A Transfer Learning Technique for Inland Chlorophyll-a Concentration Estimation Using Sentinel-3 Imagery

Chlorophyll-a (Chla) concentration, which serves as a phytoplankton substitute in inland waters, is one of the leading indicators for water quality. Generally, water samples are analyzed in professional laboratories, and Chla concentrations are measured regularly for the purpose of water quality monitoring. However, limited spatial water sampling and the labor-intensive nature of data collection make global and long-term monitoring difficult. The developments of remote-sensing optical sensors and technologies make the long-term monitoring of Chla concentrations for an entire water body more achievable. Many studies based on machine learning techniques, such as regression and artificial neural network (ANN) methods, have recently been proposed for Chla concentration estimation using optical satellite images. The methods based on machine learning can achieve accurate estimation. However, overfitting problems may arise because the in situ Chla dataset is generally insufficient to train a complicated machine learning model, which makes trained models inapplicable. In this study, an ANN model containing three convolutional and two fully connected layers with 4953 unknown parameters is designed. A transfer learning method, consisting of model pretraining, main-training, and fine-tuning stages, is proposed to ease the problem of insufficient in situ samples. In the model pretraining stage, the ANN model is pretrained and initialized using samples derived from an existing Chla concentration model. The pretrained ANN model is then fine-tuned using the proposed transfer learning technique with in situ samples collected in five different campaigns carried out during early 2019 from Laguna Lake, the Philippines. Before the transfer learning, data augmentation and rebalancing methods are conducted to enrich the variability and to near-uniformly distribute the in situ samples in Chla concentration space, respectively. To estimate the alleviation of model overfitting, the trained ANN model, using an in situ dataset from Laguna Lake, was tested using an in situ dataset from Lake Victoria, Uganda, obtained in 2019, which has a similar trophic state as Laguna Lake. The experimental results from Sentinel-3 imagery indicated that the overfitting problem was significantly alleviated and the trained ANN model outperformed related models in terms of the root-mean-squared error of the estimated Chla concentrations.

Multi-Reservoir Water Quality Mapping from Remote Sensing Using Spatial Regression

Regional water quality mapping is the key practical issue in environmental monitoring. Global regression models transform measured spectral image data to water quality information without the consideration of spatially varying functions. However, it is extremely difficult to find a unified mapping algorithm in multiple reservoirs and lakes. The local model of water quality mapping can estimate water quality parameters effectively in multiple reservoirs using spatial regression. Experiments indicate that both models provide fine water quality mapping in low chlorophyll-a (Chla) concentration water (study area 1; root mean square error, RMSE: 0.435 and 0.413 mg m−3 in the best global and local models), whereas the local model provides better goodness-of-fit between the observed and derived Chla concentrations, especially in high-variance Chla concentration water (study area 2; RMSE: 20.75 and 6.49 mg m−3 in the best global and local models). In-situ water quality samples are collected and correlated with water surface reflectance derived from Sentinel-2 images. The blue-green band ratio and Maximum Chlorophyll Index (MCI)/Fluorescence Line Height (FLH) are feasible for estimating the Chla concentration in these waterbodies. Considering spatially-varying functions, the local model offers a robust approach for estimating the spatial patterns of Chla concentration in multiple reservoirs. The local model of water quality mapping can greatly improve the estimation accuracy in high-variance Chla concentration waters in multiple reservoirs.

Geosynchronous Satellite GF-4 Observations of Chlorophyll-a Distribution Details in the Bohai Sea, China

We analyzed the distribution of chlorophyll-a (Chla) in the Bohai Sea area based on data from the geosynchronous orbit optical satellite Gaofen-4 (GF-4), which was launched in 2015, carrying a panchromatic multispectral sensor (PMS). This is the first time the geosynchronous orbit optical satellite GF-4 remote-sensing data has been used in China to detect the Chla change details in the Bohai Sea. A new GF-4 retrieved model was established based on the relationship between in situ Chla value and the reflectance combination of 2 and 4 bands, with the R2 of 0.9685 and the total average relative error of 37.42%. Twenty PMS images obtained from 2017 to 2019 were applied to analyze Chla in Bohai sea. The results show that: (1) the new built Chla inversion model PMS-1 for the GF-4 PMS sensor can extract Chla distribution details in the Bohai Sea well. The high Chla content in the Bohai Sea is mainly located in coastal areas, such as the top of Laizhou Bay, Bohai Bay and Liaodong Bay, with the value being around 13 µg/L. The concentration of Chla in the Bohai Strait and northern Yellow Sea is relatively low with the value being around 5 µg/L. (2). Taking full advantage of the continuous observation of geostationary orbit satellite, GF-4 with a high-resolution sensor PMS of 50 m can effectively detect short-term change (changes within 10 min) in Chla concentration. The changes mainly appear at the southwest and northeast costal area as well as in the center of Bohai Sea with the change value of around 3 µg/L. (3) The change of Chla concentration in the Bohai sea is related to the environmental factors such as seawater temperature, salinity, illumination and nutrient salts, as well as the dynamic factors such as wind, flow field and tidal current.

A fuzzy function model for remote evaluation of chlorophyll-a concentration in the Northern of South China Sea-Hong Kong

Chlorophyll-a (Chla) concentration is an important indicator to monitor eutrophication, which is a persistent problem that affects the ecological health of coastal water or shallow lakes. We have applied a Gaussian-like fuzzy function model for the estimation of Chla concentration in Hong Kong Coastal Waters, based on the spectral signature using the HJ-1A /1B CCD images and measured data. The method of this paper is as follows. Firstly, reflectance peak and fluorescence band were used to figure out the bands or bands combination which impact the Chla concentration significantly; and then calculated the value of all the pixels without measured data according to their similarity degree with the measured points; finally, the full Chla concentration maps in the study area were produced by GIS spatial interpolation. It is concluded in this paper that the method can retrieve the Chla concentration accurately and the result of changes detected coincides with the observed result extremely, what’s more, the maps generated are continuous and smooth which are quite different from traditional RS maps that can only accurate to pixel size.

