using predicted SST in the Tropical Pacific and Atlantic Oceans and persisted SST anomalies in the other areas. The model integrations at this month were performed using 25 initial conditions for each set of different SST. The predictions for MAM and AMJ 2003 discussed here are the one forced with January SST anomalies (Fig.1 a, b), and the updated ones forced by February SST anomalies (Fig 1c, d).

Results:

The model shows in MAM 2003 (Fig. 1a, b) forced with January SST anomalies, in both cases, above normal precipitation over Northern South America and part of Northeastern South America, which comprises eastern Amazonia and part of Northeast Brazil. Normal to below normal conditions is predicted over the extreme northeast of Northeast Brazil, due to the displacement of the ITCZ northward. The above normal precipitation predicted by the model over parts of Northeast Brazil (off the coast of the Brazilian state or Maranhão) can be related to the positive SST anomalies predicted in the Atlantic Ocean close to the Northeast Brazil coast (Fig. 2a).

The updated forecast for MAM 2003 using the February 2003 SST anomalies show a reversal of rainfall (Fig. 1c, d) shows a reversal of the rainfall patterns shown in the MAM 2003 forecasts from Fig. 1 a, b. Northeast Brazil shows rainfall above the normal over the entire Northeast Brazil region as well as along the ITCZ band that is depicted by the model as located southward of its climatological position. The model forced with forecasted February 2003 SST anomalies (Fig 1c) produces abundant rainfall over the entire Northeast Brazil region, including the southern Northeast Brazil, while when forced with persisted SST anomalies, the more abundant rainfall is detected in the northern part of Northeast Brazil mainly. The main differences with the previous MAM 2003 forecast (Figs, 1a, b) is that the situation of the tropical Ocean Atlantic and Pacific changed from January 2003 to February 2

The precipitation prediction for AMJ 2003 with predicted and persisted February SST anomalies is shown in Fig. 4 a, b. Over the Northeast Brazil region and the tropical South Atlantic the model predicts normal to above normal precipitation, indicating the southward position of the ITCZ. The AMJ 2003 rainfall fields produced by the model run with predicted January 2003 SST anomalies (not shown) show a pattern that resembles the MAM 2003 rainfall anomalies from Fig. 1a, b, with deficient rainfall conditions in the whole Northeast Brazil region and Northward displaced ITCZ.

In MAM 2003, the semiarid region and north of Northeast Brazil were expected to be below average or normal, due to the possible northward displacement of the ITCZ and the El Niño effect. This behaviour was consistent with the tropical Atlantic SST anomalies which displayed a weak north/south SST gradient. The AMJ 2003 (updated) changed the situation, since the tropical Atlantic SST anomalies was stronger and favor a southward displacement of the ITCZ over Northeast Brazil and thus more rainfall on the region.

References:

Cavalcanti, I.F.A.; L.P.Pezzi, P. Nobre; G.Sampaio., H.Camargo Jr, 1998. Climate prediction of precipitation in Brasil for the Northeast Brazil rainy season (MAM) 1998. Experimental Long -Lead Forecast Bulletin, 7, No. 1; 24-27.

Cavalcanti IFA, Marengo JA, Satyamurty P, Trosnikov I, Bonatti JP, Nobre CA, D’Almeida C, Sampaio G, Cunningham CAC, Camargo H, Sanches MB 2002, Global climatological features in a simulation using CPTEC/COLA AGCM. Journal Climate, 15, 2965-2988.

Folland, C., Colman, A., Rowell, D., Davey, M., 2001, Predictability of Northeast Brazil rainfall and real-time forecast skill, 1987-98. J. Climate, 14, 1937-1958.

Kinter, J.L.;J.Shukla; L.Marx; E.K. Schneider, 1988. A simulation of winter and summer circulations with the NMC global spectral model. J.Atm.Sci., 45, 2486-2522.

