Spring 2023 - Project Abstracts
Austin Pruitt - Assessing Burn Severity of the 2018 Woolsey Fire and Determining Correlations between Pre-existing Vegetation Communities
The Woolsey Fire, beginning November 8, 2018, and lasting until November 21 of the same year, was one of the largest wildland fires in California’s recent history. Burning just under 40,000 hectares, the fire destroyed thousands of structures and cost an estimated $6 billion in damages. The California chaparral, a biome type adapted to fire, comprises a complex diversity of vegetative communities—15 existing within the perimeter of the Woolsey Fire before the fire’s ignition. To understand the elements of the fire more thoroughly, particularly the severity pattern of the fire as it relates to previously existing vegetation within the fire’s perimeter, a combined difference in normalized burn ratio and vegetation analysis using Sentinel 2 images and geospatial data from Cal Fire, respectively, was undertaken. More specifically, this analysis was conducted to determine which vegetation communities, as defined by Cal Fire, were more susceptible to burning and/or played a disproportionate role in increasing the fire’s severity.
Results showed that several vegetation communities were more susceptible to fire than others, including coastal oak woodland, valley foothill riparian, mixed chaparral, and chamise-redshank chaparral. These same communities also severely burned at a disproportionate rate. The numerical rates for these communities are discussed in the results section of this paper. Despite these findings, it is important to note that much more research is needed to understand the complex relationships between these disproportions as they relate to the fire landscape. Given the coarse nature of this analysis, factors that may also influence burn severity, including terrain slope and intervention by firefighting teams, were not assessed. As such, the findings and techniques of this analysis should be used to inform further research on the subject.
Emmanuel Takyi - Monitoring Woody Enchroachment in a Forest Savanna Mosaic
Studies on woody encroachment have historically focused on grasslands and savannas. However, the extent of this phenomenon in forest-savanna mosaics is unclear especially in protected reserves. Two Landsat 8 (OLI) imageries that span a 6-year period were used to assess whether large scale woody encroachment is occurring in Kogyae Strict Nature Reserve in Ghana, West Africa. Quantitative results indicate no incidence of woody encroachment but a widespread increase in savanna by 17%, and a significant reduction in tree cover by 27% across the landscape (p <0.05). For small scale detection of woody encroachment, active remote sensing techniques are recommended. Analysis of MODIS burned area product highlights the impact of annual increases in burned areas on forest loss, although seasonality in satellite data likley influenced the overestimation of the savanna expansion. The results also provide insights into the nature of fire regimes in the reserve and emphasize the need for a revision of fire management strategies especially in the protected and restoration zones.
Benjamin Mousseau - High Spatial Resolution Remote Sensing of Lava Flows in Visible and Near-Infrared Wavelengths
In this study, I explore the potential of using very high-resolution optical and very near infrared imagery from PlanetScope for the study of active lava flows, using the December 2022 eruption of Mauna Loa as a case study. The primary benefits of using this imagery are twofold. First, the higher spatial resolution reveals fine-scale features of lava flows, such as a higher thermal anomaly on the edge of the flow than in the middle, that are not visible from slightly coarser-grained imagery like Sentinel-2 data. Second, by combining images from Planet with traditional sources such as Landsat, Sentinel, and ASTER, a very high (often sub-daily) temporal resolution can be reached, allowing for better tracking of flow dynamics such as a new fissure opening or a flow path changing. Finally, this study demonstrates that Planet imagery can also be used for more quantitative analyses of lava flows such as temperature determination and creating a dual-component model for the flow, though it is less well-suited for these purposes than Sentinel-2 or similar satellites due to a lack of bands beyond the NIR. A single-component model yielded a temperature of ~970 °C for the hottest lava temperature at the fissure from which lava emanated, and a dual-component model assuming a hot component temperature of 1150 °C yielded a cold component of 847 °C covering 98.46% of the surface of the flow.
