Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor | Department of Land Surveying and Geo-Informatics | en_US |
| dc.contributor.advisor | Wong, Man Sing (LSGI) | en_US |
| dc.creator | Hafeez, Sidrah | - |
| dc.identifier.uri | https://theses.lib.polyu.edu.hk/handle/200/11804 | - |
| dc.language | English | en_US |
| dc.publisher | Hong Kong Polytechnic University | en_US |
| dc.rights | All rights reserved | en_US |
| dc.title | Integrating remote sensing and machine learning technologies for coastal water quality monitoring in Hong Kong and the Pearl River Estuary | en_US |
| dcterms.abstract | The present study aimed to develop robust algorithms for monitoring marine Water Quality Indicators (WQIs) i.e., Chlorophyll-a (Chl-a), Total Suspended Solids (TSS) and Turbidity in complex coastal waters of Hong Kong and the Pearl River Estuary (PRE) using multi-sensor data. For this purpose, 17 years (from 1999 to 2015) of satellite data from Landsat mission including Landsat 5 Thematic Mapper (TM), Landsat 7 Enhanced Thematic Mapper Plus (ETM+) and Landsat 8 Operational Land Imager (OLI) and in-situ WQIs measurements were first used. Concentrations of Chl-a, TSS and turbidity were estimated with several machine learning techniques including Artificial Neural Network (ANN), Random Forest (RF), Cubist regression (CB), and Support Vector Regression (SVR). Landsat (5,7,8) reflectance data were compared with in-situ reflectance data to evaluate the performance of machine learning models. Results showed that the highest accuracies for estimating WQIs were achieved by ANN for both, in-situ reflectance data (89%-Chl-a, 93%-TSS, and 82%-turbidity) and satellite data (91%-Chl-a, 92%-TSS, and 85%-turbidity). The coastal WQIs retrieved by the ANN models were further compared to those retrieved by "standard Case-2 Regional/Coast Colour" (C2RCC) processing chain model C2RCC-Nets. The root mean square error (RMSE) for estimating Chl-a and TSS was 2.7 µg/L and 3.3 mg/L, respectively using ANN. Whereas, when C2RCC was applied on Landsat 8 OLI data, RMSEs were 12.9 µg/L and 12.7 mg/L for estimating Chl-a and TSS concentrations, respectively. | en_US |
| dcterms.abstract | Relative variable importance was also conducted to investigate the consistency between in-situ reflectance data and satellite data, and results showed that both datasets are similar. The red band (wavelength ≈ 665 nm) and the product of red and green band (wavelength ≈ 560 nm) were influential inputs in both reflectance data sets for estimating TSS and turbidity. Whereas the ratio between red and blue band (wavelength ≈ 490 nm) as well as the ratio between Near Infrared (NIR) (wavelength ≈ 865 nm) and blue band and green band are demonstrated to be more useful for the estimation of Chl-a concentrations. The results indicated that the ANN-based machine learning approaches perform better and thus, can be used for water quality monitoring with satellite data in optically complex coastal waters. With only Landsat data, the complex coastal environment of Hong Kong and PRE cannot be monitored on a high temporal scale. Nonetheless, the synergy of fine-to-moderate resolution (i.e., 10-60 m) satellite data of Sentinel-2 Multispectral Imager (MSI) and Landsat 8 OLI provide the possibility to monitor the dynamics of sensitive aquatic systems. Before developing new algorithms for their combined use, it is imperative to assess the spectral consistency of both sensors. The spectral consistency between Landsat 8 OLI and Sentinel-2 A/B MSI was evaluated, mainly in terms of the top of atmosphere reflectance (pt), rayleigh-corrected reflectance (prc),and remote sensing reflectance (Rrs). To check the spectral consistency under various atmospheric and aquatic conditions, near-simultaneous same-day overpass images of Landsat 8 and Sentinel-2 A/B were selected over diverse coastal and inland areas across Mainland China and Hong Kong. Results showed that spectral data obtained from Landsat 8 OLI and Sentinel-2 A/B MSI were consistent. Landsat 8 OLI and Sentinel-2 A/B MSI average products difference (average difference of all matching bands) in terms of mean absolute percentage error (MAPE) was ~8% in pt, and ~10% in both prc and Rrs. Overall, the green channel is the most consistent with the lowest MAPE of ≤ 4.6% in all products. | en_US |
