Earth Observation of Anthropogenic Influence on Wetland Environments: Insights into Water Color and Dike-Pond Systems

Over past centuries, humans have extensively reclaimed land for food production and urbanization in response to population growth and environmental threats. Among various reclamation categories, areas with seasonal or permanent water inundations which also represent a distinct type of wetland can be referred to as "wetland reclamations" or "dike-pond systems". These areas consistently consume substantial water resources, contributing to agricultural, aquaculture and salt production. Positioned adjacent to lakes, floodplains, and estuaries for convenient water access, wetland reclamations emerge as critical drivers of wetland losses. The Yangtze Plain lakes in China exemplify a reclamation hotspot where extensive reclamations for food production and urbanization have occurred since the 1950s. Rice and freshwater aquaculture production on the Yangtze Plain contribute over 50% to China's total output. Lake area loss and water quality deterioration were widely observed in this region. Despite being critical for food production, the correlation between wetland reclamations and water quality in hotspots like the Yangtze Plain remains poorly understood, hindering the assessment of the trade-off between food production and ecosystem health. Besides, China coastal regions and global mega cities have witnessed extensive reclamations, intensifying eco-environmental pressures such as habitat/biodiversity loss and water quality deterioration. Current reclamation datasets primarily target specific coastal/inland regions or limited study periods, emphasizing the need for a more comprehensive global assessment of reclamation efforts and their socio-economic contributions and eco-environmental consequences.

Earth observation (EO) has become a cornerstone in acquiring geo-information to monitor various land use dynamics, leveraging advancements in cloud platforms and image segmentation technology. Recent studies highlight successful applications of large-scale image processing and machine learning techniques in land use investigation, offering new opportunities for studying wetland reclamations. Against this backdrop, this thesis is designed to address three key questions: 1) what are the change patterns of lacustrine water color on the Yangtze Plain over the past four decades and whether reclaimed aquaculture ponds affect water colors? 2) To what extent have inland and coastal wetland reclamations been constructed and developed at the global scale over the past four decades, and what is their contribution to food production and potential environmental impacts? 3) What are the change trajectories of a specific type of wetland reclamations, namely reclaimed aquaculture ponds, over the past decades, and to what extent have they affected climate change, ecosystem, and diversity?

The first study assessed water color dynamics in a lake reclamation hotspot in China, namely Yangtze Plain, with a focus on examining whether reclaimed aquaculture zones affected water color and the concentration of chlorophyll-a (Chl-a) and dissolved organic carbon (DOC). Utilizing all available Landsat 5/8 images from 1984 to 2018, the study applied the Forel-Ule Index (FUI) model to derive hue angle values in 61 lakes. Results indicated statistically significant changes in the annual hue angle values for 25 lakes (P value <0.05), with three distinctive change patterns exhibited among 61 lakes. Analyses incorporating environmental factors such as wind speed, runoff, and normalized difference vegetation index (NDVI) demonstrated statistically their significant correlations with water color changes for 33 lakes. Analyses with three environmental factors, including wind speed, runoff, and normalized difference vegetation index (NDVI), showed that 33 lakes exhibited statistically significant correlations with water color changes. The comparison analysis using the dataset of reclaimed aquaculture zones for the Yangtze Plain lakes revealed differences in water color between aquaculture zones and adjacent natural waters in lakes. Field trips measuring chlorophyll-a (Chl-a) and dissolved organic carbon (DOC) concentrations confirmed the potential of reclaimed aquaculture zones to increase light-absorbing material concentrations, thereby affecting water quality. This study significantly contributes to understanding water color changes in Yangtze Plain Lake in response to both environmental variability and human activities.

The second study investigated the extent, distribution, and trajectory of global wetland reclamations over three periods (1984-2000, 2001-2010, and 2011-2020), and assessed their contribution to fishery production and potential impacts on protected areas. A deep learning framework was applied to map global wetland reclamations based on the global surface water (GSW) dataset. The generated maps revealed a total area of 132, 886 km2 of global wetland reclamations from 1984 to 2020. A comparative analysis between inland and coastal regions (30 km ocean-connected areas) highlighted that inland region exhibited larger total areas and experienced more substantial increases in wetland reclamations. Analyses using random points within the 2011-2020 map indicated that 94% of wetland reclamations were utilized for aquaculture ponds or paddy fields, with the remaining 6% serving various purposes such as salt ponds and tailing ponds. Correlation analyses between the total area estimates of aquaculture ponds and national fishery productions demonstrated statistically significant relationships. Pixelwise comparisons with the global tidal flat dataset further disclosed that 60% of tidal flat losses from 1984 to 2019 could be attributed to wetland reclamations. Over the latest period (2011-2020), 6% and 14% of inland and coastal wetland reclamations, respectively, were situated within protected areas. This study underscores the urgent need to prioritize the sustainable management of the ever-expanding wetland reclamations. The generated maps could serve as a high-quality database for monitoring potential consequences of wetland reclamation, and support the planning and validation of future restoration efforts.

The third study aimed to monitor global reclaimed aquaculture ponds at finer temporal resolutions across nine periods (1984-1994, 1995-2000, and every three-year interval from 2001 to 2021). Employing the random forest algorithm on the GSW and NDVI images, this study produced nine global maps. Over the period from 1984 to 2021, the total area of reclaimed aquaculture ponds increased from 18219.79 km 2 ± 893.57 km 2 to 45925.48 km 2 ± 1895.42 km 2 . Analyses of per unitarea aquaculture production indicated that the total value rose by 1.6 times, with its trend primarily dominated by Asia. Evaluation of methane emissions from global RAP revealed that the median value of annual emissions increased by 1.8 times. Pixel-wise comparison with a global mangrove dataset (1996-2021) unveiled that RAP led to approximately 15% of total mangrove losses worldwide. Assessment of potential threat areas in key biodiversity areas (KBAs) disclosed that, during the recent period of 2019-2021, 21% of RAP were distributed within KBAs, with 83% of these areas encroaching on bird habitats. This study provides comprehensive insights into the trajectory, contribution, and impacts of global reclaimed aquaculture ponds over the past four decades, emphasizing the need for sustainable management practices that balance aquaculture development with eco-environment conservation.

In summary, the thesis leveraged earth observation methods to comprehensively map global wetland reclamations and reclaimed aquaculture ponds at the global scale while assessing their contributions to food production and eco-environmental impacts. It began by offering a holistic evaluation of the influence of reclaimed aquaculture zones on water quality in a reclamation hotspot in China, and further quantified the broader impacts of wetland reclamations and reclaimed aquaculture zones on climate change and natural ecosystems on a global scale. The datasets and frameworks generated in this thesis have significant potential to serve as a foundational resource for designing, monitoring, and evaluating future sustainable development policies. The three studies collectively underscore the importance of conducting more detailed investigations into global reclamations and their environmental impacts, calling for future trade-off efforts between human needs and eco-environmental conservation.