OEFS Abstracts - Spring 2007

Alark Saxena - Using Landsat imagery to locate coconut plantations on the Island of Dominica

Energy poses the biggest challenge that Dominica faces in terms of sustainable development. According to some studies, Dominica has a large amount of unutilized coconut plantations that could be used to produce Biodiesel, which can save a lot of Dominica’s budget. The latest information about the total area under coconut plantations in Dominica is about 15 years old. For the government to make a successful policy decision, it is imperative to know the exact amount of coconut plantation available for utilization. The project aims to identify and measure the coconut plantations available in the island country. Two available Landsat ETM images were used to analyze the land-cover of Dominica. Due to severe cloud cover in the images, a composite was made that included the best of both the pictures. Two different techniques were analyzed to identify the coconut plantations. Google earth was used as a tool for ground truthing. Based on ground truthing and information from GIS layers, training regions for eight classes were created under the supervised classification. The area summary statistics suggest that there is more coconut plantation in Dominica than harvested. Due to cloud cover, the exercise was not able to analyze the entire island but the results suggested that there is enough resource available for Dominica to start a full-scale coconut-oil/biodiesel production facility. The exercise was also able to identify some areas on the island that have not yet been identified as areas of coconut plantations under the current GIS vegetation layers. The next part of the exercise is to use GIS to identify strategic spots for the location of biodiesel plants based on availability of coconuts.

Daniel Peerless - Exploring methods to age-classify recent fire scars in the boreal forests of Denali National Park

The purpose of this project was to investigate potential methods for the identification and aging of recent boreal forest fire scars in Denali National Park using satellite imagery. Denali National Park is the third largest national park in the United States, and contains within its enormous land area are large swaths of spruce-dominated boreal forest, a heavily fire prone habitat type. Large fires occur annually within the park boundaries, and have a significant impact on wildlife distribution, human health and tourism conditions, and general species diversity. Apart from annual climatic cycles, fire is likely the most significant influence on landscape-scale ecological processes in the region.

A variety of processing and analytical methods were used to attempt to identify fire scars in a Landsat scene of the broad flatwoods that exist in the western reaches of Denali. Although further work is recommended, preliminary conclusions are that unsupervised classifications can be quite successful at identifying scars within the 1-6 year old range, and that thresholds can be set to narrow ages classes further using the thermal infra-red band and/or an improvised Normalized Difference Fire Index.

Xuemei Han - Comparing the effectiveness of different image types in classifying multiple forest characteristics in northeastern China

A promising application of remote sensing techniques in forestry is to understand the spatial and temporal variations of forest with different characteristics. Appropriate Amur tiger conservation layouts in northeastern China could benefit from the classification of forests with different characteristics on satellite images. In this course project, I worked with two scenes of Landsat images (1991 and 2001), applied three methods (i.e. general supervised classification, unsupervised classification and supervised classification with changing detection information) to classify forest types and further applied the supervised classification method to identify forest stand structures. Then I used the ground truth data collected in summer 2006 to assess the accuracy of each method. Despite the generally low accuracy obtained, and the unsuccessful experiment of forest stand structure classification using Landsat images, the results suggested that changing detection could effectively enhance the accuracy of classification. Furthermore, some worthwhile future study directions were pointed out.

Innocent Liengola - Remote sensing as a tool to detect Sericostachys scandens distribution in the Grauer’s gorilla habitats

The forest in the upland sector of the Kahuzi Biega National Park (KBNP) is the site of greatest terrestrial diversity of species and in particularly the endemic species such as the eastern lowland gorilla, Gorilla berengei graueri. However, the Park is suffering particularly from the civil wars and has problems with the people encroaching to farm and settle in the park, mining activities and destruction of habitat through the collection of fire wood and making charcoal. Biological diversity of the KBNP faces many threats; one of these is invasive liana species, Sericostachys scandens. The degradation of natural habitats and ecosystems that occurred throughout the KBNP in the past 20 years has made it easier for the Sericostachys scandens to establish and become invasive. In many cases, many alien invasive are colonizing species that benefit from the reduced competition that follows habitat degradation. Managers of KBNP require accurate and timely spatial information to assist with locating and controlling small infestations of that liana species before to eradicate effectively and to monitor the effectiveness of their management for the protection of the Grauer’s gorilla, endemic species in the eastern of the DRC. In the past decades, traditional invasive plant mapping has utilized ground-based hand or GPS receiver mapping and specify sufficiently high accuracies for small management areas but that methods might be financially, technically and logistically impracticable for many regions and many managers. Remote sensing is providing new tools for advanced management and has been used to measure and map the biophysical characteristics of vegetation.

