**4. SVD analysis**

#### **4.1 Inter-annual anomalies**

To gain further insight into the relationship between SLA and CHL anomalies at inter- and intra-annual time scales, we use SVD analysis. Figure 5 shows the spatial patterns for the first mode, which explains 41% of the covariance between the SLA and CHL inter-annual anomalies. Both patterns are scaled to represent the amplitude of SLA and CHL anomalies associated with one standard deviation of the corresponding expansion coefficients. Large

#### Using SVD Analysis of Combined Altimetry and Ocean Color Satellite Data for Assessing Basin Scale Physical-Biological Coupling in the Mediterranean Sea 131

130 Remote Sensing of Biomass – Principles and Applications

Fig. 3. Correlation coefficient between SLA and CHL anomalies at inter-annual time scales. The correlation in the red and blue areas is statistically significant at 95% level or more.

Fig. 4. Correlation coefficient between SLA and CHL anomalies at intra-annual time scales. The correlation in the red and blue areas is statistically significant at 95% level or more.

To gain further insight into the relationship between SLA and CHL anomalies at inter- and intra-annual time scales, we use SVD analysis. Figure 5 shows the spatial patterns for the first mode, which explains 41% of the covariance between the SLA and CHL inter-annual anomalies. Both patterns are scaled to represent the amplitude of SLA and CHL anomalies associated with one standard deviation of the corresponding expansion coefficients. Large

**4. SVD analysis** 

**4.1 Inter-annual anomalies** 

areas of the Mediterranean Sea present the negative covariations (i.e. SLA increases and CHL decreases, or vice versa) typical of temperate areas, particularly in the Tyrrhenian, Ionian and Aegean basins. In the Levantine basin and in the southern part of the Western basin, areas with positive covariations are observed. These regions are characterized by high levels of mesoscale eddy variability (Pujol & Larnicol, 2005). Mesoscale eddies have important biological and biogeochemical consequences, driving vertical motions of water and lifting subsurface nutrients into the surface euphotic layer (McGillicuddy et al., 1998; Oschlies & Garçon, 1998). Positive covariations are also observed in the Adriatic basin, which is driven by the local winter climatic conditions (Santoleri et al., 2003), and in coastal areas such as Rhone, Ebro, Po and Nile river deltas, suggesting the influence of riverine inputs.

Fig. 5. First spatial patterns of (a) SLA and (b) CHL anomalies at inter-annual time scales. The patterns are scaled to represent the amplitude of SLA and CHL anomalies associated with 1 standard deviation of the first expansion coefficients.

Using SVD Analysis of Combined Altimetry and Ocean Color

major rivers.

Satellite Data for Assessing Basin Scale Physical-Biological Coupling in the Mediterranean Sea 133

annual variability is dominated by changes in vertical nutrient fluxes, that are regulated by the seasonal themocline dynamics. Mesosecale eddies plays also an important role, especially in the Levantine basin and the southern area of the Western basin. The Adriatic basin behaves completely different as its response is regulated by climatic conditions. Finally, riverine inputs influence the biological response in areas under the influence of

Fig. 7. Time evolution of normalized (a) first and (b) second expansions coefficients of SLA

(blue line) and CHL (red line) inter-annual anomalies.

The spatial pattern for the second mode is shown in Figure 6, accounting for 28% of the covariance between the SLA and CHL anomalies at inter-annual time scales. The areas with negative covariations dominate in the Levantine and Aegean basins. Negative covariations are also observed in the Ionian basin, with the exception of the Gulf of Gades, although the northern and southern parts behave in opposite ways (i.e. SLA increases in the north and decreases in the south, both negatively correlated with CHL). This different behavior occurs also in the Western basin, between the eastern and western parts. The Adriatic basin and the coastal areas covary positively, as observed in the first spatial pattern.

Fig. 6. Second spatial patterns of (a) SLA and (b) CHL anomalies at inter-annual time scales. The patterns are scaled to represent the amplitude of SLA and CHL anomalies associated with 1 standard deviation of the second expansion coefficients.

The time evolution of the first two modes of the SLA and CHL inter-annual anomalies is shown in Figure 7. The expansion coefficients for the first and second modes are correlated at 0.79 and 0.84 (significant at 99% level), respectively. These high values indicate that inter-

The spatial pattern for the second mode is shown in Figure 6, accounting for 28% of the covariance between the SLA and CHL anomalies at inter-annual time scales. The areas with negative covariations dominate in the Levantine and Aegean basins. Negative covariations are also observed in the Ionian basin, with the exception of the Gulf of Gades, although the northern and southern parts behave in opposite ways (i.e. SLA increases in the north and decreases in the south, both negatively correlated with CHL). This different behavior occurs also in the Western basin, between the eastern and western parts. The Adriatic basin and the

Fig. 6. Second spatial patterns of (a) SLA and (b) CHL anomalies at inter-annual time scales. The patterns are scaled to represent the amplitude of SLA and CHL anomalies associated