Bernard Nyanzu - Assessing Forest Cover Change Within an Agriculture Landscape in Western Ghana
Ghana’s forest has declined by more than 2.5 million hectares (Mha), or 33.7% since the early 1990s due to the establishment of tree plantations for local and industrial-scale production of rubber, coconuts, and Cocoa in the western region of Ghana has led to the fragmentation of the original rainforests. Thus, the agricultural plantation landscape is now considered an area of utmost importance for research focused on biodiversity and ecosystem service protection. To assess the fragmentation over time and the extent of agricultural encroachment, Two satellite images from 1999 and 2022 were used and classified to determine land cover change. Results from the classification indicated 5171.94 Ha (15.76%) of the forest area present in 1999 lost in 2022. 3521.42 Ha (10.9%) increase in land cover class classified as a farm area in 2022 as compared to 1999 and 1647 Ha, indicating a 72.5% increase in land classes of town/city in 2022 as compared to 1999. This suggests more forest reserve areas are changed and converted to agricultural production and rural development. Policies prioritizing forest health and conservation over agricultural inputs and income must be established and strictly enforced by the local and central governments.
Bibek Shrestha - Improvement of Water Clarity of Trishuli river, Nepal due to a Pause in Sand Mining during COVID-19 Lockdown: Elucidating the Impact of Sand Mining on River Water Quality using Remote Sensing
Trishuli and Budhigandaki rivers are tributaries of Gandaki River system, one of Nepal’s four major river systems. The two snow-fed, perennial rivers originate in the Himalayan range and drain mountainous river valleys before they merge near the northern threshold of Nepal’s Terai plains. Both the rivers flow through human-dominated landscape in the midhills of Nepal. Trishuli river has harbored sand mining industries for decades and the resulting impact on its water quality is evident as greater murkiness of its water compared to Budhigandaki at their confluence. The visual contrast is prominent especially in the lean flow season (winter to early summer). However, during the first few weeks of a strict lockdown enforced to curb the spread of COVID-19 pandemic on 24 March to 21 July 2020 in Nepal, an improvement of the clarity and water quality in Trishuli river was widely reported and was immediately attributed to the halt of often-unregulated sand mining operations in the river in the early phase of the lockdown.
The study examines this reported phenomenon using remote sensing techniques – Sentinel 2 images from before and during the first two weeks of the lockdown and Normalized Difference Water Index (NDWI; McFeeters, 1996) to distinguish the water bodies i.e. river network in the scene and Normalized Difference Turbidity Index (NDTI, Lacaux et al., 2007) to infer on the water clarity levels in the rivers in the time periods of interest. The results, especially the NDTI images quantitatively corroborate the reported improvement in water clarity of Trishuli river during the COVID 19 lockdown in 2020 in Nepal. This study also elucidates the potential for applicability of remote sensing starting in river water quality monitoring and for detecting/monitoring environmental impacts of activities such as unregulated riverbed mining especially in the context of limited capacity of the authorities to make on-ground inspections.
Eliza Poggi - Detecting Increased Coastal Runoff Following Hurricane Harvey
Sediment plumes, which have harmful implications for ecological and urban communities, result from the flux of terrestrial sediment into water bodies, especially after storms. Remote sensing is used to identify a sediment plume off of the coast of Corpus Christi, TX, that developed shortly after Hurricane Harvey in August 2017.The Normalized Difference Turbidity Index (NDTI), introduced by Lacaux et al. (2007), has been used previously to identify suspended particles rivers (e.g. Garg et al., 2020; Chen et al., 2022) and bays (Syed-Raza et al., 2022), but never before to image sediment plumes. Here, after processing Sentinel-2 satellite images from the hurricane aftermath and one year after the hurricane, the efficacy of imaging the sediment plume using NDTI is compared to that of using the Inverse Chlorophyll a Index. While both indices show relative change between the area encompassed by the sediment plume and surrounding regions, only the NDTI shows a substantial absolute difference between the sediment plume and the same area one year into hurricane recovery.
Dana Polomski - Using Remote Sensing for Detecting Lithologies and Minerals in Bou Azer, Morocco
Information about the Earth’s surface can be acquired from a distance using aerial or satellite-based sensors, which is known as remote sensing. This study seeks to identify a variety of minerals and lithologies in the Bou Azer region of Morocco using the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) and ground truth them against a geological map. This project attempted to identify minerals and lithologies using false color images, pseudo color images, and band ratios. Iron oxides were specifically targeted in one method, due to their applications in paleomagnetic research. In another method, band ratios and false color were tested as a method for finding valuable mineral deposits and compared to previously mapped mining locations. The ASTER images were processed in QGIS and then compared visually to a 1:50,000 scale geological map of the imaged region. The images, while fairly accurate, function more as an accessory to the geological map than as stand-alone sources of information from which to select sampling locations. Additionally, the remote sensing methods may be useful to identifying errors in geological maps that have been produced using field observations.