Marengo, J., Cavalcanti, IFA, Satyamurty, P., Nobre, C. A., Bonatti, J. P., Manzi, A., Trosnikov, I. Sampaio, G., Camargo, H., Sanches, M. B., Cunningham, CAC, D’Almeida, C., Pezzi, L. P., 2003: Ensemble simulation of regional rainfall features in the CPTEC/COLA atmospheric GCM. Skill and Predictability assessment and applications to climate predictions (Accepted, Climate Dynamics)

Nobre, P.; I.F.A.Cavalcanti, 1996. Previsão Climática Sazonal no CPTEC-A estação chuvosa de 1995 e 1996 no Nordeste do Brasil. Congresso de Meteorologia Argentino (Congremet VII e Climet VII), Buenos Aires, 2-6 setembro, 1996, pp.351-352.

Nobre, P.; M.L.Abreu, I.F.A.Cavalcanti; M.Quadro; L.P.Pezzi, 1995. Climate ensemble forecasting at CPTEC. Proceedings of the twentieth annual climate diagnostics workshop., Seattle,Washington, Oct. 23-27, 1995. pp 417-420.

Pezzi, L.P.; C.A.Repelli; P.Nobre; I.F.A. Cavalcanti; G.Sampaio, 1998. Forecasts of Tropical Atlantic SST anomalies using a statistical Ocean Model at CPTEC/INPE Brazil. Experimental Long-Lead Forecast Bulletin, 7, n⁰ 1, 28-31.

Xue,Y.; P.Sellers.J.L. Kinter; J.Shukla, 1991. A simple biosphere model for global climate studies. J.Climate,4, 345-364.

(a)

(b)

(c)

(d)

Fig.1. Rainfall Anomalies Forecasts: a) MAM 2003 using Predicted SST Anomalies in Pacific (NCEP) and Atlantic (SIMOC/CPTEC); b) MAM 2003 using persisted January 2003 SST Anomalies, c) MAM 2003 using Predicted SST Anomalies in Pacific (NCEP) and Atlantic (SIMOC/CPTEC); d) MAM 2003 using persisted February 2003 SST anomalies.

(a)

(b)

Fig.2: SST anomaly of (a) January 2003 and (b) February 2003 used to force the CPTEC/COLA AGCM. The upper panels represent the persisted SST for January and February, while the lower panels (MAM and AMJ) represent the persisted SST fields in the tropical Pacific and Atlantic (and persisted anywhere else).

Fig. 3 shows the diagram of plumes for the forecast of rainfall in North Northeast Brazil for all the runs forced by persisted SST anomalies in January 2003 (a) and February 2003 (b). As discussed on the previous figures, the mean of the ensembles show rainfall below the normal when the model is forced with January 2003 SST anomalies, while the rainfall appears to be above the normal for all the season after March 2003 when the model was forced with February SST anomalies.

(a)

(b)

Fig. 3. Diagram of plumes for rainfall forecasts in Northeast Brazil combing both persisted and predicted SST anomalies during January 2003 (a) and February 2003 (b).

(a)

(b)

Fig. 4. Rainfall Anomaly Forecast for AMJ 2003 using (a) persisted) and (b) predicted SST Anomalies in Pacific (NCEP) and Atlantic (SIMOC/CPTEC) for February 2003.

Climate Diagnostic Meetings are held monthly at CPTEC, in order to analyse and discuss the monthly observed large scale atmospheric and oceanic conditions, and the model results. Considering the predictability of several regions of South America, the intensity of the SST forcing, and verifying the consistence with other GCMs, a seasonal prediction consensus is performed based on the models results. It is important to remember that the peak of the rainy season in Northeast Brazil occurs between March-May. During March 2003, the consensus prediction was performed to AMJ 2003 (Fig. 5b), while During February 2003 the consesnsus predictsion was performed for MAM 2003 (Fig. 5a).

(a)

(b)

Fig. 5- Precipitation consensus forecast for (a) MAM 2003 and (b) AMJ 2003-updated over Brazil issued by CPTEC|.