A Deep Learning Model Using Satellite Ocean Color and Hydrodynamic Model to Estimate Chlorophyll-<i>a</i> Concentration
In this study, we used convolutional neural networks (CNNs)—which are well-known deep learning models suitable for image data processing—to estimate the temporal and spatial distribution of chlorophyll-<i>a</i> in a bay. The training data required the construction of a deep learning mode...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-05-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/13/10/2003 |
| _version_ | 1797533409410547712 |
|---|---|
| author | Daeyong Jin Eojin Lee Kyonghwan Kwon Taeyun Kim |
| author_facet | Daeyong Jin Eojin Lee Kyonghwan Kwon Taeyun Kim |
| author_sort | Daeyong Jin |
| collection | DOAJ |
| description | In this study, we used convolutional neural networks (CNNs)—which are well-known deep learning models suitable for image data processing—to estimate the temporal and spatial distribution of chlorophyll-<i>a</i> in a bay. The training data required the construction of a deep learning model acquired from the satellite ocean color and hydrodynamic model. Chlorophyll-<i>a</i>, total suspended sediment (TSS), visibility, and colored dissolved organic matter (CDOM) were extracted from the satellite ocean color data, and water level, currents, temperature, and salinity were generated from the hydrodynamic model. We developed CNN Model I—which estimates the concentration of chlorophyll-<i>a</i> using a 48 × 27 sized overall image—and CNN Model II—which uses a 7 × 7 segmented image. Because the CNN Model II conducts estimation using only data around the points of interest, the quantity of training data is more than 300 times larger than that of CNN Model I. Consequently, it was possible to extract and analyze the inherent patterns in the training data, improving the predictive ability of the deep learning model. The average root mean square error (RMSE), calculated by applying CNN Model II, was 0.191, and when the prediction was good, the coefficient of determination (R<sup>2</sup>) exceeded 0.91. Finally, we performed a sensitivity analysis, which revealed that CDOM is the most influential variable in estimating the spatiotemporal distribution of chlorophyll-<i>a</i>. |
| first_indexed | 2024-03-10T11:14:12Z |
| format | Article |
| id | doaj.art-6666789e4e9140738e80c7676a6fd2e4 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T11:14:12Z |
| publishDate | 2021-05-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-6666789e4e9140738e80c7676a6fd2e42023-11-21T20:33:59ZengMDPI AGRemote Sensing2072-42922021-05-011310200310.3390/rs13102003A Deep Learning Model Using Satellite Ocean Color and Hydrodynamic Model to Estimate Chlorophyll-<i>a</i> ConcentrationDaeyong Jin0Eojin Lee1Kyonghwan Kwon2Taeyun Kim3Environment Data Strategy Center & Environmental Assessment Group, Korea Environment Institute, Sejong 30147, KoreaEnvironment Data Strategy Center & Environmental Assessment Group, Korea Environment Institute, Sejong 30147, KoreaOcean Environment Group, Oceanic, Seoul 07207, KoreaEnvironment Data Strategy Center & Environmental Assessment Group, Korea Environment Institute, Sejong 30147, KoreaIn this study, we used convolutional neural networks (CNNs)—which are well-known deep learning models suitable for image data processing—to estimate the temporal and spatial distribution of chlorophyll-<i>a</i> in a bay. The training data required the construction of a deep learning model acquired from the satellite ocean color and hydrodynamic model. Chlorophyll-<i>a</i>, total suspended sediment (TSS), visibility, and colored dissolved organic matter (CDOM) were extracted from the satellite ocean color data, and water level, currents, temperature, and salinity were generated from the hydrodynamic model. We developed CNN Model I—which estimates the concentration of chlorophyll-<i>a</i> using a 48 × 27 sized overall image—and CNN Model II—which uses a 7 × 7 segmented image. Because the CNN Model II conducts estimation using only data around the points of interest, the quantity of training data is more than 300 times larger than that of CNN Model I. Consequently, it was possible to extract and analyze the inherent patterns in the training data, improving the predictive ability of the deep learning model. The average root mean square error (RMSE), calculated by applying CNN Model II, was 0.191, and when the prediction was good, the coefficient of determination (R<sup>2</sup>) exceeded 0.91. Finally, we performed a sensitivity analysis, which revealed that CDOM is the most influential variable in estimating the spatiotemporal distribution of chlorophyll-<i>a</i>.https://www.mdpi.com/2072-4292/13/10/2003deep learningconvolutional neural networkchlorophyll-<i>a</i>satellitehydrodynamic model |
| spellingShingle | Daeyong Jin Eojin Lee Kyonghwan Kwon Taeyun Kim A Deep Learning Model Using Satellite Ocean Color and Hydrodynamic Model to Estimate Chlorophyll-<i>a</i> Concentration Remote Sensing deep learning convolutional neural network chlorophyll-<i>a</i> satellite hydrodynamic model |
| title | A Deep Learning Model Using Satellite Ocean Color and Hydrodynamic Model to Estimate Chlorophyll-<i>a</i> Concentration |
| title_full | A Deep Learning Model Using Satellite Ocean Color and Hydrodynamic Model to Estimate Chlorophyll-<i>a</i> Concentration |
| title_fullStr | A Deep Learning Model Using Satellite Ocean Color and Hydrodynamic Model to Estimate Chlorophyll-<i>a</i> Concentration |
| title_full_unstemmed | A Deep Learning Model Using Satellite Ocean Color and Hydrodynamic Model to Estimate Chlorophyll-<i>a</i> Concentration |
| title_short | A Deep Learning Model Using Satellite Ocean Color and Hydrodynamic Model to Estimate Chlorophyll-<i>a</i> Concentration |
| title_sort | deep learning model using satellite ocean color and hydrodynamic model to estimate chlorophyll i a i concentration |
| topic | deep learning convolutional neural network chlorophyll-<i>a</i> satellite hydrodynamic model |
| url | https://www.mdpi.com/2072-4292/13/10/2003 |
| work_keys_str_mv | AT daeyongjin adeeplearningmodelusingsatelliteoceancolorandhydrodynamicmodeltoestimatechlorophylliaiconcentration AT eojinlee adeeplearningmodelusingsatelliteoceancolorandhydrodynamicmodeltoestimatechlorophylliaiconcentration AT kyonghwankwon adeeplearningmodelusingsatelliteoceancolorandhydrodynamicmodeltoestimatechlorophylliaiconcentration AT taeyunkim adeeplearningmodelusingsatelliteoceancolorandhydrodynamicmodeltoestimatechlorophylliaiconcentration AT daeyongjin deeplearningmodelusingsatelliteoceancolorandhydrodynamicmodeltoestimatechlorophylliaiconcentration AT eojinlee deeplearningmodelusingsatelliteoceancolorandhydrodynamicmodeltoestimatechlorophylliaiconcentration AT kyonghwankwon deeplearningmodelusingsatelliteoceancolorandhydrodynamicmodeltoestimatechlorophylliaiconcentration AT taeyunkim deeplearningmodelusingsatelliteoceancolorandhydrodynamicmodeltoestimatechlorophylliaiconcentration |