The main objective of this project is to contribute to the management of the upland sector of the KBNP by showing the land cover and the distribution of Sericostachys scandens, invasive liana species in different land cover and in the relation with Grauer’s gorilla habitats.

Laura Robertson - Spectral assessment of forest damage in the High-Peaks region of the Adirondack forest

Coniferous forests in the High-Peaks region of the Adirondacks have been adversely affected by changing environmental conditions over the last 4 decades. Previous remote sensing studies have quantified forest decline using comparison images from the 60s, 70s and 80s. A 1987 Landsat Thematic Mapper (TM) image and a 1999 Landsat Thematic Mapper (ETM+) image were preprocessed using DN value standardization and dark object subtraction techniques as well as several types of non-coniferous vegetation masks. Each image was subset into an area of approximately 700 km2 surrounding the High Peak Region, classified using supervised and unsupervised techniques, and converted into 4 different types of change analysis composite images. Results were conflicting and inconclusive, but the most promising measure of forest health appears to be a TM 5/4 or 7/4 band ratio, which shows an overall increase in forest health since 1987

Nathan Reagle - Identification and spread of Mountain Pine Beetle in British Columbia

The use of remote sensing for the identification of mountain pine beetle (Dendroctonus ponderosae) damage has been established for Landsat TM and IKONOS data. I applied unsupervised classifications, 214-RGB algorithms, along with NDVI and EVI formulas to MODIS data to detect mountain pine beetle damage. Comparative NDVI and EVI images were used to detect change and identify pine beetle damage spread. MODIS can be used to detect beetle damage, however, classification and determining a quantitative rate of spread is problematic.

Christopher McManus - Measuring sources of wild food in Nkandla, South Africa

The HIV/AIDS epidemic in South Africa is posing a problem for household food security, especially in the rural areas. Several HIV affected rural households rely on wild food sources to sustain household food security. Based on field research I conducted between May and August 2006, several of the wild food sources are being collected from a nationally protected park called Nkandla Forest. Based on reported use of forest materials by community members, conservation efforts appear to be superficial. While continuing collection is beneficial for vulnerable households, inability to accurately assess the effectiveness of the conservation efforts may be harmful in the long run. In this project, my objective was to use remote sensing techniques to assess the effectiveness of conservation efforts within Nkandla Forest between 1991 and 2006. Additionally, I wanted to quantify any change. I found qualitative evidence using normalized difference vegetation index (NDVI) comparison that the Nkanlda Forest has not been degraded. In fact, several forest gaps present in 1991 appear to have filled in by 2006. Due to classification errors, however, quantification of the change was not possible. Further, groundtruthing in the future will be necessary to determine whether indigenous Zulu forest or invasive species have filled in the gaps.

Kimberly Lau - Studying primate habitat canopy density changes near Lake Nakuru, Kenya

Lake Nakuru National Park in Kenya is part of the Great African Rift Valley and is home to the vervet monkey and the black and white colobus (or the guereza), both primates that are quite abundant in this part of Kenya. Recently researchers have noticed that flamingos, the prime attraction of the park, are leaving this sanctuary in large numbers, possibly due to adverse human industrialization and agricultural interferences. These could have caused changes to the area’s ecosystem that affected the populations of vervets and black and white colobus as well. In this study I used various types of techniques on Landsat images, include NDVI comparison, unsupervised classifications, and supervised classifications, to gain a solid understanding of the landcover of the park area and the nearby Mau escarpment in 1986 and 2000. With the data and observations collected I made predictions about how the populations of the two primates could have changed between these two years, as well as comment about how the park environment itself has changed. Overall, I predicted the vervet populations remained stable in the Mau forest and increased inside the park, while the black and white colobus populations increased in the Mau escarpment and remained stable inside the park. I was unable to detect negative changes from human causes on the park’s environment. I also concluded that the results obtained from the satellite images were a bit questionable and may not produce the accurate results that are necessary for such an analysis.

