Determining precision of aquatic turbidity measurement by NOAA-N series AVHRR

Abstract

Previous studies using National Oceanic and Atmospheric Administration's (NOAA) Advanced Very High Resolution Radiometer (AVHRR) visible and near-infrared data to map aquatic urbidity have used surface validation data from one or a very few surveys, but have not tested their relationships with long term data sets. Long term variance of one-time surface validation data is of concern because the AVHRR visible and near-infrared sensors have been shown to degrade in sensitivity through time. The lack of onboard calibration has been addressed by the remote sensing community by estimating calibration coefficients using desert targets with well-understood albedo characteristics. This study uses ten NOAA AVHRR images, recorded from 1985-1998 using five NOAA-N satellites, paired with Environment Canada Secchi transparency and beam transmissometer data from Lake Erie, to test the stability and precision of AVHRR data, corrected using these postlaunch calibration coefficients, in the measurement of turbidity. Shipboard data collected on Lake Erie in September, 1997 is used to develop regressions between AVHRR-derived albedo and Secehi transparency and beam transmissometer data. For the whole data set, AVHRR Channel 1 albedo predicts the natural logarithm of Secchi with a standard error of 0.40-0.44, 10% of the range in Lake Erie, and predicts transmission measured with a beam transmissometer with a standard error of 8-12%, or about 9% of the range. Where data associated with individual images deviated significantly from the general regression as a group, they were found to be associated with solar zenith angles in excess of 70$\sp\circ$ or satellite nadir angles in excess of 40$\sp\circ.$