Spring 2019 - Project Abstracts
Britta Dosch - M&E verification of silvopastoral capacity building programs in Panama using 4 vegetation indices
This project compares land cover change in the Azuero Peninsula in Panama between the years 2010 and 2019 with a specific focus on land parcels owned by past participants in the Environmental Leadership Training Initiative’s (ELTI) capacity building programs. Four vegetation indices were utilized to analyze land cover change: NDVI, EVI, OSAVI and MSAVI2. The project objectives are twofold: to determine if remote sensing is a viable tool to use in monitoring and evaluation of ELTI’s effectiveness at capacity building in the Azuero Peninsula and, if so, to quantify the change in vegetation cover within the participant land parcels using skills and knowledge gained from F&ES 726 Observing Earth from Space. All four indices show a quantifiable increase in vegetation cover in the ELTI parcels and in the wider Azuero Peninsula region. However, limitations with scale emphasize the need for ground-truth data for a study site of this size.
Naoya Orita - Leakage from REDD+ Project: Case study within the Madre de Dios region of Peru
In implementation of REDD+ Projects, Leakage is an important issue to be considered. Leakage is defined as the net increase of anthropogenic greenhouse gase (GHG) emissions of which occurs outside of the project boundary. In this study, leakage in a REDD+ project in Tambopata National Reserve and Bahuaja-Sonene National Park of the Madre de Dios region was calculated by using Landsat images from 2007 to 2018. The analysis revealed that the deforestation in the leakage belt increased significantly from 2010 project commencement to 2018 while it remained relatively moderate in the project area. In the leakage belt, deforestation mainly because of illegal mining activities expanded rapidly from 2010 and spread along the boundary of the project area. This result probably implies that the deforestation expansion into the project area was prevented by the project implementation. Although there were significant increases in deforestation in the leakage belt, deforestation in both the project area and the leakage belt remained below prepared baseline scenario. The baseline scenario in the leakage projected rapid and significant deforestation expansion from 2010 to 2014 and gradual decrease after 2015. The actual deforestation showed a contrastive path of the significant increase in deforestation year by year. If not managed appropriately, deforestation in the leakage belt will exceed the baseline scenario before the project implementation will be completed in 2030. In management of the target area, it would be important to address increasing illegal mining activities that have been increasing despite various policies taken by the Peruvian government against them.
Aidan Mock - Palm oil-linked deforestation and forest degradation in Sabah, Malaysia
Palm oil plantations have expanded rapidly across Southeast Asia as demand for the crop has soared. Sabah is one of the hotspots for palm oil-linked deforestation and the Roundtable for Sustainable Palm Oil (RSPO) has emerged as an organization attempting to push palm oil producers to more sustainable practices. In this study, I quantify palm oil expansion in northeastern Sabah between 2000 and 2018. I use a Landsat 5 image from 2000 and a Landsat 7 image from 2018 for my analysis. I find that 1,183 km2 of forest has been converted to palm oil plantation in this time period, almost all of which was originally primary or coastal forest. I also find that RSPO does not appear to be effective in reducing deforestation or the impacts of palm oil. Government-defined protected areas are effective in conserving forest habitat, however these protected forests appear to suffer from degradation of inner vegetation when bordered by palm oil plantations.
Rahul Nagvekar - Land use change in Zimbabwe’s Shangani River Valley during the Gukurahundi Massacres
The Gukurahundi massacres were an episode of state violence against the Ndebele people of western Zimbabwe between 1983 and 1987 that involved an estimated 20,000 killings of civilians by army personnel as well as widespread rape, beatings, torture, displacement, and intimidation. Many victims of Gukurahundi were subsistence farmers, and the massacres likely had severe impacts on small-scale agriculture in predominantly Ndebele areas, such as the Shangani River valley in Matabeleland North province. This project sought to identify land use change suggestive of farm abandonment in the Shangani valley between the 1984-85 and 1986-87 growing seasons using Landsat 5 MSS imagery.
