Previous: Introduction Next: Results Up: Ext. Abst.
2. Data and Methodology
The El Niño region SST and South America TCC data for the large scale-to-regional study comes from the NCEP NCAR CDAS Reanalysis database. Visible integrated incoming radiation temporal series from 0800 ST to 1200 ST and the Atmospheric Boundary Layer (ABL) structural parameter temporal series evolution used in the regional to local scale analysis comes from the CEILAP (CITEFA-CONICET) center placed in Villa Martelli, Buenos Aires, Argentina.
The methodology involved in the large scale teleconnection process between the El Niño 3 SST region and South America TCC is a Principal Components (PC) based Canonical Correlation Analysis (CCA) as presented in Bretheton et al., Barnet and Preisendorfer with further analysis over SST studies in Wallace et al.
In this case only a small but representative number of PC modes are retained to preserve the algorithm stability. CCA technique uses an intrinsic normalization of the cross corvariance matrix of the studied fields in the SVD analysis that translates the othonormality from the spatial to temporal field. In few words this means that the expansion coefficients (temporal series) of each mode of each field has the following properties: If ak(t) is the expansion coefficient of the kth TCC pattern and bk(t) is the expansion coefficient of the kth pattern, then the CCA technique assures that:
< an(t),ak(t) > = nk, < bn(t),bk(t) > = nk, < an(t),bk(t) > = cn.nk (were cn is a nonnegative value)
This choice was taken to achieve the geophysical goal of isolating the most significant Center Pacific SST natural oscillation from the anomaly induced by the El Niño Process an then, via the Heterogeneous correlation maps, detect coherent regions between this fields. CCA expansion coefficients, patterns, weight vectors and correlation maps were found for the 1960 - 2000 TCC and SST time series. Then the most important SST CCA expansion coefficients were correlated with the TCC to detect geographical coherence regions. This last correlation was implemented both with the total time series and for specific Niño, Niña and Normality periods. For this specific periods it was employed a 2 years timeframe.
CCA expansion coefficients, patterns, weight vectors and correlation maps were found for the 1960 - 2000 TCC and SST time series. Then the most important SST CCA expansion coefficients were correlated with the TCC to detect geographical coherence regions.
This last correlation was implemented both with the total time series and for specific Niño, Niña and Normality periods.
After demonstrating the impact over the South America TCC, 1996-1998 visible incoming radiation and ABL height time series were used to analyze the influence of this large scale variable over local field variables during the last Normality and El Niño events. In this case, correlation analysis was the selected tool to evidence the coupling between this set of variables.