12 Observing the Agulhas Current With Sea Surface Temperature . . .
239
number of valid observations over the Agulhas Current region. This can be explained
by differences in both sensors measurement capacity and cloud masking algorithms.
MODIS has a high spatial resolution and larger number of spectral bands compared
to AVHRR and is therefore able to better detect clouds during daytime (Heidinger
et al. 2002). In addition, the Pathfinder cloud mask algorithm relies on spatial uniformity test and background fields of reference SST (Hickox et al. 2000; Thomas et al.
2004). The Pathfinder cloud masking algorithm therefore tends to erroneously flag
pixels in regions of strong thermal gradients (Dufois et al. 2012). Coastal regions in
the southern Agulhas Current ecosystem often exhibit strong temperature gradients.
Coastal regions west of Cape Agulhas (at the southernmost tip of Africa) are considered to form part of the Benguela upwelling ecosystem (Hutchings et al. 2009).
West of Cape Agulhas, strong temperature fronts develop near the coast during the
austral summer due to the predominance of upwelling-favorable southerly winds.
The Agulhas Current also dynamically drives several coastal upwelling cells on the
eastern Agulhas Bank (Lutjeharms 2006). The Pathfinder dataset is particularly bad
at imaging the inshore front of the Agulhas Current and the coastal regions near
Cape Town during the summer upwelling season. The OSI-SAF SST dataset seems
to suffer from the same cloud masking limitations as the Pathfinder dataset. For the
MODIS data presented here, no reference SST was used when masking clouds.
Figure 12.2 shows that the percentage of valid IR SST observations in the Agulhas
Current region varies seasonally. North of the Retroflection, IR sensors provide a
better coverage during the austral winter months (from June to August). Further
south, at the Retroflection and within the Agulhas Return Current, the tendency
is reversed with a greater number of valid SST observations available during the
austral summer (December to February). Throughout the year, the high frequency
OSI-SAF SST dataset provides 10 to 20 times more observations than the MODIS
and Pathfinder datasets. Inadequate cloud masking algorithms can cause significant
bias in the climatology derived from IR SST datasets, with important consequences in
climatological or numerical modelling studies (Dufois et al. 2012). Figure 12.3 shows
that during the austral summer months, the MODIS and Pathfinder SST climatology
exhibits strong differences in the coastal regions of the Agulhas Current, near Port
Elizabeth and around the Cape Peninsula.
Microwave radiometers which are able to “see” through clouds, provide an alternative to IR-based sensors in regions of strong air/sea interactions. The TRMM
Microwave Imager (TMI), where TRMM stands for Tropical Rainfall Measuring
Mission, launched in November 1997, was the first microwave radiometer to provide accurate measurements of SST in the Agulhas Current region. The TMI SST
dataset was used successfully to monitor the Retroflection of the East Madagascar
Current and track perturbations in the southern Agulhas Current region (Quartly and
Srokosz 2002). Since 2002, the AMSR-E sensor (mentioned in Sect. 2) has provided
additional microwave observations of SST in the Agulhas Current region. While the
coverage from the TMI sensor is limited to regions north of 38
◦ S, that of the AMSR-E
dataset is global. The AMSR-E sensor is therefore well suited to the monitoring of
the Agulhas Retroflection. Microwave SST observations suffer from 2 main limitations: a low spatial resolution of about 50 km and an inability to measure SST closer
239
number of valid observations over the Agulhas Current region. This can be explained
by differences in both sensors measurement capacity and cloud masking algorithms.
MODIS has a high spatial resolution and larger number of spectral bands compared
to AVHRR and is therefore able to better detect clouds during daytime (Heidinger
et al. 2002). In addition, the Pathfinder cloud mask algorithm relies on spatial uniformity test and background fields of reference SST (Hickox et al. 2000; Thomas et al.
2004). The Pathfinder cloud masking algorithm therefore tends to erroneously flag
pixels in regions of strong thermal gradients (Dufois et al. 2012). Coastal regions in
the southern Agulhas Current ecosystem often exhibit strong temperature gradients.
Coastal regions west of Cape Agulhas (at the southernmost tip of Africa) are considered to form part of the Benguela upwelling ecosystem (Hutchings et al. 2009).
West of Cape Agulhas, strong temperature fronts develop near the coast during the
austral summer due to the predominance of upwelling-favorable southerly winds.
The Agulhas Current also dynamically drives several coastal upwelling cells on the
eastern Agulhas Bank (Lutjeharms 2006). The Pathfinder dataset is particularly bad
at imaging the inshore front of the Agulhas Current and the coastal regions near
Cape Town during the summer upwelling season. The OSI-SAF SST dataset seems
to suffer from the same cloud masking limitations as the Pathfinder dataset. For the
MODIS data presented here, no reference SST was used when masking clouds.
Figure 12.2 shows that the percentage of valid IR SST observations in the Agulhas
Current region varies seasonally. North of the Retroflection, IR sensors provide a
better coverage during the austral winter months (from June to August). Further
south, at the Retroflection and within the Agulhas Return Current, the tendency
is reversed with a greater number of valid SST observations available during the
austral summer (December to February). Throughout the year, the high frequency
OSI-SAF SST dataset provides 10 to 20 times more observations than the MODIS
and Pathfinder datasets. Inadequate cloud masking algorithms can cause significant
bias in the climatology derived from IR SST datasets, with important consequences in
climatological or numerical modelling studies (Dufois et al. 2012). Figure 12.3 shows
that during the austral summer months, the MODIS and Pathfinder SST climatology
exhibits strong differences in the coastal regions of the Agulhas Current, near Port
Elizabeth and around the Cape Peninsula.
Microwave radiometers which are able to “see” through clouds, provide an alternative to IR-based sensors in regions of strong air/sea interactions. The TRMM
Microwave Imager (TMI), where TRMM stands for Tropical Rainfall Measuring
Mission, launched in November 1997, was the first microwave radiometer to provide accurate measurements of SST in the Agulhas Current region. The TMI SST
dataset was used successfully to monitor the Retroflection of the East Madagascar
Current and track perturbations in the southern Agulhas Current region (Quartly and
Srokosz 2002). Since 2002, the AMSR-E sensor (mentioned in Sect. 2) has provided
additional microwave observations of SST in the Agulhas Current region. While the
coverage from the TMI sensor is limited to regions north of 38
◦ S, that of the AMSR-E
dataset is global. The AMSR-E sensor is therefore well suited to the monitoring of
the Agulhas Retroflection. Microwave SST observations suffer from 2 main limitations: a low spatial resolution of about 50 km and an inability to measure SST closer