Alex DeWire - Examining post-Katrina wetlands recovery

The goal of this project was to detect and quantify the amount of land change in the Breton Sound, Louisiana as a result of Hurricanes Katrina and Rita in 2005. Additionally an ASTER image from 2007 was used to measure any regrowth of salt marsh vegetation since the image after the Hurricanes. Change was detected using NDVI comparisons visually, and then quantified using NDVI differences. The study area was further refined to include only marsh and open water with mixed results.

Paula Randler - The salt marsh at Sherwood Island State Park from 1974 to 2005

The objective of this project is to track changes in land coverage in the marsh at Sherwood Island State Park from 1974 to 2005. Aerial photographs were georeferenced to Connecticut State Plane and classified using training regions and supervised classification. Lacking metadata and good ground control points, the warping of each image is not accurate. Change detections were run despite these problems and area summary statistics were used to determine correlations between different area types. The area types include Water, Mudflat, Spartina alterniflora, High Marsh Mosaic, Sand, and Shadow.

Spartina alterniflora and High Marsh Mosaic are negatively correlated to one another (-0.726) as are Water and High Marsh Mosaic (-0.571) and Spartina alterniflora and Mudflat (-0.534). Higher altitude land types Sand and High Marsh Mosaic are positively correlated to one another (0.548). Remote sensing is not the best method to use in finding relationships among different land cover types in aerial photographs in light of problems with photographic angle, warping, and high resolution which makes the pixels of one type too different from one another to be classified confidently.

Chris Yuan-Farrell - Change detection in a salt marsh community along the Quinnipiac River

Healthy salt marsh ecosystems are vital to preserving ecological processes essential for maintaining water quality and biotic communities. Landscape scale changes have affected the quality and quantity of these habitats throughout the United States. The current study addresses the suitability of techniques used to assess these changes in a small marsh community in south, central Connecticut. Results are consistent with recent analyses in this community and enhance the ability to accurately assess changes in vegetation cover and species composition using advanced techniques in the field of remote sensing.

Punit Lalbhai - Predicting bird abundances and composition using satellite imagery

In this paper I explore the effectiveness of satellite imagery for its use in models predicting bird species occurrence. The effectiveness of such models lies in the ability of remotely sensing habitat features that are important in driving bird species distributions. Here I use an image stack comprising three Landsat TM images of three different seasons (late winter leaf-off, summer leaf on, and fall leaf on), and a high spatial resolution IKONOS image to try and distinguish differences in 1) habitat structure, 2) understorey mountain laurel (Kalmia latifolia), and 3) tree species assemblages. NDVI change comparison, multi-temporal NDVI classification, and unsupervised classification showed good success in differentiating between coniferous and deciduous vegetation, and deciduous vegetation with a mountain laurel understorey, at both spatial resolutions. Supervised classification yielded poor results. These methods, collectively, were not effective in detecting changes in forest structure and tree species assemblages.

Abigail Letzter - Urban expansion of Chicago from 1989 to 2001

Throughout the world, the expansion of urban areas into more rural areas is a fact. Although some regions within cities show an increase in repopulation within already standing infrastructure, for the most part, cities are continuing to grow outwards; building new houses and buildings, leaving older infrastructure to the whims of the real estate market. The outgrowth of the urban center into the surrounding areas is often not well planned, and few people have actually tried to take into consideration what or who will be hurt in the process. Specifically, however, this expansion of urban areas is detrimental not only to the environment but also to a certain demographic within the United States; the rural poor and agriculturally based communities. Urban expansion can damage water tables, increase erodible land, and diminish the rural community. Urban expansion can also change the immediate temperature as can be seen in Thermal Landsat Images. In order to better understand this land use change pattern and its affects, I decided to look at the Chicagoland area via satellite, census data, and other urban sprawl literature. I hoped that the satellite imagery would detect if there were any significant visual change that coincided with obvious population increases that are evidenced by census data.