Eliana Stone - Crop Production and Pesticide Use in Mato Grosso, Brazil
Mato Grosso, Brazil is an important global source of agricultural production, which stimulates economic growth while also exposing its residents to concerning levels of pesticides. This project, firstly, utilizes the MapBiomas methodology to generate a dataset of land cover classifications for soybean, sugar cane, and cotton production and construct maps of crop change and transitions during the period from 2000 to 2010. This research then aims to access whether adoption of double cropping soybeans with sugar cane or cotton explains increases in pesticide use during this time, through quantifying the area of production for each crop and correlating these estimates to crop-specific estimates of pesticide use. Overall, soybean production and associated pesticide use explains the majority of pesticide use in Mato Grosso, despite changes to production of other crops.
Ben Everett-Lane - The Impacts of Cobalt Mining on the Democratic Republic of the Congo
Cobalt mining is a necessary industry for producing the rare earth metals critical for manufacturing lithium-ion batteries, which are used to store renewable energy output. Given that renewable energy is vital in order to transition away from fossil fuels and limit global warming to 1.5°C, this paper conducted an analysis to evaluate the impact of mining on land use in the region of greatest cobalt production: southern Democratic Republic of the Congo. Maximum Likelihood Supervised Classification, Change Detection, and Comparative NDVI analyses were used to examine the differences between Landsat 5 and 8 images of this area of the DRC, primarily around the city of Kolwezi, in 1993 and 2022. It was found that mine use increased by 67.84% across this 30-year period, accompanied by a loss in vegetation of 37.65%, which is supported by prior research (Brown et al. 2022). These results have profound implications on the projected impacts to people and wildlife as the mining industry continues to grow and the consequences of mining, which include habitat loss and pollution, are exacerbated. This leads to a debate of the trade-off between the benefits from this type of mining that aids in reducing fossil fuel emissions and the negative effects on the local environment.
Fredrick Gyamfi Addai - Galamsey: An Assessment of the Extent of Deforestation Caused by Gold Mining along 31.6 km of the Rive Offin, Using Remote Sensing
Ghana is the leading producer of gold in Africa, and gold mining contributes significantly to the nation’s economy. In 2020, gold mining accounted for US$ 1.3bn of Ghana’s GDP. Artisanal Small-Scale Mining (ASM) plays a crucial role in Ghana’s gold production. An estimated of over 100 million people worldwide depend on ASM. However, the recent gold rush in Ghana has led to an increase in illegal small-scale mining in the country. Illegal mining in Ghana has been the major driver of deforestation, water pollution, and general environmental degradation in recent decades. These illegal mines occur along the buffer strips of major rivers in the country, such as the River Offin. In 2017, the government of Ghana introduced a policy to stop illegal mining, but their effort faced strong opposition from local mining communities. The clashes between law enforcers and local mining community members have sometimes resulted in bloodshed.
Currently, the media in Ghana reports that the government has lost the fight against illegal gold mining in the country. This study uses remote sensing techniques (i.e., Change detection) to assess the change in land use within approximately 7,500 Ha along 31.6 Km of the River Offin. The studies indicated that a huge area (4,194 Ha) of the forest had been degraded by gold mining between 1991 and 2021. The studies also suggested that an area of 870 Ha of forest land was degraded by gold mining between 2017 and 2021. The studies conclude that illegal gold mining is the main driver of deforestation along the River Offin, and the government policy to stop the menace has not been successful yet.
Guillaume Hoffmann - Monitoring Glacier Recession in Zermatt from 1990-2020
Glaciers throughout the world have been receding in the last centuries (NASA). The objectives of this report are to perform change detection following supervised classification on a span of glaciers in Switzerland in Zermatt to better understand this melting of glaciers. Using Landsat 5 and 8 imagery from 1990 to 2022, a net loss of 14.386 square kilometers of glaciers was observed in the region of interest. This is a retreat of 0.486 square kilometers per year, equivalent to the area of 84 American Football fields per year. While this cannot easily be converted into a volume calculation, this amount of snow will inevitably end up in the Mediterranean sea, contributing to rising sea levels. Better ways to track this retreat and make predictions about its impact are discussed.
