Improving Chlorophyll-A Estimation From Sentinel-2 (MSI) in the Barents Sea Using Machine Learning

Abstract

This paper addresses methodologies for remote sensing of ocean Chlorophyll-a (Chl-a), with emphasis on the Barents Sea. We aim at improving the monitoring capacity by integrating in-situ Chl-a observations and optical remote sensing to locally train Machine Learning (ML) models. For this purpose, in-situ measurements of Chl-a ranging from 0.014-10.81mg/m 3 , collected for the years 2016-2018, were used to train and validate models. To accurately estimate Chl-a, we propose to use additional information on pigment content within the productive column by matching the depth-integrated Chl-a concentrations with the satellite data. Using the optical images captured by the Multi-Spectral Imager Instrument on Sentinel-2 and the in-situ measurements, a new spatial windowbased match-up dataset creation method is proposed to increase the number of match-ups and hence improve the training of the ML models. The match-ups are then filtered to eliminate erroneous samples based on the spectral distribution of the remotely sensed reflectance. In addition, we design and implement a neural network model dubbed as the Ocean Color Net (OCN), that has performed better than existing ML-based techniques, including the Gaussian Process Regression (GPR), regionally tuned empirical techniques, including the Ocean Color (OC3) algorithm and the spectral band ratios, as well as the globally trained Case-2 Regional/Coast Colour (C2RCC) processing chain model C2RCC-networks. The proposed OCN model achieved reduced Mean Absolute Error compared to the GPR by 5.2%, C2RCC by 51.7%, OC3 by 22.6%, and spectral band ratios by 29%. Moreover, the proposed spatial window and depth-integrated match-up creation techniques improved the performance of the proposed OCN by 57%, GPR by 41.9%, OC3 by 5.3%, and spectral band ratio method by 24% in terms of RMSE compared to the conventional match-up selection approach.

Applications of Machine Learning in Chemical and Biological Oceanography

“…Likewise, a neural network model named Ocean Color Net (OCN)) with match-up data sets showed to improve the Chl-a estimation on the surface and within the productive zone of the Barents Sea using satellite imagery data. 83 A new spatial window-based match-up data set was created by matching depth-integrated in situ Chl-a concentration with the multispectral remote sensing images from Sentinel-2. After the removal of the erroneous samples in the match-ups data sets based on satellite reflectance, the OCN was trained and tested with the match-ups data set.…”

“…Also, the model predicted that CDOM plays a vital role in estimating the spatial-temporal distribution of Chl-a from the satellite color data. Likewise, a neural network model named Ocean Color Net (OCN)) with match-up data sets showed to improve the Chl-a estimation on the surface and within the productive zone of the Barents Sea using satellite imagery data . A new spatial window-based match-up data set was created by matching depth-integrated in situ Chl-a concentration with the multispectral remote sensing images from Sentinel-2.…”

Meta-Analysis of Satellite Observations for United Nations Sustainable Development Goals: Exploring the Potential of Machine Learning for Water Quality Monitoring

“…In cases where data are normally fitted, the R 2 value typically falls within the range of 0 to 1. However, it is important to acknowledge that poorly fitted data can sometimes yield negative R 2 values, as demonstrated by Asim et al [162]. Their study focused on the development of a Case 2 Regional CoastColour (C2RCC)-net model for Chl-a retrieval, utilizing a match-up criterion based on C2RCC in situ Chl-a measurements.…”

“…In this research, various subsets of deep learning models have been utilized, each with its own set of variants. Among the popular advanced networks for feed-forward neural networks (FNNs) are the MLP [31,40,161,162,238,240,245,248], mixture density networks (MDNs) [21,28,251,268], extreme learning machine (ELM) [30], cascade forward neural network (CFNN) [30], radial basis function neural network (RBFNN) [242], and Bayesian neural networks (BNNs) [251]. Additionally, recurrent neural networks, such as LSTM, and GRU have also gained significant popularity in this field due to their ability to handle sequential data and capture temporal dependencies [22,42,48,194,249,266].…”

CHLNET: A novel hybrid 1D CNN-SVR algorithm for estimating ocean surface chlorophyll-a

“…Therefore, machine learning was also used to build a unified Chla inversion algorithm across water types. Among them, multilayer perceptron (MLP) (Vilas et al, 2011;D'Alimonte et al, 2012;Awad, 2014), Gaussian process regression (GPR) (Asim et al, 2021), support vector regression (SVR) (Hafeez et al, 2019;He et al, 2020;Hu et al, 2020), and random forest regression (RFR) (Cheng et al, 2021) have demonstrated potential in retrieving Chla. Several approaches that derived from neural network algorithms, including mixture density network (MDN) (Pahlevan et al, 2020), OLCI Neural Network Swarm(ONNS) (Hieronymi et al, 2017), and Case 2 Regional CoastColour (C2RCC) (Doerffer and Schiller, 2007), were used to inverse Chla concentrations for Cases I and II waters.…”