Chlorophyll-a and total suspended matter retrieval and comparison of C2RCC neural network algorithms on Landsat 8 data over Wangi-Wangi Island, Indonesia

As a marine protected area, the water quality of Wangi-wangi Island, located in South East Sulawesi, should be monitored regularly to detect potential eutrophication. However, in situ water quality monitoring is expensive and unpractical due to its remote area. The Landsat 8 satellite is able to retrieve water quality information, such as chlorophyll-a and total suspended matter (TSM) over period of times using several neuronal nets of Coast 2 Regional CoastColor/C2RCC algorithm. The C2RCC is a neural network algorithm, trained on simulated top of atmosphere reflectance. There are 2 sets of neuronal nets on C2RCC algorithm which is C2RCC-Nets and C2X-Nets. The aim of this study is to obtain the best neuronal nets of C2RCC algorithm to retrieve chlorophyll-a and TSM from Landsat 8 image over Wangi-wangi island. The in situ measurement of chlorophyll-a and TSM were measured during satellite pass of Landsat 8. Four scenes of Landsat 8 data overpass of Wangi-wangi Island in 2016 were selected in this study. The result showed that the C2RCCnets is more accurate to retrieve chlorophyll-a information with R2=0.0747, RMSE = 0.1046 mg/m3 and MAE = 0.0834 mg/m3. The C2RCC-nets showed good performance to retrieve TSM information with R2=0.1586, RMSE = 0.4327 g/m3 and MAE = 0.4258 g/m3. During our study timeframe, the trophic status of Wangi-wangi Island was oligotrophic with no signs of eutrophication.