Evaluation of satellite retrievals of water quality parameters for Lake Victoria in East Africa

Abstract

Lake Victoria in East Africa is a major ecosystem, whose size and importance has warranted the exploration of MODIS imagery to provide continuous and accurate water quality information. To this effect, two sea expeditions (in November 2014 and February 2015) were carried out to collect in situ lake surface temperature (LST), chlorophyll a (Chl) and Secchi disk depth (SDD) to compare with the corresponding satellite derived variables. Comparisons were made based on the following error matrices: coefficient of correlation (r), mean square error (MSE), root mean square error (RMSE), mean absolute deviation (MAD) and mean absolute percent error (MAPE). In general the results showed that satellite derived LST gives a good estimate of in situ LST. Conversely, the error matrices indicated that satellite derived Chl yielded more uncertainty when compared with in situ Chl. This could be because of the fact that the MODIS Chl product algorithm was designed for oceans and may not be appropriate for inland lakes. Satellite derived SDD compared fairly well with in situ SDD. From these results it is evident that with more in situ observations and improved algorithms, MODIS can be useful to improve how water quality is monitored on Lake Victoria.