5:15 PM - 7:15 PM
[AHW22-P03] Empirical modeling of total nitrogen distribution in seawater using optical Earth observation satellites: A case study of Yoron Island, Central Ryukyus
Keywords:water quality, remote sensing, total nitrogen, coral reef island, empiric model
The Earth observation satellites facilitate the prediction of water quality distribution by analyzing various physical, biological, and chemical parameters. Among these, nutrient dynamics has been a focal point of remote sensing research. However, detecting nutrients such as total nitrogen (T-N), which often occurs at low concentrations, presents significant challenges for remote sensing applications. This study aims to develop an empirical model for estimating the spatial distribution and recent temporal variations of T-N concentration using optical Earth observation satellite data over Yoron Island, a coral reef-fringed island in the Central Ryukyus.
To achieve this objective, the first step involved constructing an empirical model for extracting T-N concentration data. This model was developed using an empirical approach, which establishes relationships by assigning input band values as coefficients, generating multiple models and selecting the optimal fit based on error metrics. In situ data were obtained from reference sources and integrated with Landsat imagery at a 30-meter resolution. The methodology included preprocessing Landsat data to obtain surface reflectance values, overlaying these with in situ reference data, and conducting regression analysis across multiple models. This study evaluated five models and six spectral bands (Blue, Green, Red, NIR, SWIR 1, and SWIR 2), focusing on identifying the most suitable band for T-N detection. The model demonstrating the highest regression accuracy was selected for further analysis. The final model was then evaluated by assessing its error using the Root Mean Square Error (RMSE). A model with minimal RMSE was deemed appropriate for time-series analysis.
The chosen model is represented by the equation T-N (mg/L) = 11.877 (Blue)² - 0.6456 (Blue) + 0.1703. This model was applied to a 2014 Landsat image of the eastern region of Yoron Island, yielding a T-N concentration range between 0.13 and 0.98 mg/L. Furthermore, a time-series analysis from 1984 to 2022 reveals annual variations in T-N distribution, with higher concentrations observed in the lagoon adjacent to Ooganeku Beach.
The findings demonstrate that the empirical model accurately predicts T-N distribution in the lagoon of a small reef island. Elevated T-N concentrations observed in the lagoon near Ooganeku Beach are closely associated with the morphological characteristics of the island's groundwater basin, indicating that nutrient influx from terrestrial human activities is transported to the marine environment through groundwater discharge.
To achieve this objective, the first step involved constructing an empirical model for extracting T-N concentration data. This model was developed using an empirical approach, which establishes relationships by assigning input band values as coefficients, generating multiple models and selecting the optimal fit based on error metrics. In situ data were obtained from reference sources and integrated with Landsat imagery at a 30-meter resolution. The methodology included preprocessing Landsat data to obtain surface reflectance values, overlaying these with in situ reference data, and conducting regression analysis across multiple models. This study evaluated five models and six spectral bands (Blue, Green, Red, NIR, SWIR 1, and SWIR 2), focusing on identifying the most suitable band for T-N detection. The model demonstrating the highest regression accuracy was selected for further analysis. The final model was then evaluated by assessing its error using the Root Mean Square Error (RMSE). A model with minimal RMSE was deemed appropriate for time-series analysis.
The chosen model is represented by the equation T-N (mg/L) = 11.877 (Blue)² - 0.6456 (Blue) + 0.1703. This model was applied to a 2014 Landsat image of the eastern region of Yoron Island, yielding a T-N concentration range between 0.13 and 0.98 mg/L. Furthermore, a time-series analysis from 1984 to 2022 reveals annual variations in T-N distribution, with higher concentrations observed in the lagoon adjacent to Ooganeku Beach.
The findings demonstrate that the empirical model accurately predicts T-N distribution in the lagoon of a small reef island. Elevated T-N concentrations observed in the lagoon near Ooganeku Beach are closely associated with the morphological characteristics of the island's groundwater basin, indicating that nutrient influx from terrestrial human activities is transported to the marine environment through groundwater discharge.