Brenna Vredeveld - Land cover change in and around Quito, Ecuador

Quito, Ecuador has experienced rapid urban expansion over the last fifty years. Changes in land cover are tied to changes in the export economy that have influenced migration and settlement patterns within the country. In the past century, Quito has absorbed a growing population that has altered its regional land cover. For example, from 1950 to 1990, the urban city has grown six-fold in population and twenty-fold in area. Increasingly, natural land covers are being cleared for agriculture, and agricultural lands are being developed for residences, industries and commercial business. To begin to understand the scope of land cover change in this region, I compare two remotely sensed images taken at different dates: one Landsat TM taken on March 26, 1987 and one Landsat ETM+ taken on November 14, 1999. Both were acquired from the Global Land Cover Facility database. I use a variety of preprocessing techniques to prepare the images for analysis, including subsetting, georeferencing and rectification, atmospheric correction and cloud masking. The cloud masking approach, however, was eventually abandoned given its small utility for this project. I apply the Tassled Cap transformation to create three indices for each image: Brightness, Greenness and Wetness. Using these indices one is able to observe changes in land cover associated with urbanization in Quito. These observations will aid in identifying those areas which have experienced the most change from 1987 to 1999. In future research, supervised classifications of land cover change in those focal areas will aid in quantifying the various land cover changes.

John Nixon - Comparing land change over ~14 years in the northern and southern study areas of the BOREAS research project

The BOREAS project was an intensive study of the boreal forest ecosystem that involved hundreds of scientists over several years and produced over 500 publications. The BOREAS mid and final reports made predictions of the effects of global warming on the study sites, including longer thaws of permafrost and increased carbon outgassing from soils. This project attempted to follow their predictions and see if there was any significant discernable change between the late 1980’s, before the project started, and 2001, several years after the project ended.

After literature searches disproved many of the attempted methods to demonstrate the BOREAS predicted change, the author uses various indices to compare effectiveness at differentiating regions within the BOREAS study area. NDVI and RSR are found to have similar effectiveness while TCI is found to be superior for visual differentiation of regions. While the initial goal of the study is not met, the author concludes that BOREAS was successful in operating without altering the ecosystem.

Laura Frye-Levine - Modeling land-cover change in a Honduran national park

Cusuco National Park, a 7,700 hectare montane cloud forest in Northern Honduras, was established in 1987 as a legally protected area with high biological diversity. As a newly protected area in a third world country - the conservation status of the park is far from secure. Local villagers have historically depended on forest products from the park for survival, and a successful management plan addressing park conservation and rural development has yet to be drafted. This summer I will travel to Cusuco the perform mapping, ground truthing of land use patterns, and social science research to aid in the creation of such a management plan. I am using remote sensing to help assess the trajectories of land use within the park since its inception.

This semester in OEFS, I have attempted to identify significant trajectories in land use change within the park. To accomplish this I have used mostly qualitative methods of change comparison and analysis, including unsupervised classification, NDVI, PVI (Perpendicular Vegetation Index), MSAVI2 (Modified Soil Adjusted Vegetation Index), DEM drapes of composite vegetation indices, and visual comparison of images from the two years using different spectral combinations. Absent ground truthing data, my ability to make significant quantitative judgments about land use change in Cusuco National Park has been limited, though the familiarity I have gained with the local terrain and through the analysis of generalized patterns of change has been invaluable preparation for the fieldwork I will undertake this summer.

Lei Lei - Examining wetlands changes around Lake Powell from 1991 to 2000

Wetlands play an important role in maintaining a healthy ecosystem and provide variety of recreational opportunities. Lake Powell is a man-made reservoir and has been experiencing severe water loss problem since its completion due to both the arid climate and ongoing projects. The insufficient water problem could lead to wetlands loss and impose a threat to the local ecosystem. This paper aims at examining the changes of land covers in Lake Powell, Utah from year 1991 to 2000. Three approaches are employed: unsupervised classification, supervised classification, and NDVI. Results from the unsupervised classification are hard to use because of less distinctions between different land cover types. The supervised classification method indicates an increase in water, urbanization and riparian rocks and a significant decrease in riparian vegetation. However, it is hard to identify whether the riparian vegetation are wetlands or not without credible ground truth. Finally, an NDVI approach is used to show the changes of different years. Again, the NDVI image reflects that most changes happened along the river bank. For future work, to improve the accuracy of land cover classification, extensive field work and local expertise are needed.

