2612 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 49, NO. 7, JULY 2011 Resolving the Subscale Spatial Variability of Apparent and Inherent Optical Properties in Ocean Color Match-Up Sites Mhd. Suhyb Salama and Zhongbo Su Abstract—A stochastic approach is developed to resolve the scale variability between point and aerospace measurements in ocean color match-up sites. The model used the differences be- tween in situ and aerospace-observed spectra and ocean color model inversion to estimate the subscale variability of apparent and inherent optical properties (IOPs). The model was tested and validated against three sets of ocean color data: simulated, in situ measured, and satellite data sets. The results showed that the variability of chlorophyll-a absorption was derived with high ac- curacy. Errors in derived subscale variability of detritus–gelbstoff absorption and particle scattering were larger than those of chlorophyll-a. The subscale radiometric variability was found to be proportional to that of IOPs and decreased with increasing wa- ter turbidity. The subpixel variability of reduced resolution ocean color image was derived with less than 12% of relative errors in clear and moderate turbid waters. Larger errors were obtained in estuarine turbid waters. Better accuracy was obtained for match-up sites with high internal contrast, i.e., spatial variability. Index Terms—Match-up site, ocean color, radiometric variabil- ity, subscale variability, variability of inherent optical properties (IOPs). I. I NTRODUCTION F IELD measurements are crucial for calibrating and validat- ing remote sensing data and derived water color products. Although high accuracy in both aerospace and in situ measure- ments can be realized, direct matching between aerospace and just above the water field radiances is quite challenging. This is due to the dynamic nature of water, correction schemes, and to the different volumes being sampled. Concurrent measurements with sensor overpass may eliminate a considerable range of the dynamic variability. Accurate correction for atmospheric and adjacency effects can be realized through atmospheric radiative transfer models and extensive in situ measurements. The scale difference between remotely and in situ sampled water volumes is, however, an inherent feature to the matching process unless properly treated. This is of quite importance in productive waters and with the introduction of the new network, i.e., Aerosol Robotic Network-Ocean Color [1], to support satellite ocean color validation activities. Manuscript received April 3, 2009; revised October 12, 2010; accepted December 15, 2010. Date of publication February 24, 2011; date of current version June 24, 2011. The authors are with the Faculty of Geo-Information Science and Earth Observation (ITC), University of Twente, 7500 Enschede, The Netherlands (e-mail: salama@itc.nl; b_su@itc.nl). Digital Object Identifier 10.1109/TGRS.2011.2104966 The nominal pixel size of current ocean color sensors varies approximately from 300 m for the Medium Resolution Imag- ing Spectrometer (MERIS) to 1000 m for Sea Wide Field- of-view Sensor (SeaWiFS) and Moderate Resolution Imaging Spectroradiometer (MODIS). The scale difference between in situ observation and a pixel of ocean color satellite is at least three to four orders of magnitude for nadir match-up sites and much larger for off-nadir ones. This huge scale difference means that point measurement is sampling a tiny fraction of the water body which is observed by a satellite pixel. Few studies were carried out to address the scale difference between point and aerospace measurements directly. Most of these studies have used resampling to smooth out the scale differences in the match-up sites. Hu et al. [2] suggested the data aggregation of the satellite ocean color data to meet the desired accuracy. Bissett et al. [3] suggested that spatial variations in ocean color depended on the number of channels used to described differences between homogenous regions. Harding et al. [4] selected the match-up points such that they satisfied a homo- geneity criterion. Bailey and Werdell [5] calculated the number of spatially homogeneous pixels surrounding the match-up point which should be averaged before matching. Hyde et al. [6] applied a correction algorithm to SeaWiFS products of chlorophyll-a to overcome the mismatch which was partially due to sampling size differences. Blackwell et al. [7] identified the spatial variabilities of marine biogeophysical quantities and their scales. They analyzed the spatial variability of marine geo- biochemical properties with respect to (w.r.t.) offshore distance and depth using in situ measurements. While the assumption of spatial homogeneity may result in good matches in case-1 water [8]–[11], it lowers the percentage of usable match-up points considerably [12], [13] and should be avoided for productive waters. Due to the high variability of productive waters, direct matching may result in large discrep- ancy between in situ and satellite measurements [4], [14], [15]. Moreover, any spatial aggregation/averaging of pixels increases the differences between the sampled volumes and ignores the prominent geophysical variability in inland and coastal waters. The deviations between in situ measurements and aerospace observations provide valuable information on the scale differ- ences between them. Salama [16] used the residuals between model and measured spectra to establish a confidence interval around the retrieved inherent optical properties (IOPs) and separate their uncertainty sources using the nonlinear regression technique of Bates and Watts [17]. Maritorena and Siegel [18] 0196-2892/$26.00 © 2011 IEEE