Unsupervised classification indicated some land cover change from farms to natural vegetation in the Shangani valley between February 24, 1985 and January 29, 1987. In addition, between these dates, mean NDVI decreased by a greater amount for pixels classified as farms in 1985 than for pixels classified as natural vegetation in 1985. However, environmental factors such as year-to-year or seasonal variations in precipitation cannot be ruled out as potential causes for this change. While this project did not provide definitive evidence of farm abandonment in the Shangani valley between 1985 and 1987, further analyses that apply the same or improved methods to larger areas or different timespans could potentially help establish a link between Gukurahundi and concurrent land use changes in western Zimbabwe.
Mingmin Feng - Regional development assessment after connection to the highway in Motuo County, China
Surrounding by high mountains and limited by geological conditions, Motuo County in Tibet Autonomous Region was the last county connect to the highway. The highway linking Medog County and Bome County was completed in October 2013. This project used RapidEye image from 2009 to 2017 to assess the land cover change of Motuo County before and after the highway connection. Normalized Difference Vegetation Index (NDVI) was calculated to detect the vegetation change in this period. The land cover was classified into four classes: development, agriculture, forest, and grassland. This project also analyzed the spatial and temporal pattern of local development in Motuo County. Results show that the development area continued to expand from 2009 to 2017. Grassland was the major source of the new developments. The expansion rate of development was slower after the highway connection, but the new development was more concentrated and the scale is larger after 2013.
Mads O’Brien - Vegetation and mineral alteration near Mammoth Geothermal Complex, Mono County, CA
Mammoth Geothermal Complex (MGC) is a geothermal power plant assemblage located on the margin of Long Valley caldera, east of California’s Sierra range. Three power plants, operated by Ormat Technologies, utilize hot water in the hydrothermal system beneath the caldera to generate 29 Megawatts of power for approximately 22,000 homes in the adjacent town of Mammoth Lakes. In 2013, Ormat was approved to develop an expansion of their current projects, a plant called Casa Diablo IV (CD-IV), which would generate an additional 30 Megawatts at the complex. The company has already been drilling exploration wells in the western part of the geothermal area as part of this development. Because of how geothermal power generation works, any geothermal fluid that is tapped for the electricity generation process is then injected back into the subsurface reservoir. Though Ormat has conducted extensive safety testing on the CD-IV expansion, especially its effect on local drinking water supply, it is important to study how injecting wastewater back into the geothermal reservoir might affect fluid release elsewhere in the caldera. If a past or previous development project has disrupted the hydrothermal system and caused surface changes before, then the plant’s infrastructure may warrant a second look. Assessing the ecosystem impacts of tree die-off and cheat grass proliferation is beyond the scope of this report; instead, this study attempts to answer:
• What vegetation and mineralogical changes occur around MGC over time?
• How well may changes be detected with remotely sensed imagery?
• Do vegetation and mineral surface changes correlate with development at MGC?
Zach Gold - Change detection during a historic drought in Kruger National Park, South Africa
In 2015-2016, a historic drought occurred in Kruger National Park, South Africa. This project used remote sensing to evaluate change in the park before, during, and after the drought. Using MODIS data, I calculated mean annual land surface temperatures for every pixel within the park as well as for the park as a whole for two years prior to the drought, the two years of the drought, and two years after the drought. I also found the difference between the land surface temperatures in the park for the time periods 2013-2016 and 2016-2018. Additionally, I used
MODIS data to calculate mean annual NDVI for every pixel within the park as well as for the park as a whole before, during, and after the drought. I also did the same analysis of NDVI
restricted to the dry seasons of each year included in the study. I then found the difference between NDVI in the park for the time periods 2013-2016 and 2016-2018. Finally, I performed a
supervised classification of Landsat 8 images from June 2013, 2016, and 2018 in order to determine the change in relative abundance and location of dense tree cover to open savanna.