Natalie Ceperley - Land-cover change in riparian corridors of the Central Oueme Basin

In the Oueme Basin, primarily in the Republic of Benin, riparian forest cover borders the many waterways, providing ecosystem services at local, regional, and basin-wide scales. However with agricultural expansion, market growth, and increasing access to roads, riparian forests may be under new anthropogenic threats. Identifying and monitoring riparian forest cover in the context of current land-use change is important to maintaining the integrity of these forests and their hydrologic networks in central Benin. Remote sensing has the potential to play an important role in this effort. This project compares the viability of the land-use change sensing techniques of normalized difference vegetation index (NDVI) overlay, unsupervised classification, and supervised classification for detecting changes in riparian forest cover in the central Oueme Basin in the Republic of Benin using Landsat TM and ETM+ images from 1986 and 2000. Qualitative observation of NDVI change was found to be the most powerful tool at this scale and stage.

Lynda Odofin - Observing the Niger Delta from Space: The effects of oil exploration

The Niger Delta one of the world’s largest oil producing regions and has been used to illustrate the cause of natural resources. All the woes occurring in the region have been linked to oil exploitation, from mangroove loss to HIV incidence. In this project I decided to investigate landscape changes using remote sensing. I performed NDVI to detect changes qualitatively, and employed supervised and unsupervised classification for quantitative analysis.

All methods showed landscape changes, the most significant was the decrease in forest area, and urbanization, and increase in agriculture. This reflects the culture change of the area, which used to be water dependent (fishing) to land dependency (farming).

Tom James - Vegetation change assessment in the Darhad Valley, Mongolia

Landsat TM and ETM+ data were analyzed using ERMapper software to assess vegetation change in the Darhad valley Mongolia. Two Landsat images were chosen from NASA’s global land cover facility site and after being downloaded, were formatted using ERMapper’s CEO Tools toolbar. The two images represented 1986 and 2001 scenes. Change assessment was made using three vegetation ratios, which were subsequently compared to the supervised classifications of both scenes. The Green Vegetation Index (GVI), Normalized Difference Vegetation Index (NDVI) and the Short Wave Infrared Ratio (SWIR) were compared and evaluated relative to their ability to identify vegetation change over the identified 15 year period. NDVI and SWIR were visually more effective than the GVI ratio, although further analysis is needed to substantiate this observation. Throughout, the persisting challenge is one of detecting change in a relatively open forest canopy, where understory reflectance strongly influences image analysis. The SWIR was favorable in ameliorating this issue. A supervised classification for each scene did agree with the composite vegetation indices, but without ground truth data, landscape features could not be quantitatively predicted.

Tara Mayeau - Vertical uplift and normal fault identification in Western Crete from InSAR and ASTER

Crete is an island that is actively deforming in the presence of a subduction zone and numerous normal faults. The question of which faults are responsible for most of the deformation has significant implications for seismic hazard in the region. This project was an attempt to assess the usefulness of remote sensing data in addressing this question. Currently, InSAR is the preferred method for using satellite data to study active tectonic deformation. However, a number of computer problems have prevented, at least so far, the analysis of SAR data from the ERS-1 and ERS-2 satellites. As an alternative, an ASTER image was analyzed visually using several RGB band combinations, unsupervised classification, and with high pass filters to determine which of these methods would yield the most useful information about the faults in western Crete. While visual analysis of the landscape yielded several fault traces that were evident in the RGB combinations, the unsupervised classification and high pass filters tended to obscure those features, rather than enhance them. Thus, visual inspection of high resolution RGB data seems to be the best way to identify the faults in the ASTER data. Higher resolution data, such as that available from Google Earth, also makes it possible to identify the amount of erosion along the fault scarp, which is an indication of the level of seismic activity. However, this method still suffers from several shortcomings, primarily the inability to identify faults that do not have obvious surface expressions. The future addition of InSAR data should assist in better mapping out the active faults in Crete, even those without significant surface expressions.

Karen Stamieszkin - Pacific Ocean SST and chlorophyll concentrations in normal and El Niño years

As of the early 1990’s, scientists have documented the effects of the El Niño Southern Oscillation (ENSO) in the equatorial Pacific Ocean. During an El Niño event, the air pressure gradient across the western half of the Pacific shifts. This shift is part of the Southern Oscillation. Normally, there is low air pressure recorded at Darwin, Australia, and high air pressure recorded in Tahiti. During an El Niño event, the air pressure recorded at Darwin indicates a high pressure system, and the air pressure recorded in Tahiti indicates a low pressure system. Because of this pressure gradient, during normal years, the Walker circulation creates easterly winds across the Pacific, rising air and precipitation in the Western Pacific, a piling up of warmer surface waters in the west, and the upwelling of cold, nutrient-rich, deeper waters in the east. Therefore, there is typically increased production in the Eastern Pacific as compared with that during El Niño events.

