A Coastal Seawater Temperature Dataset for Biogeographical Studies: Large Biases between In Situ and Remotely-Sensed Data Sets around the Coast of South Africa

Abstract

Gridded SST products developed particularly for offshore regions are increasingly being applied close to the coast for biogeographical applications. The purpose of this paper is to demonstrate the dangers of doing so through a comparison of reprocessed MODIS Terra and Pathfinder v5.2 SSTs, both at 4 km resolution, with instrumental in situ temperatures taken within 400 m from the coast. We report large biases of up to +6°C in places between satellite-derived and in situ climatological temperatures for 87 sites spanning the entire ca. 2 700 km of the South African coastline. Although biases are predominantly warm (i.e. the satellite SSTs being higher), smaller or even cold biases also appear in places, especially along the southern and western coasts of the country. We also demonstrate the presence of gradients in temperature biases along shore-normal transects — generally SSTs extracted close to the shore demonstrate a smaller bias with respect to the in situ temperatures. Contributing towards the magnitude of the biases are factors such as SST data source, proximity to the shore, the presence/absence of upwelling cells or coastal embayments. Despite the generally large biases, from a biogeographical perspective, species distribution retains a correlative relationship with underlying spatial patterns in SST, but in order to arrive at a causal understanding of the determinants of biogeographical patterns we suggest that in shallow, inshore marine habitats, temperature is best measured directly.

Assessing the Use of Area- and Time-Averaging Based on Known De-correlation Scales to Provide Satellite Derived Sea Surface Temperatures in Coastal Areas

“…In addition, previous studies have demonstrated biases in SST data when comparing it to in situ measurements within coastal areas (Smale and Wernberg, 2009;Lathlean et al, 2011;Stobart et al, 2016), with some studies recording up to 6 • C differences (Smit et al, 2013). These biases are due to contamination of the satellite processing due to the presence of coastal features such as the shoreline, estuaries and embayments as well as complex coastal dynamics such as tides and upwelling (Smit et al, 2013) that vary over short spatial scales. Despite this well documented bias of satellite SST data in coastal areas, no studies have assessed if using spatial area averaging (based on known decorrelation length and/or time scales) can reduce the bias or increase the availability of satellite derived temperature data.…”

“…SST, however, may not be available for coastal areas due to contamination in the satellite-derived images [noisier radar returns from land and sea (Brooks et al, 1990) and improper instrument corrections (Shum et al, 1998)]. In addition, previous studies have demonstrated biases in SST data when comparing it to in situ measurements within coastal areas (Smale and Wernberg, 2009;Lathlean et al, 2011;Stobart et al, 2016), with some studies recording up to 6 • C differences (Smit et al, 2013). These biases are due to contamination of the satellite processing due to the presence of coastal features such as the shoreline, estuaries and embayments as well as complex coastal dynamics such as tides and upwelling (Smit et al, 2013) that vary over short spatial scales.…”

“…Lathlean et al, 2011), or an area averaged value whereby the mean SST measurements is calculated over a set number of pixels, including pixels offshore (Smit et al, 2013;Delgado et al, 2014;Stobart et al, 2016). Stobart et al (2016) assessed the daily temperature differences between satellite-derived and in situ measurements and averaged all the satellite-derived SST measurements within a 20 km 2 box centered on the in situ logger locations.…”

“…Stobart et al (2016) assessed the daily temperature differences between satellite-derived and in situ measurements and averaged all the satellite-derived SST measurements within a 20 km 2 box centered on the in situ logger locations. While Smit et al (2013) used an average over the adjacent nine SST data points and Delgado et al (2014) used 3 × 2 and 3 × 3 satellite pixels to compare with coastal and inner-shelf in situ measurements, respectively. For the latter study this area averaging, as well as incorporating time-scales over which the satellites passed the area, produced high correlations (R 2 of 0.99) between the two datasets.…”

Comparison of Split Window Algorithms for Retrieving Measurements of Sea Surface Temperature from MODIS Data in Near-Land Coastal Waters

“…Moreover, the differences between the in situ and satellite measurements of coastal sea temperature have also been evaluated by other authors [13][14][15], where this error can be as large as 6 • C [15]. All of these papers specifically analyzed SST data retrieved from advanced very high-resolution radiometer (AVHRR) and/or moderate resolution imaging spectroradiometer The study area comprised three sections of Italian coastal waters (Figure 1a): the Manfredonia Gulf, Taranto Gulf, and the area close to Lesina Lagoon [42].…”