Jacob Peters - Creating a Mosaic of 1934 Aerial Imagery over New Haven, CT to Estimate Historic Tree Cover
Street trees provide many benefits to society, and as a result many cities and local governments are placing a high priority on increasing the number of street trees and green space. They mitigate pollution and stormwater runoff, provide habitat and food for urban wildlife such as birds and squirrels, increase property values, and help to mitigate climate change by removing atmospheric carbon dioxide. However, many cities are falling short when it comes to evenly distributing planting efforts across all neighborhoods, which may be a lasting impact from historic red lining. Here we attempt to create a mosaic of 1934 aerial images of New Haven Connecticut to train an object-based classifier using Google Earth Engine to identify trees. Our goal was to map and compare tree cover over time (1934, 1970, 2018) to observe restoration efforts following the outbreak of Dutch Elm Disease in the mid-1900s, and how historic tree cover compares with today. We expected that red lined neighborhoods would have had reduced tree cover over time and would have experienced reduced restoration efforts following certain outbreaks, and that these impacts may be long-lasting and observable today. Current iterations of our 1934 classification for New Haven, however, are not acceptable due to high error and will require more troubleshooting. In its current form, the first product of this work will be the mosaic of the 1934 aerial imagery for the city. This work will provide valuable insight to city planners and researchers seeking to understand more about historic red lining or landscape change over time in this area.
Makenzie Birkey - Redlining’s Legacies in New Haven: Surface Temperature Differences by Neighborhood
Redlining, a discriminatory method of mortgage appraisal by the Home Owners’ Loan Corporation (HOLC), created many disparities between communities in cities across the United States. These legacies have manifested as economic gaps between neighborhoods, including homeownership rates and credit scores, as well as climate risks. This paper investigates how redlining and the HOLC assessment system relates to present day disparities between summer surface temperatures in New Haven, CT. This was done through ENVI analysis of Landsat 8 Surface Temperature data from July 29, 2019, in New Haven, CT in conjunction with historic redlined maps of the city. Through the resulting statistics, it is apparent that New Haven neighborhoods with worse HOLC ratings experience higher summer surface temperatures, with a direct correlation between surface temperatures and rating. Neighborhoods with the best HOLC rating experienced average surface temperatures of 36.7ºC, and redlined neighborhoods, or those with the worst HOLC rating, experienced average surface temperatures of 41.2ºC, resulting in a 4.5ºC difference because of redlining. These higher summer surface temperatures pose a threat to historically redlined communities and represent climate injustice present in New Haven’s backyard.
Peyton Meyer - Using NDVI to Analyze Neighborhood Greenness in Bridgeport, CT
Neighborhood greenness is linked to a variety of benefits to residents. The present investigation utilizes the Normalized Difference Vegetation Index (NDVI), a validated metric for neighborhood greenness, to investigate levels of greenery among the thirteen neighborhoods of Bridgeport, Connecticut. Using a shapefile of local neighborhood boundaries paired with a Sentinel-2A satellite image from June 6th, 2022, the mean NDVI was calculated for each neighborhood. Because neighborhoods visually seemed to vary in the quality of their greenery, in addition to quantity, a classification of greenspaces was performed. Classification data was used to scale NDVI to create an adjusted index for neighborhoods. Statistical analyses, including hot spot analysis in ArcGIS Pro and Univariate Local Moran’s I in GeoDa, were performed to investigate the significance of the differences in the mean NDVI and scaled mean NDVI for each
neighborhood. Finally, income metrics were plotted against mean NDVI and scaled mean NDVI for each neighborhood to investigate correlations. Overall, findings indicated neighborhoods with high NDVIs tended to cluster together, as did those with low NDVIs. Further, neighborhoods in the north of Bridgeport, such as the North End and Reservoir, tended to have higher levels of greenness when compared to neighborhoods in the south (mostly clustered around the harbor), such as Hollow, Downtown, East Side, and West Side. A strong correlation of -0.819 was found between the low income rate in a neighborhood and scaled mean NDVI. The findings overall suggest that wealthier neighborhoods towards the north of Bridgeport tend to be greener than the less green, lower income neighborhoods towards the south and around the harbor, pointing at potential inequities for low income residents who lose out on the benefits of green neighborhoods that higher income residents tend to enjoy elsewhere.