During an El Niño event, the Walker Circulation is disrupted by the change in pressure gradient across the ocean. Low pressure moves from the west to the middle of the Pacific, weakening easterly winds, creating rising air and precipitation in the mid-Pacific, and dampening upwelling in the Eastern Pacific. Therefore, during El Niño events, less nutrient water reaches the surface, resulting in decreased productivity. Correlated with the drop off in primary productivity in the East Pacific during El Niño events, scientists have documented a similar effect on zooplankton, and in significant and commercially valuable fisheries.

It has been hypothesized that the opposite occurs to some extent in the Western Pacific; upwelling should increase somewhat, elevating the thermocline in the area, and resulting in increased productivity. Further, during El Niño events, one should see increased productivity, cooler surface temperatures (cooled by the elevated thermocline, i.e. cooler water closer to the surface waters), and decreased sea surface height (SSH, due to the contraction of the cooler water that is closer to the surface). Satellite imagery can be used to test these hypotheses. For this project, chlorophyll and sea surface temperature (SST) products from the OceanColor website by NASA were used to look at large-scale patterns of SST and chlorophyll (http://oceancolor.gsfc.nasa.gov). Further, NOAA’s altimeter data from their Global Ocean Observing System Center was used to examine SSH (http://www.aoml.noaa.gov/phod/goos.php). While the data are not fit to coordinates, and on an arbitrary scale, they can be compared to each other, and qualitative analysis can be done.

Micah Ziegler - Seasonal ice pack and change over time of the size of the Ross Ice Shelf

The Chesapeake Bay is the largest estuary in the United States and plays a critical ecological and economic role in the Mid-Atlantic Region. Sediment in the Bay increases turbidity, preventing light from reaching submerged aquatic vegetation, a crucial component of the ecosystem. In the upper bay, the Susquehanna River provides most of the sediment.

Suspended sediment in the upper bay was studied using 23 tiles of 16-day composite MODIS reflectance images of the region compiled at even intervals during the 2006 calendar year. This project analyzes the ability of this the atmospherically and radiometrically corrected composite reflectance product to visualize temporal and spatial changes of suspended sediments over a variety of concentrations. The satellite data were compared and correlated against in situ measurements of secchi disk transparency and total suspended solids. Correlations between satellite data and water-based measurements were studied empirically and statistically. A good empirical relationship was found between in situ sediment levels and satellite observations in the red MODIS band. This band was then used to map the location and relative intensity of the northern Chesapeake’s estuarine turbidity maxima for 2006. Ideally, this procedure could be used to map the Bay’s sediments over a longer time span with better spatial resolution than is currently available in situ.

Bridgid Curry - Characterizing the smoke plume generated by a Canadian forest fire

MODIS imagery was used to characterize a smoke plume that affected the East Coast of the United States from July 6 through July 9, 2002. As a result of the smoke plume, air quality declined in the region and posed a risk for people with respiratory ailments. Imagery from the day of peak plume activity was characterized using a normalized difference dust index (NDDI) to distinguish the aerosol plume from the surrounding clouds and land. Brightness temperature values for the plume were calculated from radiance measurements and were used to estimate the altitude of the plume. Both methods were only moderately successful in providing additional information about the plume. The NDDI isolated the plume but the index values for aerosols generated through biomass burning did not agree with those reported for sand and dust events. The temperature and altitude profiles indicate the plume was located 2 kilometers (km) above the land surface and had a temperatures ranging from 291 to 295 K. Further analysis is necessary to understand how aerosol data collected through remote sensing can be used to enhance exposure information for use in public health studies.

Xiaoyue Du - Using MODIS data to study sand and dust storms in eastern China

Sand and dust storms are nature disasters, and became more frequent in modern time. Remote sensing can be a better way to understand the sand and dust storms. The satellites with the Moderate Resolution Imaging Sprectroadiometer (MODIS) sensor provides us daily image of the global scenes. The detected spectrum characteristics of sand and dust storm in detail and bands information can help us distinguish sand and dust from clouds and surface. The Normalized Difference Dust Index (NDDI) was designed to identify sand and dust storms and tested well in my project to determine the scope and intensity of sand and dust storm.

