Previous: Data and Methodology Next: Discussion and Summary Up: Ext. Abst.

3. Results

In our first phenomenological approach to detect the influence of the El Niño/La Niña events over the South America cloud cover the whole 1960 - 1980 period is analyzed. The PC based CCA analysis is implemented with the 5 more significative PC modes for each field. Results of this study can be seen in Figure 1 for the first tree modes. An important spatial correlation of the 80% was found between the third CCA expansion coefficient of the 1960-2000 and the El Niño 3 anomaly.

Previous studies (Ristori, Fochesatto et al., 2000) shows South America coherence regions for the Normality 1995-1996 period and for the last El Niño 1997-1998 period via SVD analysis. In this case we can see a sort of similarity between the first two CCA maps and the normality period maps and between the third CCA maps and the anomaly period map.

In a second approach, a clustering technique based in the correlation study for a given time frame identified by El Niño, Normality or La Niña, compare the SST CCA expansion coefficients calculated before and the South America TCC with results shown in Table 1.

CCA_5,n means the 'n' canonical correlation of the 5 PC modes retained in the CCA, r(a_k,b_k) in the temporal correlation between the most significative expansion coefficients, SCF is the squared covariance fraction between the true and the synthetic reconstructed covariance fraction for each mode and the CSCF is the correspondant cumulative covariance fraction taking into account the current as well as the previous calculated modes. Finally, for this table, the correspondent heterogeneous correlation maps are shown in figures 2 to 8.

After computing the large scale coupling, we proceed by a downscaling process to quantify the regional to local impact process between the most significant variables evolves in the air pollution process. For the large to regional scale studies (only for the 1996-1998 period in which local measurements had been made), scatter plots between the TCC and the Visible Radiation temporal series are shown with -52% for normality and -78% for anomaly periods (Figure 9).

Figure 10 shows the linear correlation between Visible Radiation an ABL height monthly mean with annual time lag.

In our first phenomenological approach to detect the influence of the El Niño/La Niña events over the South America cloud cover the whole 1960 - 1980 period was analyzed. Results of this study can be seen in Figure 1. An important spatial correlation of the 80% was found between the third CCA expansion coefficient of the 1960-2000 and the El Niño 3 anomaly.

Previous studies (Ristori, Fochesatto et al, 2000) shows South America coherence regions for the Normality 1995-1996 period and for the last El Niño 1997-1998 period via SVD analysis. In this case we can see a sort of similarity between the first two CCA maps and the normality period maps and between the third CCA maps and the anomaly period map.

In a second approach, a clustering technique based in the correlation study for a given timeframe identified by El Niño, Normality or La Niña, compare the SST CCA expansion coefficients calculated before and the South America TCC with results shown in Table 1. The correspondent heterogeneous correlation maps are shown in figures 2 to 8.

For the large to regional scale studies (only for the 1996-1998 period in which local measurements had been made), scatter plots between the TCC and the Visible Radiation temporal series is shown with -52% for normality and -78% for anomaly periods (Figure 9). Figure 10 shows the linear correlation between Visible Radiation an ABL height monthly mean with annual time lag.