Jasmine Gormley - Using Remote Sensing Techniques to Analyze Urbanization and Land Use Change in Western Montana, USA, Before and After the Onset of the COVID-19 Pandemic
In this project, I set out to analyze changes in land use in the area around Missoula, Montana to Flathead Lake, Montana, during July of 2018, 2020, and 2022. This was accomplished using LANDSAT-8 imagery and ENVI image processing software. The image from before the onset of the COVID-19 Pandemic, 2018, is used to establish a baseline level of urban development and other types of land use in western Montana prior to COVID-19. The 2020 and 2022 images then show the changes in land use and land cover that occurred during 2020 and in the years following the onset of COVID-19. There have been several studies on the detection of gentrification and changes in urban land use usung remote sensing, but I have found few on remote worker migration, or on this region of Montana (Lin et al., 2021). There has been little quantitative analysis of how the population of western Montana has been impacted by the phenomenon of remote workers moving to rural areas due to COVID-19, so I aim to fill some of this gap. There does appear to have been a variety of interesting changes in land use between these years which vary by city. Further study is required to determine if remote worker migration is the cause of these changes.
Vincent Haller -Exploring the potential of image texture for forest biomass modeling in Chile
This paper discusses the potential of using image texture as a predictor variable for forest biomass modeling in evergreen forests of Chiloe Island, Chile. For this, the Fourier transform based on textural ordination (FOTO) index is used to represent texture as a variable. The FOTO index is applied using a Fast Fourier Transformation in 2D and Principal Component Analysis on a Sentinel 2 and PlanetScope NIR image. The outcome of applying this index shows that higher resolution and smaller window size are preferred for biomass modelling, since hereby less topographical variability and more topographical variability is detected by the PCA. Consequently, there is potential for the FOTO texture index to be used as a predictor variable for forest biomass modeling.
Joshua Li - The Correlation between Antarctic Ecotourism and Urban Expansion in Ushuaia
Ushuaia is a crucial gateway to Antarctica due to its port being used as the main departure gate for Antarctic Cruise Ships. In this study we use Google Trends and IAACO data to track the popularity of Antarctic Ecotourism and compare that with the Urban growth of Ushuaia tracked through Sentinel-2 and VIIRS NightLight images from 2017-2023. Results show that there is a correlation between the direction of both Urban Growth and Antarctic Tourism between three periods: 2017-2019, 2019-2022, and 2022-2023.
Leah Clayton - Recovery of Alaskan Tundra Post-Fire in the Yukon-Kuskokwim Delta
Climate change and the resulting amplified Arctic warming are greatly affecting ecosystems and processes across the Arctic. The extent of wildfire in the tundra ecosystems of Alaska has been increasing dramatically since the turn of the 21st century and specifically impacting the tundra subregion known as the Yukon-Kuskokwim (YK) Delta in southeastern Alaska. While fires are growing in intensity and area, historical fire scars are present and have been identified in the YK Delta from 1953 to the present. This analysis uses Landsat 8 imagery from 2017 and a robust set of fire scar polygons for the region from 1953 to 2017 to analyze tundra recovery post-wildfire over time. The bands from the Landsat 8 surface reflectance data were used to calculate several indices that describe the vegetation and surface conditions of the region (NDVI, NDMI, albedo, and tasseled cap transform). Spatial statistics for each fire scar year were extracted and compared to unburned areas, and simple semi-log regressions were calculated to model recovery post-fire. Overall, this study concludes that tundra fire scars have historically recovered from fire over the course of the 65 years studied. The two vegetation indices used for analysis, NDVI and NDMI, support the conclusion that vegetation recovers to besimilar to that of unburned tundra within 45 years post-burn. While this analysis demonstrates recovery, it is critical to understand the historical context of the data and that there are large uncertainties about how tundra fire will change in the coming decades due to climate change.
