Pennucci, G; Fargion, G; Alvarez, A; Trees, C; Arnone, R (2012). A methodology for calibration of hyperspectral and multispectral satellite data in coastal areas. OCEAN SENSING AND MONITORING IV, 8372, 83720K.
The objective of this work is to determine the location(s) in any given oceanic area during different temporal periods where in situ sampling for Calibration/Validation (Cal/Val) provides the best capability to retrieve accurate radiometric and derived product data (lowest uncertainties). We present a method to merge satellite imagery with in situ measurements, to determine the best in situ sampling strategy suitable for satellite Cal/Val and to evaluate the present in situ locations through uncertainty indices. This analysis is required to determine if the present in situ sites are adequate for assessing uncertainty and where additional sites and ship programs should be located to improve Calibration/Validation (Cal/Val) procedures. Our methodology uses satellite acquisitions to build a covariance matrix encoding the spatial-temporal variability of the area of interest. The covariance matrix is used in a Bayesian framework to merge satellite and in situ data providing a product with lower uncertainty. The best in situ location for Cal/Val is then identified by using a design principle (A-optimum design) that looks for minimizing the estimated variance of the merged products. Satellite products investigated in this study include Ocean Color water leaving radiance, chlorophyll, and inherent and apparent optical properties (retrieved from MODIS and VIIRS). In situ measurements are obtained from systems operated on fixed deployment platforms (e. g., sites of the Ocean Color component of the AErosol RObotic NETwork-AERONET-OC), moorings (e. g, Marine Optical Buoy-MOBY), ships or autonomous vehicles (such as Autonomous Underwater Vehicles and/or Gliders).