Remote sensing of lake dynamics in Alaska

Abstract

Lakes are abundant in high northern latitude permafrost regions. They are important ecosystem components forming a complex and dynamic landscape with repeated cycles of lake formation and drainage affecting regional hydrological and terrestrial characteristics, biogeochemical processes and carbon cycling, wildlife habitats, and human communities living in the permafrost region. Remote sensing is a useful tool to map the spatial distribution of lakes and assess its change, understand lake dynamics, and to extract useful information to study their associated feedbacks in a changing climate. In this dissertation, I focused on remote sensing studies associated with (1) methane ebullition from a thermokarst lake, (2) post-drainage succession patterns in drained thermokarst lake basins, and (3) lake change dynamics. I developed a semi-automatic classification method based on an Object-based Image Analysis (OBIA) framework to detect methane ebullition bubbles trapped in a snow-free ice-covered lake using high-resolution airborne images of Goldstream lake, Fairbanks, Alaska acquired following freeze up in October of 2011 and 2012. This study showed that remote sensing is a valuable tool to map ebullition bubbles (bubble patches) on the entire lake surface with an accuracy of > 95%, a task that is difficult to achieve through field-based survey alone. The image analysis performed by combining the mapping results from the OBIA and field-based observations showed a relationship between bubble patch brightness and ground-measured methane flux, which was then used to estimate the whole-lake methane flux. A strong inverse exponential relationship (R2 >= 0.79) was found between the percent of the surface area of lake ice covered with bubble patches and distance from the active thermokarst lake margin, indicating high methane production as a response to thermokarst activity that released labile organic-rich carbon along the eroding lake margin. Despite the influence of atmospheric pressure conditions on distribution of ebullition bubble patches following the lake freeze-up events, the spatiotemporal regularity of bubble patches revealed that a larger number of seeps are stable over at least annual timescales. This remote sensing technique is applicable to other regions for mapping ebullition bubbles trapped in snow-free ice-covered lakes, identifying their relative flux, and assessing their spatiotemporal dynamics. By using TerraSAR-X (TSX) Synthetic Aperture Radar (SAR) backscatter data and the Normalized Difference Vegetation Index (NDVI) derived from a Landsat-5 image from the year 2009, I characterized drained thermokarst lake basins (DTLBs) of various age ranging between 0 to 10,000 years since drainage in the northern Seward Peninsula, Alaska. In the study I found logarithmic relationships of basin age from 0 to 10,000 years with mean basin TSX backscatter (R2 = 0.36) and with mean basin NDVI (R2 = 0.53). However, TSX data performed much better to discriminate older basins in the age class 50–10,000 years with R2 = 0.58, while no significant relationship was found between NDVI and basin age. Results of this study demonstrated the potential application of X-band SAR data in combination with NDVI data to enhance differentiation of soil moisture and vegetation status on drained basins for mapping long-term succession dynamics of DTLBs. Finally, I demonstrated the utility of Landsat imagery to identify lake distribution patterns and changes between 1972 and 2014 in six major lake-rich study regions across various permafrost zones covering an area of 68,830 km2 in western Alaska. Even though lake area change was found to be positive (increase by less than 4%) in some study areas while negative (decrease by 4-15%) in others, there was a widespread drainage of mainly large lakes in all regions creating remnant ponds that increased the abundance of lakes <10 ha in all study regions by 2-27%. The average lake area decline observed in various permafrost zones did not represent the trend of individual sites due to spatial heterogeneity in lake characteristics. While lake drainage dominated the non-continuous permafrost zones, areas of continuous permafrost showed both trends of negative and positive lake area change accompanied by major lake drainage events that led to a regional lake area loss in the continuous permafrost zone. This remote sensing technique proved to be useful in identifying ongoing lake drainage and expansion events within study regions and a regional shift in lake distribution (i.e. lake area loss) that is happening in western Alaska. Based upon my research, there is an immense opportunity to use and combine various remote sensing tools to study lake dynamics and to evaluate associated environmental changes. Future work should be directed towards collaborative research for combining field-based observations and remote sensing tools to improve the understanding of how lakes and drained lake basins change in a changing climate as well as extend the scale of observations of methane ebullition features by covering many lakes in an environmentally diverse set of regions. This will guide us to understanding the feedback of lake dynamics to the surrounding ecosystem, global carbon budget, and to upscale the response of lakes to changing climate and permafrost environments to larger regions.