An Accuracy Accessing Approach for Chlorophyll-a Concentration Retrieval Results

In this study, a MDC3A algorithm (Multi-Data Crossing Algorithm for Accuracy Accessing) was developed for accessing the accuracy of chl-a (chlorophyll-a) retrieval model in case of no sufficient available in situ measurements. Three simple estimation algorithms of chl-a concentration, i.e., two-band algorithm, three-band algorithm and four-band algorithm, were used as input dataset of MDC3A algorithm to illuminate its performance. These three simple algorithms were calibrated and validated by calibration and validation dataset collected on October 27–28, 2003. According to model calibration and validation results, it was found that the four-band algorithm (R2 = 0.8676) had a superior performance to the two-band (R2 = 0.5061) and three-band (R2 = 0.5142) algorithm. The uncertainties in modeling prediction of these three simple algorithms were underestimated as 0.07 %, 0.04 % and 8.07 % for calibration dataset and 8.38 %, 9.33 % and 9.37 % for validation dataset by MDC3A algorithm through comparison with in situ measurements. Because the MDC3A algorithm was able to detect the random errors from measured values, but had an inadequate ability to detect systematical errors and gross errors from measured values. The uncertainty estimated by MDC3A algorithm was usually lower than that estimated by in situ measurements.