Mingyu (Emily) Zhang - Detecting Coral Reef Bleaching due to Artificial Island Construction using Sentinel-2B
Coral bleaching events were identified in the southwestern region of Ocean Flower Island in the South China Sea. The findings revealed that a widespread coral bleaching event took place during the summer of 2020 in the studied area. By utilizing high-resolution difference imagery, it was observed that many corals in the area experienced bleaching between late July and early September 2020. The changes in the difference image and reflectance levels over time indicated the occurrence and progression of the bleaching event. It is suggested that coral reefs in this area started bleaching in July, and the severity of bleaching increased from the end of July until September. In October, the bleached area notably decreased, with the majority of corals recovering from the bleaching. These results demonstrate that the Sentinel-2B difference imagery is an effective tool for detecting coral bleaching.
Noah Weiner - The Dixie Wildfire: A Before and After Analysis Using Remote Sensing
The Dixie Wildfire of 2021 devastated a large area in northern California. It leveled a small historical town, burned cabins in a National Park, and forced thousands to evacuate. In this work, I use Landsat 8 images to examine the spectral trace left by the fire. Using various bands of the radiation spectrum, I compute vegetational health indices to visualize which parts of the image were most impacted by burning. I overlay the state of California’s official Dixie Fire perimeter shapefile to better display how the land within that perimeter changed. I also overlay data from the VIIRS Nighttime Lights remote sensing product to attempt to visualize town evacuations and recovery over time. I overlay elevation map data to explore the extend to which the fire shifted between high and low ground. I perform supervised classification on images from before and after the fire to better visualize which types of land were burned, to estimate the burn area, and to compare the spectral patterns of burned and unburned vegetation. I also use spectral indices to detect and highlight areas of smoke in Landsat images. I use some of my images to create layer stacks in ENVI and piece together animations of the fire’s development over time.
Oliver Leitner - Mapping Suburban Growth and Vegetation Change in Reno, Nevada
This study uses remote sensing to quantify and map land use change in the Reno, Nevada metro region over a 30 year period from 1984 to 2010. Performing supervised classification on four Landsat 5 TM images from this time frame revealed a 130 km2 increase in suburban land development, a 30 km2 decrease in adjacent forested area, and fluctuation in high NDVI vegetation such as agricultural fields, golf courses, and gardens. The increase in suburban development was found to correlate well with Reno’s dramatic population increase during this time. RGB change maps illustrate that nearly all of this development is outside a 3 mile radius from downtown Reno. The trends in vegetation could not be directly correlated to annual precipitation or snowfall data, although forest decrease could be due to declining Sierra Nevada snowpack levels. This study simultaneously highlights climate change symptoms and the influence suburban sprawl has on vegetation distribution.
Raymond Fedrick - Vegetation Cover Change Since the Protection of the Ballona Wetlands
The Ballona Wetlands are a highly degraded saltwater and brackish wetland, representing the last 2.4 km2 of this habitat remaining in the greater Los Angeles Area. Restoration efforts have been ongoing since the 1990s, including both major hydrologic alterations such as the installation of tide gates and more localized efforts made to reintroduce native species. Despite this 30-year long history of restoration efforts, the long-term impacts of this endeavor have been relatively poorly documented, with studies focusing on individual time points rather than looking at overarching trends. This project aims to begin filling in that gap by investigating the change in vegetation cover between 1989 and 2022. I used NDVI comparative analysis coupled with Isodata unsupervised classification to characterize the differential responses of the lowlands and highlands of the wetlands over time. I found that the highlands increased in vegetation cover between the two timepoints, whereas the marshy lowlands experienced a decrease in vegetation cover. I also observed a strong effect of rainfall on the highlands vegetation cover that was not reflected nearly as strongly in the lowlands. The Ballona wetlands provide a case study for the progress of vegetation growth in a protected, but relatively unmanaged habitat.
