Spring 2020 - Project Abstracts

Sangam Paudel - Forest firests and land cover change in Nepal

In Nepal, forest fires occur every year in the dry months that precede the South Asian monsoon. With the help of remote sensing tools and images, this project analyses the effect of forest fires in Kanchanpur district while lies on the plains of Far-Western Nepal. Using Normalized Difference Vegetation Index (NDVI) and Difference Normalized Burn Ratio (dNBR), this project will examine the immediate effects of fires on land cover and the persistent effect of the fire on the vegetation in the following year. Results from the combined NDVI and dNBR approach in the forest in Kanchanpur shows that while forest fires led to decreased vegetation immediately after the fire, the effect on vegetation a year down the line was not significant. Although the specific results of this study are site-specific, it does show that a combined NDVI-dNBR approach is helpful in understanding fire dynamics in Nepal’s forests. Complementing remote sensing data with local ecological insight will further enhance our understanding of forest fires.

Amelia Han - Land cover change in Newtown, CT

This research focus on analyzing the land cover change process in Newtown, CT area from 2000 to 2010 by using the Landsat 5 TM satellite images. The land cover types are simply classified as five different informational classes: water, bare soil/rock, coniferous tree, deciduous tree, and forest. Land cover change results over ten years can be concluded using various change detection methods in ENVI, including transition table/line chart, change detection statistics table, and RGB change detection images. Urbanization and new agriculture trends are also easily recognized in the satellite images. This project indicates that the remote sensing techniques can be really helpful for the city management, for learning the impacts of population growth and housing development on the land use change. The results show that population growth may have impacts on the land cover change of vegetation degradation; however, this change seems to be reversible. From 2000 to 2010, there were more areas experiencing urbanization than new agriculture in the
Newtown area. Finally, the development of urbanization in Newtown tends to be radial and scattered. Future research is needed to be conducted to improve the accuracy of the supervised classification by doing the field trip to find the ground truth data as training data. It would also be better to assign more land cover classes for more detailed land cover types. In addition, the whole project can be done for another place where the population growth rate is much bigger than Newtown, in order to future illustrate whether or how the population growth of a place can impact its land cover change.

Sherry Xu - Using remote sensing to study desert “superblooms” in California

Superblooms are intense greening and flowering events in places that are usually arid. Though they are probably best known for occurring in the Atacama Desert of Chile, superblooms have recently occurred in California, in 2017 and 2019. These blooming events were subjected to much tourism and press coverage, and were so widespread that they could be visible from many satellite instruments in space. In this study, I investigated a variety of methods to determine if the superbloom could be recognizable in satellite imagery via a unique spectral signature. I tested the Normalized Difference Vegetation Index (NDVI), the Enhanced Bloom Index (EBI), and an unsupervised classification method, ISODATA, and evaluated how useful each method was for identifying superbloom areas. Ultimately, I concluded that, though it is not perfect, EBI was the most accurate and useful for identifying areas with blooming flowers. Additionally, supervised classification has potential, especially if robust training sets are utilized.

Minnie Min Ying Li - Urban change detection in New Haven, CT usinng declassified imagery

Abstract not available.

Anya Lomsadze - Change analysis surrounding the Merowe Dam, Sudan

The construction of dams on the Nile has been and continues to be a fraught, controversial matter. As the region sees its largest hydrological construction in its history develop in Ethiopia, to be known as the “Grand Ethiopian Renaissance Dam,” it is worth looking at the other dams in the region to understand the impacts of such an undertaking. This paper discusses the Merowe Dam in northern Sudan, which was opened in 2009 with controversy from humanitarian to ecological complaints. Using supervised classification, soil moisture analysis using the NDMI index, and NDVI analysis, we try to understand the spatial impact of the dam by comparing three Landsat images taken before and after the construction of the dam. In particular, we look to see how the dam affected land-use, settlement and dispersal patterns, and soil moisture for agriculture in this region. This study shows that the construction of the Merowe Dam and the filling of its reservoir significantly altered the region surrounding the dam and the upstream region, though not the downstream region, particularly reducing farmable land. We observe a sprawl like pattern of habitation caused by the displacement from construction and an increase in soil moisture around the reservoir. The patterns in land-use change surrounding the Merowe Dam could be considered in the construction of future projects to minimize harmful changes with dam openings.

Colin Baciocco - Assessing changes in perennial ice cover on Baranof Island, Alaska

Changing climates may cause Southeast Alaska, one of the rainiest non-monsoon receiving areas of the world, to spend significantly more time in a state of drought, as well as receive less precipitation overall. In such a scenario, snowpack becomes increasingly important for maintaining water levels in economically crucial bodies of water, like hydroelectric reservoirs and salmon streams. A pattern of increasingly severe recent droughts support the likelihood of this valuable-snowpack scenario. Increasingly accurate inventories of snowpack will likely become necessary for the Pacific Northwest and Alaska as the climate continues to
warm.

Accordingly, this study assessed the viability of using Landsat images to assess changes in the area-at-minimum-extent of snowpack on Baranof Island, a mountainous island on the outer coast of the Southeast Alaskan archipelago. Four cloud-free Landsat images of Baranof, taken at different points over a thirty-year period, were chosen for analysis. Unfortunately, difficulties getting appropriate scaling parameters for the higher bands of two Landsat sensors meant that accurate TOA reflectances in the near IR were never obtained, and that accurate ice indices were calculated or usable. Supervised Maximum Likelihood Classification was, however, also run on the scenes. Two of the classified scenes were then compared to assess changes in snowpack area between the start and end of the thirty-year period, as well as changes in their snowline. It was difficult to have confidence in the accuracy of the changes in snowpack area found using Supervised Classification, or the patterns of snow-line change also found, but further refinement of this study’s methods – particularly in their combination with relatively new high-resolution DEMs – holds promise for reliably assessing snowpack changes in both Southeast Alaska and around the world.