Melanie Parker - Investigating cloud properties off the coast of Southern California

This project involved using a MODIS daily scene image to calculate cloud properties. The image contained areas of clouds off the coast of southern California. The Planck Function was used to calculate the brightness temperature of the clouds in the image, and the cloud altitude was determined using the brightness temperature values. The main purpose of the project was to compare the calculated values for these cloud properties to data contained in the MODIS Atmosphere Group’s cloud product. The calculated values of brightness temperature and altitude were similar to the data in the cloud product. This result indicates that the Planck Function is a good approximation of brightness temperature and that a simple height-dependent temperature model is also a good approximation.

Kasia Wegrzyn - The change in maximum sea ice extent in the Greenland Sea, 2000 to 2006

In this study, I used five 8-day composite images of a MODIS surface reflectance package to study the change in sea ice extent in the North Atlantic over the past five years (2002 - 2006). One image per year from March was analyzed, with an attempt to take all images from the same period. The study area is centered at 75.71° N and 19.51° W off east coast of Greenland, in the Greenland Sea. The methods used to analyze the maximum presence and extent of sea ice in the North Atlantic included five indices-the Normalized Difference Snow Index (NDSI), the Normalized Difference Snow/Ice Index (NDSII), NDSI-band 7, Ice Index, and Albedo-and two visual comparisons of true-color RGB-143 and also RGB-367 images to comment on trends in ice distribution and thickness at its maximum extent. This analysis was then used to determine if in fact there is a decrease in the maximum extent in sea ice in the North Atlantic, as much literature suggests.

Kevin Currey - Examining high-latitude mountain meteorology on Mt. McKinley

This paper examines the factors influencing the heat budget of the Earth’s surface on Denali (Mount McKinley), Alaska, using a Landsat-7 ETM+ image of Denali from September 27, 2001. First, using data from a local radiosonde sounding, I examine the influence of the temperature of the free atmosphere on surface temperature. Second, I examine the effects of solar heating, a function of surface albedo and cosine I, the sun incidence angle with respect to surface normal, on surface temperatures. I then discuss two traverses through different sections of the image and, using multivariate regression, attempt to ascertain the relative influence of the temperature of the free atmosphere and solar heating on surface temperature. Although the temperature of the free atmosphere has the dominant influence over surface temperature over long distances with large fluctuations in elevation, solar heating has the dominant influence over shorter distances with much smaller fluctuations in elevation. This study concludes by discussing how NVDI and melting snow might affect surface temperatures and suggesting future research that should be conducted to better understand the surface heat budget in high-latitude, high-altitude mountain microclimates.

Matt Grant - Studying mountain micro-climates in Glacier National Park

The variable topography of mountainous areas leads to highly variable micro-climates within a given region, with substantial effects on temperature and other meteorological variables. These effects often have great practical significance; previous studies have considered the consequences of mountain micro-climates when controlling forest fires, the importance of micro-climates in controlling alpine snowmelt and consequently downstream water supply, and the importance of microclimates in controlling albedo, perhaps with broader implications for climate.

In order to gain a better understanding of how topography affects climate, the problem was investigated in a small, mountainous region within Glacier National Park using Landsat ETM+ remotely sensed data for September 12, 2002 on Path 041, Row 26 when snow cover is close to its annual minimum. This data included all eight ETM+ bands (sample visible and TIR images are provided on p. 3) with standard ETM+ resolution. Additionally, a digital elevation map (DEM) having 90m by 90m resolution was obtained from the USGS Seamless Data Distribution System. These data were augmented with upper air profiles in radiosonde data for station TFX Great Falls September 12, 2002 at 0 Z (6:00 p.m. local time) from the University of Wyoming data archive. To simplify the problem, the image and DEM were subsetted to extents 48:39:47.95N to 48:31:39.71N and 113:50:39.62W to 113:37:19.77W; this region was selected because it shows a high degree of topographic variability, encompassing three peaks, an elevation range of 1006 m to 3061 m, and several snowfields, is cloud free in the image, and is removed from potential human disturbances which might confuse signals created by climate variation.