Changing Characteristics of Chlorophyll a in the Context of Internal and External Factors: A Case Study of Dianchi Lake in China

During the past 20 years, the ecological environment of Dianchi Lake has been adversely affected by climate change and human activities, which directly affected the ecosystem and biodiversity of the Dianchi Lake watershed. Analyzing the spatiotemporal variation of chlorophyll a (Chla) concentration of Dianchi Lake and exploring the internal and external factors effect on Chla concentration is the basis for controlling and improving the water ecological environment of Dianchi Lake, and it is also the key to prevent and control the water pollution of Dianchi Lake. In this study, the water quality of Dianchi Lake was examined using 12 water quality indicators from 10 water quality monitoring sites for the duration between 2000 to 2017. The changing characteristics of Chla in the context of internal and external factors were analyzed. The spatiotemporal evolution process of Chla concentration in the past 20 years was also evaluated. The results indicated that Chla concentration was significantly and positively correlated with the chemical oxygen demand (CODCr), the Dianchi Lake watershed gross domestic product (GDP), and the impervious surface area (ISA) of the watershed, in addition to the total phosphorus (TP), biochemical oxygen demand (BOD5), ammonia hydrogen (NH3-N), water temperature (WT), and civil vehicle ownership. Moreover, a significant and negative correlation was noticed between Dianchi Lake watershed GDP and NH3-N, BOD5, TP, total nitrogen (TN), and comprehensive nutrition state index (TLI). The Dianchi Lake population was negatively correlated with TP, TLI, and BOD5. The concentration of Chla in Dianchi Lake was affected by both internal factors, and external factors such as anthropogenic activities, the latter of which was the main cause of the continuous deterioration of the lake water quality.

Exploring the Influence Mechanism of Meteorological Conditions on the Concentration of Suspended Solids and Chlorophyll-a in Large Estuaries Based on MODIS Imagery

In estuary areas, meteorological conditions have become unstable under the continuous effects of climate change, and the ecological backgrounds of such areas have strongly been influenced by anthropic activities. Consequently, the water quality of these areas is obviously affected. In this research, we identified periods of fluctuation of the general meteorological conditions in the Yangtze River Estuary using a wavelet analysis. Additionally, we performed a spatiotemporal evaluation of the water quality in the fluctuating period by using remote sensing modeling. Then, we explored how the fluctuating meteorological factors affect the distribution of total suspended solids (TSS) and chlorophyll-a (Chla) concentration. (1) The results show that from 2000 to 2015, temperature did not present significant fluctuations, while wind speed (WS) and precipitation (PR) presented the same fluctuation period from January 2012 to December 2012. (2) Based on the measured water sample data associated with Moderate Resolution Imaging Spectroradiometer (MODIS) imagery, we developed a water quality algorithm and depicted the TSS and Chla concentrations within the WS and PR fluctuating period. (3) We found that the TSS concentration decreased with distance from the shore, while the Chla concentration showed an initially decreasing trend followed by an increasing trend; moreover, these two water quality parameters presented different inter-annual variations. Then, we discussed the correlation between the changes in the TSS and Chla concentrations and the WS and PR variables. The contribution of this research is reflected in two aspects: 1. variations in water quality parameters over a wide range of water bodies can be evaluated based on MODIS data; 2. data from different time periods showed that the fluctuations of meteorological elements had different impacts on water bodies based on the distance from the shore. The results provide new insights for the management of estuary water environments.

VIP-BP model for retrieving chlorophyll a concentration in the river by using remote sensing data

Chlorophyll a (Chla) is an important indicator of phytoplankton biomass in waters, and its concentration can reflect the degree of eutrophication. This paper is aimed to develop a highly accurate and universally applicable retrieval model for the concentration of Chla in rivers using remote sensing data. Taking the middle and lower reaches of the Han River as the study area, the Chla retrieval model (VIP-BP model) is established by combining the Variable Importance Projection Index and BP neural algorithm and then calibrated by the measured data from 2012 to 2013. This model uses the VIP index for selection of the appropriate spectrum transformation form and input bands. Then, the BP neural network algorithm is integrated to estimate Chla concentration. After validation and comparison with the three-band model, the results suggest that the VIP-BP model could more accurately and really reflect the changes in Chla concentration than the three-band model in the study area. When Chla concentration decreases, the retrieval error of both models increases, while the error of the VIP-BP model is significantly lower than that of the three-band model, which indicates that the VIP-BP model is more stable and preferred.

Studying Chlorophyll-A Concentration of Bosten Lake (Xinjiang, Northwest China) by MERIS Data

The Chlorophyll-a (Chla) concentration in Bosten Lake was estimated and mapped using the data of the Medium Resolution Imaging Spectrometer (MERIS) on board the ENVIronmental SATellite (ENVISAT) platform. The fixed aerosol option was chosen and local aerosol optical thickness was used in SeaDAS. The Chla concentration was retrieved by the OC3E algorithm and verified by Field data with high correlation coefficient of 0.79. It showed strong horizontal heterogeneities, which is high at the Huangshuigou region, mediate along the boundary area, and low at the middle of the lake, and decreases from the boundary to the center of the Lake. Its spatial distribution is controlled by the location of inlet and outlet and the type and quantity of discharging around the lake. On sep. 22, 2010, its value is up to 10.98 mg m-3. The minimum, maximum, average and median value of Chla concentration on Aug. 6, 2011 from MERIS data in Bosten Lake is 2.72, 8.93, 3.90 and 3.69 mg m-3.