In-pixel variations of chl a fluorescence in the Northern Gulf of Mexico and their implications for calibrating remotely sensed chl a and other products

Remote sensing instruments such as SeaWiFS and MODIS are often calibrated or tested by comparison with in situ data. These comparisons are based on the premise that there is uniform in-pixel variation of the standard deviation of in situ properties. To evaluate some of the errors resulting from this assumption, we conducted an analysis of in-pixel variation of chl a fluorescence by examining fluorescence data from a flow-through system on an underway vessel mapping surface properties for 10 days on the continental shelf in the vicinity of the Mississippi River delta. Significant variations of in-pixel standard deviation of chl a fluorescence were observed, which indicate that this uniformity assumption is not valid in the Northern Gulf of Mexico. Furthermore, our analysis indicates that a large apparent error by the remote sensor is generated if uniformity of standard deviation is assumed. Our results suggest that one should take into account both in-pixel mean and standard deviation when comparing remotely sensed ocean color data with in situ measurements. Similar measures should be taken when comparing other remotely sensed products with field determinations.