The time evolution of the first two modes of the SLA and CHL inter-annual anomalies is shown in Figure 7. The expansion coefficients for the first and second modes are correlated at 0.79 and 0.84 (significant at 99% level), respectively. These high values indicate that inter-

with 1 standard deviation of the second expansion coefficients.

coastal areas covary positively, as observed in the first spatial pattern.

annual variability is dominated by changes in vertical nutrient fluxes, that are regulated by the seasonal themocline dynamics. Mesosecale eddies plays also an important role, especially in the Levantine basin and the southern area of the Western basin. The Adriatic basin behaves completely different as its response is regulated by climatic conditions. Finally, riverine inputs influence the biological response in areas under the influence of major rivers.

Fig. 7. Time evolution of normalized (a) first and (b) second expansions coefficients of SLA (blue line) and CHL (red line) inter-annual anomalies.

Using SVD Analysis of Combined Altimetry and Ocean Color

shown) explains less than 10% of the covariance.

(red line) intra-annual anomalies.

**5. Conclusion** 

which are not coupled with the SLA should be important.

Satellite Data for Assessing Basin Scale Physical-Biological Coupling in the Mediterranean Sea 135

biomass. Alternatively, the shoaling of the thermocline could extend the deeper nutrientrich layer into the euphotic surface zone, allowing phytoplankton uptake. These compression mechanisms of the ocean surface layer would be similar dynamically to the compression caused by Rossby waves (Cipollini et al. 2001). This first pattern accounts for 69% of the total covariance at intra-annual time scales, whereas the second pattern (not

Figure 9 shows the time evolution of the first mode of the SLA and CHL intra-annual anomalies. The figure only shows the period from 2002 to 2006 to facilitate the observation of short-term variability. The correlation between the first expansion coefficients for SLA and CHL intra-annual anomalies is 0.40 (significant at 99% level). Although the compression may play an important role on the enhancement of CHL observed by the satellite, the relatively modest value of correlation indicates that processes other than mechanical accumulation take place. For example, biological processes related to food web dynamics

Fig. 9. Time evolution of normalized first expansions coefficients of SLA (blue line) and CHL

This study analyzes the basin scale physical-biological coupling in the Mediterranean Sea at inter-annual, seasonal and intra-annual time scales based on 12 years of concurrent satellite SLA and CHL data. Not surprisingly, the long-term (inter-annual and seasonal) variability of SLA and CHL is negatively correlated in most oceanic areas of the Mediterranean Sea. This is the typical behavior of temperate regions associated with the availability of nutrients in the mixed layer: summer stratification blocks upward entrainment of nutrients from deep layers and winter mixing brings nutrients to the surface (Cushing, 1959). In the coastal regions, particularly in those areas influenced by major rivers, the biological response is controlled by supply of nutrients of continental origin. However, other biological responses

Fig. 8. First spatial patterns of (a) SLA and (b) CHL anomalies at intra-annual time scales. The patterns are scaled to represent the amplitude of SLA and CHL anomalies associated with 1 standard deviation of the first expansion coefficients.

### **4.2 Intra-annual anomalies**

The spatial pattern for the first mode between the SLA and CHL intra-annual anomalies shows a significant positive covariation of SLA and CHL in the whole Mediterranean Sea. Note that variations in CHL at this time scale are markedly lower than in other modes. This behavior does not agree with that observed at inter-annual time scales, when the winter mixing enhances the upward transport of nutrients to the ocean surface. Therefore, other physical processes must be regarded in order to explain the biomass enhancement at this time scale. One candidate could be a mechanical effect indebted to oscillations in the thermocline. The modification of the thermocline depth due to the SLA variation could accumulate phytoplankton biomass close to the surface (or vice versa) and thus affect the CHL measured by the ocean color satellites without modifying the vertically-integrated biomass. Alternatively, the shoaling of the thermocline could extend the deeper nutrientrich layer into the euphotic surface zone, allowing phytoplankton uptake. These compression mechanisms of the ocean surface layer would be similar dynamically to the compression caused by Rossby waves (Cipollini et al. 2001). This first pattern accounts for 69% of the total covariance at intra-annual time scales, whereas the second pattern (not shown) explains less than 10% of the covariance.

Figure 9 shows the time evolution of the first mode of the SLA and CHL intra-annual anomalies. The figure only shows the period from 2002 to 2006 to facilitate the observation of short-term variability. The correlation between the first expansion coefficients for SLA and CHL intra-annual anomalies is 0.40 (significant at 99% level). Although the compression may play an important role on the enhancement of CHL observed by the satellite, the relatively modest value of correlation indicates that processes other than mechanical accumulation take place. For example, biological processes related to food web dynamics which are not coupled with the SLA should be important.

Fig. 9. Time evolution of normalized first expansions coefficients of SLA (blue line) and CHL (red line) intra-annual anomalies.