There was substantial increase of land surface temperatures in KNP during the drought, particularly in the eastern half of the park, followed by moderate recovery in the years after the
drought. Similarly, there was dramatic decrease of NDVI in the eastern half of KNP during the drought followed by substantial recovery. The supervised classification did not provide any
conclusive evidence pointing towards effects of drought with respect to change in amount of dense tree cover in a subsection of the park. Overall, this project showed that remote sensing is
an effective tool to monitor impacts of a drought on a large scale in savanna ecosystems.
Charlotte Stanley - Drought impact on crops in the Cape Winelands
The drought in the Western Cape of South Africa, from 2014 to 2018, altered residents’ lifestyles, damaged crops, and reduced national GDP. The impact to one valuable sector—the wine industry—can be analyzed through remote sensing analysis of satellite images. Studying changes to wine grape crops during the drought years can give insight into the crop’s response to drought, to support future sustainable management strategies. Since the region contains many vegetated land types, other croplands and forests were analyzed in addition to vineyards. The study looked at a large extent surrounding the municipalities of Stellenbosch and Drakenstein, as well as a targeted region of interest filled primarily with vineyards. The analysis used Landsat 8 imagery, MODIS Land Surface Temperature (LST) data, and MODIS classification data.
Multiple methods were implemented to study change detection from 2014 to 2018: Normalized Difference Vegetation Index (NDVI), Normalized Difference Moisture Index (NDMI), Tasseled Cap brightness index, land surface temperature, supervised classification, and MODIS change detection. The results section describes the outcomes of each technique analyzed, through change detection maps and graphs of rseults. Different methods of analysis led to differing results. NDVI decreased and LST increased during the study period, showing results consistent with the drought conditions. NDMI and brightness resulted in no significant changes, and occasionally responses conflicting with traditional drought trends. Each of the indices displayed significant variation when analyzing specific farm plots. The classification method used resulted in errors due to limited data. The discrepancy across methods requires further study to acquire more accurate results that better indicate drought impacts. Ground truthing and local data collection could provide useful information to better understand the specific species, their responses to droughts, and variations on a farm level.
Daniel Monteagudo - Observing vegetation loss in Cuba due to Hurricane Michael using Sentinel-2 imagery
Hurricane Michael formed and moved through the Caribbean in early October 2018, passing over the western portion of Cuba on October 8th. With the storm came heavy rains, flooding, and high winds, bringing crop losses and the destruction of structures to the western provinces of Pinar del Rio and Isla de la Juventud, according to news reports. This is one of the principal regions in Cuba for tobacco farming, one of the country’s main exports. This project aims to reveal and quantify the extent of this damage using Sentinel 2 imagery from immediately before and after the storm. After normalizing the images with cloud masking, change detection analysis was performed using supervised and unsupervised classification and by looking at NDVI change.
The NVDI analysis resulted in the most concrete results, showing that 79% of the study area experienced a decrease in NDVI after the storm. Visual inspection of the NDVI change showed that the most significant loss was constrained within farm plots, rather than a general loss across the board. This suggests that the most significant damage to vegetation may not have
been due to high winds stripping away leaves, but rather water damage forcing farmers to replant seed beds. These high damage zones, with an NDVI loss greater than 0.3, comprised a land area of 1,987 sq km. In contrast, only 530 sq km worth of land experienced a similarly intense (>0.3) increase in NDVI. Future research could benefit from more work on training a useful supervised classification, and from masking away cloud shadows, possibly through this classification.