Ruiyan Huang - Assessing Changes in Open Water Area due to Urbanization using Satellite Data inWujiang, Suzhou
The investigation used Landsat satellite imagery to (1) quantify the distribution of open surface water across Wujiang, Suzhou, China in August 1998 and August 2022, (2) summarize spatio-temporal variation in open surface water between the two decades, and (3) assess factors influencing open surface water, including land-use change and infrastructure construction. The analyses indicated that between 1998 and 2022 open surface water has little change in total area, but experienced significant modification across 31% of total land area since 1998. Most of the land use change was due to the migration of aquatic farming from Lake Taihu to inland lakes and the construction of infrastructure around and beneath the water bodies. The data were mostly consistent with previous descriptions of water bodies in the region. Assessing water distribution using satellites allows us to understand impacts of changing water policy, aquacultural and agricultural practices, urbanization, and land use.
Sadie Bograd - Analyzing the Effects of Urban Service Boundaries on Urban Expansion in Lexington, Kentucky
Lexington, Kentucky, was the first city in the United States to implement an urban service boundary (USB) to limit urban sprawl onto agricultural land. In 1996, the county government expanded the USB significantly in response to concerns about housing affordability and economic stagnation. Although remote sensing has been used to study urban growth boundaries in other cities, no research has been conducted on Lexington’s USB since the early 2000s, soon after the 1996 expansion. This project explores the effects of both the initial boundary and its 1996 expansion, focusing on two overarching questions:
1. Has the USB been an effective tool in limiting urban sprawl in Fayette County?
2. How has development varied between the 1958 USB and the 1996 expansion, and does the USB need to be expanded again?
In this project, I use Landsat 5 and Landsat 8 satellite images to study Lexington’s development between 1995 and 2022. I perform a supervised classification to classify pixels as developed or undeveloped, then detect changes between the two images. I analyze urban expansion via two geographic comparisons. First, I compare urban expansion in Lexington’s urban service area (inside the USB) and rural service area (outside the USB). I find that there has been a higher rate of development within the urban service area, but that considerable development has also occurred in the rural service area. I then compare development inside the 1958 USB with development in the expansion area added in 1996. I find that the expansion area is still less built-up than the 1958 urban service area, but has developed considerably in the last two decades, suggesting both that Lexington has room to grow within the current USB and that the 1996 expansion responded to a legitimate desire for additional developable land.
Selin Goren - Quantifying the Damage Caused by the February 2023 (M7.8) Earthquake in Kahramanmaraş, Turkey using High-Resolution Satellite Imagery
The objective of this study is to quantify the building damage in the city of Kahramanmaraş, Turkey caused by the recent earthquake in the region (Mw = 7.8) using high resolution satellite imagery. A post-earthquake image taken on February 8th, 2023, by the PlanetScope satellite was acquired to use in the study. Supervised classification with Maximum Likelihood method was performed on the image using information from the 5 spectral bands (Blue, Green, Red, Red-Edge, and NIR) of the PlanetScope imagery. A damage map of Kahramanmaraş was produced as a result, differentiating between the damaged and undamaged areas. Calculations using class statistics revealed that 12% of the total urban area was significantly damaged due to the earthquake. This number was relatively similar to the official statistics released by the World Bank which conclude that 19% of the buildings in the area were highly damaged due to the earthquake. The results of this study indicate that high-resolution satellite images analyzed using a supervised classification algorithm can serve as a rapid response tool to identify significantly damaged areas immediately following an earthquake. In order to obtain a more precise damage assessment and identify the less severely damaged buildings, field surveys are needed.
Tyler Mar - Cracked Rice Bowl: Drought and Californian Rice Production
California is a major producer and exporter of rice. It produces nearly of the United States sushi rice and contributes more than $5 billion annually to the state economy. However, California’s rice bowl is cracking as exceptional drought and water curtailments imperil the state’s White Gold. Rice production in California is concentrated in the Sacramento Valley, where around 500,000 acres are produced annually. However, reduced water availability from droughts and curtailments is extracting a heavy toll as farmers idle production and fallow fields. Facing its driest three-year period on record, California rice production fell to a 30-year low alongside its benchmark rice price index reaching an all-time successive high. Hereafter, Landsat-8 images of the Sacramento Valley, where 95% of California’s rice is grown, are analyzed using supervised classification and Normalized Difference Vegetation Index (NDVI) to quantify changes in the distribution of total rice planted area and other land-use classes during the 2020-2022 drought. This study observed a 39% decline in rice acreage from 2021 to 2022 as farmers took fields out of production to conserve water.
