Evaluation Of Sea Surface Temperature From In Situ Measurements Compared To Satellite Observations And Model Output In Panjang Island, Banten
Zulfikar Kartadimaja(1*), Rima Rachmayani(1), Sri Yudawati Cahyarini(2)

1)Program Study of Oceanography, Faculty of Earth Sciences and Technology, Bandung Institute of Technology, Bandung, Indonesia
2)Paleoclimate and Paleoenvironment Research Group, Research Center for Climate and Atmosphere, National Research and Innovation Agency, Bandung, Indonesia

Abstract

Surface sea temperature (SST) is a crucial parameter in climate change studies and is obtained from various sources, including in-situ measurements, satellite observations, reanalysis, and models. However, due to the limited availability of in-situ data, researchers often rely on satellite observations, models, or reanalysis, all of which are susceptible to errors. This study aims to improve the accuracy of SST data from satellite observations and models by evaluating and refining the data through comparison with in-situ SST data using timeseries analysis. The focus of the research is Pulau Panjang, Banten, and the data collection period spans from April 15, 2022, to February 7, 2023. The data was gathered from satellites (OISST V2.1), models (Hycom GOFS V3.0-3.1), and in-situ (loggers HOBO U-24 and TidBit V2.1). The results show that satellite-derived SST values perform better than the model. Correlation coefficients between satellite and in-situ SST data (r = 0,853- n = 299- p-value < 0,001) surpass those between model and in-situ SST data (r = 0,685- n = 299- p-value < 0,001). Additionally, satellite observations exhibit lower errors ({\bar{x}}_{bias} = 0,445{^\circ}C- \sigma_{bias} = 0,304{^\circ}C- RMSE = 0,538{^\circ}C) compared to the model ({\bar{x}}_{bias} = 0,472{^\circ}C- \sigma_{bias} = 0,32{^\circ}C- RMSE = 0,57{^\circ}C). The variance tests show no significant difference for satellite (p-value = 0,785) and model (p-value = 0,346) data against in-situ data, while the mean tests reveal disparities (p-value < 0,001). The calibration process reduces error values for both satellite ({\bar{x}}_{bias} = 0,276{^\circ}C- \sigma_{bias} = 0,225{^\circ}C- RMSE = 0,356{^\circ}C) and model data ({\bar{x}}_{bias} = 0,398{^\circ}C- \sigma_{bias} = 0,298{^\circ}C- RMSE = 0,397{^\circ}C), but the correlation coefficients remain stable at 0,853 (satellite) and 0,685 (model). In conclusion, satellite observations providing a more precise picture of the SST conditions on Panjang Island, Banten and the calibration process can improve the quality of satellite data and models.

Keywords: Sea surface temperature- Evaluation- Satellite- Model- Pulau Panjang, Banten

Topic: Ocean Remote Sensing and Marine Technology