Danielle Losos - Multi-temporal analysis of inland flooding following Hurricane Florence in the Carolinas
I detected flooding patterns caused by Hurricane Florence in North and South Carolina by evaluating changes in soil moisture over a ten-day period. I conducted a multi-temporal analysis by stacking 27 soil moisture images trimmed to North and South Carolina, spanning the ten-day period and collected 9 hours apart. The images were from the L4 Surface Soil Moisture product of the satellite SMAP, chosen for their high temporal resolution and active radiometry. To delineate zones of flood severity, I determined the change in soil moisture overtime between the minimum and maximum values and classified pixels accordingly. The zones of greatest flood severity were located in in southeastern North Carolina, and northeastern South Carolina, and were not necessarily the regions with the greatest overall soil moisture levels. The regions of greatest flood severity were also proven to have the fastest flood rates, or changes in soil moisture.
Devon Ericksen - Remote sensing of April 2018 Kauai floods
This project used remote sensing to detect the impacts of the April 2018 record-breaking rainfall on the island of Kauai, Hawaii. The project focused on the north shore of Kauai (Images 2 & 3), specifically the town of Hanalei and the surrounding agricultural and forested region. The project used the Planet satellite’s high-resolution imagery in order to detect flooding and landslide impacts in areas around the town of Hanalei. The focus on the north shore and particularly Hanalei came from news reports detailing the hardest-hit zones following the storm.1 The project used Normalized Difference Vegetation Index (NDVI) change detection to quantify the impacts of silt deposition that occurred from landslides and flooding in the study scene. Additionally, the project used the FEMA Flood Hazards Area map (Image 4) for the region to determine if flooding occurred outside of FEMA’s designated flood zone boundaries, and if so how much. The project found that of the 3.4km2 of impacted areas, 1.75km2 was outside of FEMA’s flood zones, which was 51.4% of the total flood impacts in the scene. This indicates a need to update FEMA’s maps, especially considering the predicted increases in storm frequency and intensity if current climate scenarios take place.
Trina White - Mapping ice cover on the Cordillera Blanca mountain range in Western Peru
The Cordillera Blanca mountain range in Western Peru is home to the highest and largest stretch of tropical glaciers in the world and serves as an essential water source in the region.
Glacial meltwater functions as a storage mechanism for freshwater from the wet to the dry season. Recently, rapid melt of the Cordillera Blanca has increased access to dry-season
freshwater and induced greater agricultural productivity in the region. Continuous monitoring these glaciers is important as the decline in these resources threatens future water access.
Remote sensing has the potential to be a vital tool for mapping these glaciers through time. This study analyzes several methods for mapping glaciers by their spectral signatures, including the
ratio of NIR to SWIR reflectance and the Normalized Difference Snow and Ice Index (NDSI) and threshold. I use these methods to estimate the decline in surface area of the Cordillera Blanca
glaciers from 1987 to 2018. I find inaccuracies in both methods and propose the use of a combined threshold. I also quantify NDVI increase in the surrounding region as an indicator of
the increased dry-season productivity resulting from glacial meltwater.
Nadia Grisaru - The effects of black carbon from wildfires on glacier surface albedo and ablation
The 2018 wildfire season on the West Coast of North America was one of the most destructive in recent history. A warming climate leads to more frequent droughts and higher summer temperatures, resulting in increasingly destructive wildfires. Such fires are part of positive feedback loops that fuels further climate change. Understanding how is crucial for predicting how the planet will respond to present and future human-driven climate change. The surfaces of glaciers that are proximate to wildfires are often dirtied by the soot, resulting in decreased surface albedo and increased ablation. The Columbia Icefield lies in the Canadian Rockies in the path of dark particulate matter carried from the British Columbia and South East Alaska wildfire zones. Glacier surfaces were observed to be darker that normal following the 2018 wildfire season.
In order to determine if 2018 surface albedo levels were lower than 2011 values, a year with a relatively mild wildfire season, two Landsat images from the end of the wildfire and ablation seasons each year were analyzed. Surface albedo was calculated for the Columbia Icefield for each year, and supervised classifications were performed to determine the extent of snow cover loss and glacier thinning and retreat between the two years. This analysis provides valuable insights into the effects of wildfires on nearby glaciers and the value of satellite-based imagery and remote sensing to calculate and monitor glacier albedo and ablation.
