MONTHLY AVERAGES OF SEA SURFACE TEMPERATURE OVER TUSCANY SEA

This paper concerns with the time variation of measurements of the mean Sea Surface Temperature on a regional scale by using AVHRR data. The studied algorithm is also presented with reference to the sub-pixel resolution, the precise geolocation and the cloud pixel location, which is based on temperature maps themselves instead on average temperature values of the Mediterranean.
Special calculations have been made for coast line pixels. The considered zone is the Tuscan Archipelago.
The detection of spatial/temporal Sea Surface Temperature (SST) variation plays a very important role in the influence of climatic behaviour [1].
This work reports a developed procedure for determining the SST monthly mean and its temporal evolution on a particular sea area. Different zones of the area are also considered. The region taken into account is the sea between Tuscany and the coast of Corsica, including the various islands of the Tuscan Archipelago, whose limits are defined by four values: lat1 = 42° 12.000’N to lat2 = 44° 30.000’N and lon1 = 09° 12.000’ to lon4 = 12° 24.000’E. The period taken into account is from July 2001 to September 2008.
NOAA – AVHRR data are directly received by the Satellite Receiving Station at University of Florence (University Pole of Prato, 43° 53.134’ N - 11° 05.942’ E), and elaborated with an owner suite software called ‘AMEDIT’. An example of daily SST can be viewed at URL (http://maresat.ing.unifi.it/).
This activity has been carried out in collaboration with the Sea Department of ARPAT (Regional Agency for the Environmental Protection in Tuscany), which has furnished in sito measurements. The SST values obtained from satellite data processing and in sito measurements are evaluated to characterize developed algorithms.
An operational technique has been developed to obtain spatial and temporal behaviour of SST monthly means. This technique includes improved elaborations by means of original algorithms, differently from those available in literature. Among these elaborations are: precise geolocations of the observed pixels (sub pixel resolution) with minimization of the errors connected with orbital parameters and the satellite’s attitude (Yaw, Pitch, Roll) [2]; determination of the cloud pixels based, in particular, on the average temperature detected by the obtained maps and segmented interpolation of the pixels near the coast [3].
The elaboration (? Operational technique) is composed of the following steps:
• extraction of the satellite received data for the geographical area of interest (seas of Tuscany);
• data geolocation through navigation of the satellite position based on orbital parameters and scanning geometry [4];
• segmented interpolation for improving the SST determination near the coast [3];
• radiometric calibration of raw data (count);
• measurements of SST using a land/sea mask (no land temperature);
• cloud filtering;
• average computation for each pixel of the different SST maps during each month;
• representation of data in a Mercator projection using WGS-84 Datum;
• analysis for different zones of the explored area. Elliptic and rectangular areas have been chosen for the considered geographic format, so as to obtain different results for the various zones of interest. Elliptic zones have been considered near the coast and rectangular offshore. The elliptic zones cover areas of approximately 102 (nM2), the rectangular zones have a surface of about 543 (nM2).
The SST monthly mean was produced pixel by pixel from a set of daily SST (images not affected by clouds) for the sea. The algorithm of cloud detection flags cloud pixels according to two criteria: the pixel value is 10°K below the reference SST monthly mean (SSTm) value and the temperature gradient vectors are larger than a threshold. In the optimum procedure the reference value, SSTm, is obtained from the same image.
The analyses of the SST monthly mean regarding different zones of Tuscany Archipelago show that:
a) from the mouth of Magra up to Livorno the temperatures were warmer (up to more than 2 °C) than the average of the total Tuscan sea format or of the other costal zones between Piombino and the south of the Argentario ;
b) for the zones offshore there are substantial differences (up to of 2 °C) mainly in the months of September and October. An example is the open sea between Corsica and Cinque Terre and North region and South region of Elba.

[1] J. STEELE, S. THORPE, AND K. TUREKIAN, Encyclopaedia Of Ocean Sciences, London (Academic Press), 2001.
[2] P.F. PELLEGRINI, M. TOMMASINI, C. FRANCINI, M. INNOCENTI, M. MARCONI, AND G. POLI, Automatic navigation of AVHRR and SeaWiFS imagery on the sea with segmented interpolation, Final Report – Novel Methodologies for thee Integration, Processing, and Analysis of Data From Spaceborne Sensor for the Monitoring of the Hydrosphere, Rainfall Phenomena, and the Ground, Ed. Enzo Dalle Mese, Università di Pisa, Pisa, Italy, pp.70-92, 2002.
[3] M. TOMMASINI, P.F. PELLEGRINI, AND G. POLI, Segmented Interpolation Along the Coastline for AVHRR NOAA Images, EARSeL eProceedings 4, pp.26-43, 1/2005.
[4] G.W. ROSBOROUGH, D.G. BALDWIN, AND W.J. EMERY, Precise AVHRR image navigation, IEEE Transactions on Geosciences and Remote Sensing, 32, pp.644-657, 1994.