Estimation of correction factor for surface water reflection under lab and field conditions

Surface water reflection combines with upwelling water leaving radiance that carries key information on optically active constituents (OAC) of water body and introduces error in measured water reflectance. Hence, for the accurate information on OAC, it is necessary to either remove the surface reflected radiance directly above the water surface or apply a correction factor to upwelling water radiance from above-water in-situ radiometric measurements. Correction factor for a given set of environmental conditions is not always known. Either an average or a range of correction factor is available, which leads to significant errors in estimating water reflectance. Hence, the aim of this study was to develop an approach for removal of surface water reflection and to study the effect of wavelength, sun altitude, diffuse light condition, and wind speed on correction factor under both lab and field conditions. Specific objectives of this study were to: (a) develop a methodology for removal of surface reflections directly above the water surface under lab and field conditions; (b) build an artificial neural network (ANN) model to estimate a correction factor; and (c) evaluate the influence of sun altitude, diffuse light condition, wind speed, and wavelength on the correction factor with the model.;Similarly, the results obtained from field experiments conducted at Beaver Lake in Arkansas on four different days indicated that average surface reflection components contributed 8-51% to total upwelling radiance at nadir and introduced an error of 5 to 65% in water reflectance measurements for the visible and near infrared wavelengths. The model was developed with diffuse light condition, sun altitude, wind speed and wavelength as input variable and correction factor as the output under field conditions. The R2, RNS 2, and RMSE of the model developed in field condition were found to be 0.95, 0.95, and 0.0085, respectively for testing of the model. Sun altitude and diffuse light were the two major factor affecting correction factor at any VIS-NIR wavelength. The model developed might be used for estimating correction factor in various geographical locations within the calibration range of input data.;Results obtained from the tank experiments on six different days showed that surface reflections contributed 5-66% to total upwelling water radiance at nadir and introduced an error of 20 to 54% in measured water reflectance. The coefficient of variation for the correction factor was estimated as 0.92, which indicated that correction factor was highly variable. The model was developed with diffuse light condition, sun altitude, wind speed and wavelength as input variable and correction factor as the output under lab conditions. The R2, RNS2, and RMSE were found to be 0.96, 0.96, and 0.0032, respectively for testing of the model on different set of data. Sun attitude and wind speed were the two major factors affecting correction factor at any VIS-NIR wavelength.