Ulla Heede - Outgoing longwave radiation and tropical clouds
Outgoing longwave radiation (OLR) is a commonly used proxy for studying precipitation and convection in the tropics, as tall convective clouds with high altitude-tops are usually associated
with low emission temperature and corresponding OLR. This project aims at exploring the properties of such low OLR clouds in more detail. By utilizing the MODIS level 2 suite of cloud
products, the link between OLR and other properties of tropical clouds are statistically quantified, as well as qualitatively analysed across a selection of images. Three lines of analysis
are carried out. First, clouds with a low OLR signature are isolated across all participating images, and their optical and physical properties are described. Second, correlation between
OLR, and optical and physical cloud properties for each entire image is calculated. Lastly, qualitative analysis for a subset of images is performed. Overall, this study finds that while
clouds with low OLR signature have a relatively constrained cloud top height and temperature range, the column water path and optical thickness of these clouds vary vastly. Furthermore, the
image wide correlation between optical thickness, water column path and OLR is poor (<0.4), and the general correlation between OLR and cloud properties varies depending on the
heterogeneity of the cloud types in a given image. Overall, it is found that OLR may be a poor or suboptimal proxy for identifying convective clouds in the presence of cirrus clouds. These
findings shed light on the suitability of OLR as a proxy for convective clouds as well as general features of tropical clouds.
Hoon Pyo (Tim) Jeon - Estimating ground-level NO2 concentrations over the Korean Peninsula using Sentinel-5P data
In recent years, increasing levels of air pollution in South Korea has become a serious environmental issue, having enormous social, public health and even diplomatic consequences.
In order to effectively manage air pollution and successfully design, implement and target environmental policies, it is crucial to establish a precise quantitative estimation of the
concentration levels of major air pollutants. Although the ground-level measurements are often accurate and relevant for practical purposes, these measurements also suffer from limited
coverage, irregular distribution and high economic costs. Therefore, it is important to identify and develop new methods, including remote sensing techniques, to reliably measure air
pollution. As part of these efforts, this paper aims to estimate ground-level NO2 concentrations over the Korean Peninsula using the data collected by Sentinel-5P, the latest and most advanced satellite for atmosphere monitoring. In particular, we focus on the Seoul Metropolitan Area, using extensive data from twelve different dates in April 2019 and 25 ground-level measurement points spread across the districts of Seoul. Our results reveal that Sentinel-5P satellite data has a significant predictive power in estimating ground-level NO2 concentrations, indicating that remote sensing could effectively supplement ground-level measurements of air pollutants.
Alan Fairbank - Then and now: Documenting urbanization of Surket Valley from 1976-2018
A remarkable case of rapid urbanization (from a dispersed, roadless rural area to densely populated city within 50 years) was identified in Nepal. Knowledge of the origins of the town of Birendranagar in Surkhet Valley in far western Nepal (and photographic evidence thereof) derives from my Peace Corps service there (1967-1972). Documentation of ground truthing was made possible by reference to those photos from that period and was supplemented by recent photos of that town. Thereafter, multiple Landsat images were located that provided high-resolution images of that valley—images that could document the transformation of the valley from one totally devoted to rice cultivation to one partially occupied by the roads and buildings of a growing city.
This paper describes and documents that transition of rural, sparsely populated Surkhet Valley (1967), into an area dominated by a rapidly growing city of 60,000 (2019). Surkhet Valley is an oval area of fertile rice paddies at about 2,000’ elevation that is about 10 km long and 5 km wide, and is located about 600 km west of Nepal’s capital city of Kathmandu. The new town, Birendranagar, was planned and its city streets laid out in 1966-1967 on the upland elevations on the northern side of the valley. It is surrounded on all sides by foothills of the Himalayas—ridges reaching heights of between 1,200 m and 2,600 m. While construction of mud and stone houses proceeded in the late 1960s and 1970s, with post-and-beam frames constructed from trees of neighboring forests, the rapid, modern expansion of Birendranagar did not really begin until after a road from the Indian border was constructed in 1982-1983. (Until then, all commercial merchandise was either flown in or carried up from Nepalganj in a four-day trek.) Now, the city has all the features of a modern, urban area, including a university, a convention center, several hospitals, and many hotels and restaurants.