Violet Low-Beinart - Quantifying Silvopastoral Conversion in the Azuero Peninsula, Panama
This article focuses on the implementation of silvopastoral conversion in Panama’s Azuero peninsula, where conventional ranching practices have led to large-scale land degradation. Silvopasture, an agroforestry practice integrating forage crops, pasture, and trees, offers numerous economic, social, and environmental benefits. This paper aims to quantify the on-theground effects of silvopastoral conversion using remote sensing techniques. This analysis utilized Planet Scope images in order to explore NDVI change, supervised classification, and classification ground truthing in order to quantitatively analyze land use change. Eight farms that had converted to Silvopasture (model farms) as well as seven farms that had retained their conventional practices (control farms) were analyzed for change from 2017-2020. Analysis was conducted for Jacob Slusser, the Environmental Leadership Training Initiative (ELTI) Panama coordinator, in order to provide ELTI quantifiable data related to silvopastoral conversion that could be provided to donors as well as current and potential silvopastoral farmers. Results highlight the success of silvopastoral conversion due to the increase in NDVI in model farms compared to control farms. However, the analysis encountered challenges in accurately classifying land use changes, highlighting the need for improved spectral resolution or extended timeframes of analysis in future work. Overall, findings underscore the current potential and challenges of remote sensing tools in understanding and quantifying silvopastoral practices.
Yuanyue (Mary) Yao - Investigating the Urbanization of Northwest Hong Kong
This project investigated urbanization of Northwest Hong Kong. Two Landsat images, one from 2001 and the other from 2023, were used. A shapefile was used to crop the image to only contain regions of Hong Kong. Classification was performed were made followed by various attempts to improve classification results and reduce misclassification. The final classification was by supervised classification with Maximum likelihood, with four classes: Urban, greenery, mountain, and water. RGB change detection was employed to visualize areas of urbanization. NDVI of the regions where also calculated and comparative NDVI was used to visualize the changes. Results indicated that urban areas have increased from 23.7% in 2001 to 27.2% in 2023, with a rate of 1.34 km2/year. These values match official land area analysis results. The overlap in new urban areas with decrease in NDVI suggests that deforestation was a major cause for urbanization in Hong Kong. Possible sources of error and potential future developments for this project was discussed.
Pranik Chainani - Remote Sensing Phenology and the Effect of the Recent War in Syria using NDVI Time Series Data
This project aims to investigate changes in vegetation phenology using a custom Normalized Difference Vegetation Index (NDVI) time series dataset from 2015-2016 across the Syria area. Specifically, the impact of the recent war on vegetation phenology will be examined by comparing NDVI time series data in relation to population density. Remote sensing data from the MODIS instrument onboard both the Terra and Aqua satellites will be used, and time series analysis techniques will be applied to calculate the start, end, and length of the growing season. The results of this study will provide valuable information on the impact of the war in relation to the environment in Syria, and will be useful for developing strategies for post-conflict reconstruction and rehabilitation efforts. Overall, this project highlights the potential of NDVI time series data and remote sensing techniques for monitoring changes in vegetation phenology in complex and dynamic environments.
Alaysia Navor - Analysis of Snowpack Change and Glacial Retreat, 1986-2013 Wrangell-St. Elias National Park, Alaska
This study analyzes the change in snowpack coverage and glacial retreat in Wrangell-St. Elias National Park, Alaska from 1986 to 2013. ENVI software was utilized to process and analyze satellite imagery from Landsat 5 TM from June 12, 1986, and Landsat 8 OLI/TIRS from June 22, 2013. To visualize the pattern of snowpack change and glacial retreat in the Wrangell Mountains over this 27-year period, Normalized Difference Snow Index (NDSI) was utilized, and comparisons were made between true-color RGB-432 and RGB-321 representations, as well as polygonal shape outlines of the snowpack-glacial areas. Pixel images were classified through the supervised classification method of minimum distance and class change statistical analysis was performed to quantify the changes in surface coverage over time. The analysis of class changes between the two images suggests that snowpack and glacial coverage in the area decreased by 48.09% (3840.32 square km) from 1986-2013.