| dcterms.abstract | Further, this study discusses the combined use of Landsat 8 and Sentinel-2 A/B data for different aquatic applications i.e., time series of TSS over the coastal and inland waters, floating algal area comparison and tracking changes in coastal floating algae (FA). In the second step, the potentials of synergistic data from Landsat 8 OLI and Sentinel-2 MSI were assessed to monitor coastal water quality over an improved temporal scale (~3 days). Water quality models for retrieving Chl-a and TSS concentrations using combined and separate sensor data were evaluated. All sensor bands were used in single sensor models and only matching bands were used in combined sensor models. TSS model with combined OLI and MSI data outperformed single sensor-based models. Whereas, the single sensor-based model with MSI data outperformed the combined sensor-based model for retrieving Chl-a from satellite images, therefore, showing the importance of additional Red Edge bands of MSI data to retrieve Chl-a in coastal waters. The time-series of Chl-a and TSS from 2013-2018 were mapped using MSI and OLI data and it indicated spatio-temporal variation in Chl-a and TSS in coastal waters of Hong Kong. | en_US |
| dcterms.abstract | As the temporal resolutions of the polar-orbiting satellite are not considered adequate for monitoring highly dynamic oceanic processes such as in PRE, especially when considering data gaps due to cloud contamination. Therefore, the feasibility study of the Advanced Himawari Imager (AHI) onboards the Himawari-8 in mapping TSS concentrations in PRE was conducted for the first time. This study investigated and identified an appropriate atmospheric correction method for AHI data; developed a model for TSS concentrations estimation using hyperspectral data and in-situ measurements of TSS; validated the model; and assessed its potential to capture diurnal changes using AHI imagery. Two image-based atmospheric correction methods, the NIR-SWIR method and the SWIR method were tested for correcting the AHI data. Then, the new model was applied to the atmospherically corrected AHI data to map TSS and its diurnal changes in the PRE and neighbouring coastal areas. The results showed a good agreement between the AHI-derived TSS, and in-situ measured TSS with a coefficient of determination (R²) of 0.85, a root mean square error (RMSE) of 3.9 mg/L and average percentage difference (APD) of 30% (TSS range 1-40 mg/L). Moreover, the diurnal variation in the turbidity front, using the Normalized Suspended Material Index (NSMI), showed the capability of AHI data to track diurnal variation in turbidity fronts, due to high TSS concentrations at high temporal frequency. Three image fusion methods such as STARFM, SSTF and Fit-FC were tested to fused satellite data for water applications and the results show that water quality parameter retrieved using fused MSI-like images show low RMSE when compared with in-situ Chl-a and TSS data and even geostationary Himawri-8/AHI images fused with Landsat 8/OLI image can provide useful information on hourly scale when fused using Fit-FC method. The present study indicates that AHI data with high image capturing frequency can be used to map surface TSS concentrations. These TSS measurements at high frequency are not only important for monitoring the sensitive coastal areas but also for scientific understanding of the spatial and temporal variation of TSS. Also, the results of the first two parts of this research suggest that the approach outlined in the present study can be used for routine monitoring of Chl-a and TSS in the coastal waters of Hong Kong using satellite data. | en_US |
| dcterms.extent | 238 pages : color illustrations | en_US |
| dcterms.isPartOf | PolyU Electronic Theses | en_US |
| dcterms.issued | 2022 | en_US |
| dcterms.educationalLevel | Ph.D. | en_US |
| dcterms.educationalLevel | All Doctorate | en_US |
| dcterms.LCSH | Water quality -- China -- Hong Kong | en_US |
| dcterms.LCSH | Water quality -- China -- Pearl River Estuary | en_US |
| dcterms.LCSH | Remote sensing | en_US |
| dcterms.LCSH | Machine learning. | en_US |
| dcterms.LCSH | Hong Kong Polytechnic University -- Dissertations | en_US |
| dcterms.accessRights | open access | en_US |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/11804