Chengcheng Qiu - Vegetation coverage around Hutong and modern residential neighborhoods in Beijing
Hutongs are the narrow alleyways that formed between traditional courtyard houses in Beijing, China. They crisscross the traditional districts like veins, and create an intimate and genuinely idyllic atmosphere for the local communities. However, there has been a constant conflict between modernization of Beijing and the preservation of its cultural heritage. Since mid-20th century, Hutong has been continuously demolished or redeveloped into high-rise buildings, and rarely any historic Hutong quarters have remained intact. It has become important for people to understand the unique values in these traditional districts and improve the livings in the historical residential areas by lending support from scientific research methods.
This study focuses on the application of remote sensing skills on examining the vegetation distribution, thermal environment and their relationship in the traditional Hutong neighborhoods, and comparing the conditions in traditional neighborhoods with renovated neighborhoods in the inner city, which is defined by the boundary of the second ring road of Beijing. High-resolution PlanetScope satellite images was used to calculate NDVI and show the distribution of vegetation across the whole study region. Landsat 8 images was also used for calculating land surface temperature, which was later fitted into linear regression models to see its association with vegetation abundance. Findings have suggested that the traditional neighborhoods have higher abundance of vegetation in summer while less greenery in winter when compared to the renovated neighborhoods, which may be due to the seasonality of specific vegetation types in these places. The land surface temperature in both summer and winter is higher in these traditional residential areas, and is likely to concentrate in specific areas like Da Shi Lan area and Chang Qiao area. The amount of vegetation has negatively affected the land surface temperature, and the percentage of highly vegetated area like tree canopies has the strongest influence on lowering land surface temperature.
Lauren Stoneburner - Detecting urban tree canopy cover change in Portland, OR
These analyses seek to detect urban tree canopy (UTC) change between 2009 and 2016 in a study area within Portland, Oregon. The report will evaluate the accuracy of two methods for detecting UTC change: change detection of Maximum Likelihood Classification and Normalized Difference Vegetation Index (NDVI) change, both based on National Agriculture Imagery Program (NAIP) data with 1-meter resolution. I had hoped to compare tree canopy cover to socioeconomic variables or land use zoning regulations. However, in the end, the results are inconclusive and would not do justice to the landscape if compared with zoning or social vulnerability metrics. The accuracy of change detection was limited by the prevalence of shadows and how they differed between the two images. This report will discuss these challenges, as well as potential methods of overcoming the barriers that shadows present to image classification and NDVI change, with the ultimate hope of improving future researchers’ ability to use NAIP data for UTC change detection.
Yihao Xie - Urbanization in Lanzhou New Area, China
Rapid urbanization in China has led to vacant and underutilized urban residential, commercial and industrial spaces, a phenomenon called “ghost cities”. This project uses satellite remote sensing images (Landsat 8 OLI/TIRS and Visible Infrared Imaging Radiometer Suite), applies tools and methods, including Daytime Urban Heat Island effect, Night Light Intensity, and Normalized Difference Vegetation Index, to analyze if Lanzhou New Area in Northwestern China could be considered a “ghost city”. Results from this project show that Lanzhou New Area had grown and developed in the past six years by all three metrics. While no conclusion can be drawn regarding Lanzhou New Area’s status as a “ghost city”, remote sensing images and analytical tools used in this project are found to have great potential for research on urbanization and the “ghost city” phenomenon.
