Distinguishing between the SST-forced variability and internal variabilityin mid latitudes: Analysis of observations and GCM simulations

It is shown that the dominant structure of the seasonal-mean mid-latitude c irculation (500 hPa height) pattern over the Pacific-North America (PacNA) region forced by tropical sea surface temerature (SST)-related diabatic hea ting, is distinctly different from the seasonal-mean internal variability p attern that occurs in the absence of El Nino Southern Oscillation (ENSO) re lated SST anomalies. The separation of these two patterns is accomplished b y utilizing ensemble General Circulation Model (GCM) integrations in conjun ction with re-analyses. Ensemble simulations made with the GCM of the Center for Ocean-Land-Atmosph ere Studies (COLA) are compared to the re-analyses of the National Centers for Environmental Prediction (NCEP) for the 16 winters 1981/82-96/97. The G CM ensemble for each winter consists of 9 integrations initialized from ana lyses, and utilizing the observed time-varying SST. In addition, a 39-year simulation with the same GCM using climatological SSTs (with the observed a nnual cycle) is used. The hemispheric North Atlantic/Arctic pattern is the leading seasonal-mean empirical orthogonal function (EOF) for: (i) all GCM seasonal means in the ensemble simulations; (ii) the NCEP re-analyses; and (iii) the climatologic al SST integration. This mode is removed from these datasets. The total seasonal-mean variance of all GCM ensemble integrations is genera lly quite realistic in the PacNA region. The variance of GCM seasonal-mean deviations about the ensemble mean agrees with that in the climatological S ST GCM run, but is about 20% weaker than the observed variance for 29 non-E NSO winters from the NCEP re-analyses. The ratio of the SST-forced variance (obtained from the variance of ensemble means but corrected for the finite sample size) to the internal variance (from the deviations about the ensem ble mean) exceeds 2.5 in the eastern Pacific and 4.5 over Mexico. It is hig hly significant (99% confidence level) over most of the PacNA region. A number of techniques are used to calculate the patterns forced by SST hea ting and the internal variability patterns. The SST-forced mid-latitude cir culation pattern is calculated in seven ways, namely: (1) as the leading EO F of the ensemble-mean GCM height field for the 16 winters; (2) as the lead ing mode of a singular-value decomposition (SVD) analysis of height with tr opical diabatic heating from the GCM; (3) as the leading EOF las above) for NCEP re-analyses for the same 16 winters; (4) as the leading SVD mode las above) for the NCEP re-analyses for the same 16 winters; (5) as the leading EOF of height from re-analyses for the 10 winters having the five stronges t warm and five strongest cold tropical SST anomalies in the last 39 years; (6) as the leading SVD mode las above) from re-analyses for these same 10 extreme-SST winters; (7) as a regression of GCM-simulated height on a tropi cal SST time series obtained from the first EOF mode of re-analysis tropica l diabatic heating. It is found that the results of all of these techniques agree extremely well with each other, and that the leading modes in the EO F (SVD) analyses explain large amounts of variance (squared covariance), ab out 50% (90%). The spread of projections of individual seasonal means on the leading SVD m ode of the seasonal-ensemble means, is less than the variation of the proje ction of the ensemble means on the SVD mode (90% significance level). We draw two conclusions: first, that the GCM ensemble means simulate the ob served anomalies with high accuracy; and second, that the observed and simu lated anomalies are indeed forced by tropical diabatic heating. The internal variability pattern was calculated in three different ways: (1 ) as the leading EOF of height of the deviations of each seasonal mean abou t the corresponding ensemble mean for that winter; (2) as the leading heigh t EOF from the 39-year GCM integration forced by climatological (but annual ly varying) SST; and (3) as the leading height EOF from re-analyses for 29 winters not associated with very warm and cold tropical SSTs. The patterns derived from these analyses have a common structure. It is found that it is this internal variability pattern, and not the SST-forced pattern describe d above, that closely resembles the 'PNA' pattern of Wallace and Gutzler. The SST-forced pattern in the GCM (characterized by the heterogeneous corre lation pattern of the leading SVD mode for the ensemble means) is significa ntly different (95% confidence level) from the internal variability pattern (characterized by the homogeneous correlation pattern of the leading EOF o f the GCM deviations about the ensemble mean) over a region in western Nort h America and the adjoining eastern Pacific, a region north-east of the Gre at Lakes, and a small region in the Gulf of Alaska.