Sally Goodman - Mapping water features and wildlife in Laikipia County, Kenya
As increased rainfall variability and growing use of river water in agriculture reduce river flow in Kenya’s Laikipia County, the importance of additional water points for the diversity of
wildlife and livestock in the region’s arid and semi-arid lands (ASALs) becomes more evident. In order to ensure sufficient water resources for animals and maintain rangeland ecosystems, it is
important to understand the status of water points in the region and how animals distribute around them. This study compares methods for detecting water including modified normalized
difference water index (MNDWI) and unsupervised and supervised classification, and considers how water availability, vegetation, and other factors relate to wildlife distribution. I found that
supervised classification is most effective for identifying water features in Laikipia. Additionally, NDVI appears to play a larger role in the presence of wildlife than nearness to a water point.
Zhi Yi Yeo - Understanding coral reef stressors on urban reefs along the southern shores of Singapore
Coastal development over the world has placed increased stress on coral reefs, especially reefs that are situated near urban agglomerations and areas of intense coastal developments. Singapore’s urban coral reefs are an example of such reefs under threat, as human activities such as land reclamation result in increased levels of stress on coral reef ecosystems around the island. While much research has been done on these reefs, large-scale understandings of the spatial distributions of reef stressors around these the island is still limited, prompting a need for research to identify spatial patterns and distributions of these stressors. This study employs the use of images taken by the Landsat 8 OLI and TIRS sensors to derive values of three known coral reef stressors – chlorophyll-alpha concentrations, total suspended matter concentrations, and sea surface temperatures along the southern shores of Singapore. Starting with the digital number (DN) values of the Landsat 8 Level 1 scene, a series of algorithms were used to remove the effects of atmospheric interference and interference due to the sea-air interface before converting corrected reflectances to values representing physical attributes. The results showed evidence of impacts of human activities on these physical attributes, creating conditions that place increased stress on coral reef ecosystems. Based on comparisons with other studies, the results from this analysis is likely to be reliable. However, in-situ data is still required to further validate the findings. This study present an exploratory analysis of the spatial distribution of coral reef stressors off the coasts of Singapore, with much potential for further applications.
Zhi Li - The influences of fire severity on forest regeneration in the West
Fire has played an important role in the western forests, whereas the recent shift of fire regimes has jeopardized human assets and altered forest ecosystems. With the increased prevalence of high severity fires, western forests are now more homogenous and denser. This project aims to investigate the influences of fire severity on vegetation regeneration with remote sensing techniques. Two indices, delta normalized burn ratio (dNBR) and normalized difference vegetation index (NDVI), are used to assess the influences of two fires, one wildfire and one prescribed burn, in the Bridger-Teton National Forest. The result indicates that the wildfire burnt more severely than the prescribed fire. Regeneration has occurred after the wildfire, predominantly in the high severity patches. However, the region has not restored to the prefire condition in 2009. Although remote sensing techniques are efficient in studying the effects of fires, limitations such as image availability and other confounding factors can bias the results. Field sampling should be incorporated with remote sensing to depict a better picture.
Will Strauch - Detection of cumulative spruce budworm defoliation using Landsat imagery
The Eastern Spruce Budworm is the most destructive insect affecting Eastern North American forests. Outbreaks usually last for at least 10 years, and defoliate millions of acres of coniferous forest. Traditionally, data on the long-term defoliation associated with outbreaks are obtained by layering data from annually-commissioned aerial defoliation surveys. In this project, I assessed the feasibility of using satellite imagery to detect long-term defoliation due to a spruce budworm outbreak. Using Landsat imagery, I analyzed an area of Eastern Quebec affected by an ongoing outbreak that started in 2006. I used a combination of NDMI and NDVI vegetation indices to detect relative vegetation changes between 2004 and 2017/18. Then I compared these results with publicly available data on the annual spruce budworm population density from 2014 to 2017. There was a strong visible and statistical correlation between the vegetation-change image and the population density image, which led to the conclusion that satellite imagery could be used as a method of analyzing cumulative defoliation due to outbreaks of spruce budworm.