OEFS Abstracts - Spring 2010

Ben Blom - Forest loss along a political gradient: Change in the border region of Hispaniola

Quantification of land use change using remote sensing has altered the landscape of governance and management of the world’s tropical forests by increasing transparency and accountability (Fuller 2006).  Among other things, remotely sensed data on forest change in the tropics has been used to focus attention on illegal forest clearing in protected areas, the impacts of land policies on forest loss, the spatial identification of deforestation threats and the measurement of forest carbon stocks.  Given the rapid pace of forest loss in the tropics, these new methods for monitoring and analyzing forest loss have proven to be invaluable tools for understanding and altering these patterns of land use change.  More recently, forest loss in tropical areas has received increased focus for the carbon emissions that result from tropical forest conversion.

 

Eli Mitchell-Larson - Building an objective case for uniqueness of the Perucica Forest Reserve

The Perudica primeval forest in Bosnia and Herzegovina is one of the last of its kind in Europe and offers a unique case study of the role of old growth forests in the global carbon cycle. This project employs intermediate spatial resolution satellite images (LANDSAT TM) and SRTM DEM data to derive estimates of Gross Primary Productivity (GPP) for the Perudica reserve and several surrounding areas representative of different land cover types. Building on remote sensing forest research methods designed for nearby Mediterranean forests, this project employs a simplified model of GPP that utilizes available solar radiation and the Fraction of Absorbed Photosynthetically Available Radiation (fPAR) index along with external data to derive its estimates. The quantitative results are presented with the intent that they be refined and confirmed through subsequent on-site field work which could provide valuable insight into the carbon content of these old growth forests. Through comparison of GPP values between seasons and across decades, land cover types were analyzed according to their significant in photosynthetic processes.

 

Elaine Hooper - Utilization of remotely sensed data to characterize tropical forest fragmentation

Landsat Tm imagery was utilized to assess the extent of deforestation and the rate and pattern of change in deforestation over a 20 year period (1987 – 2007) in an area located in the Brazilian Amazon. The study area included Manaus (located at the confluence of the Rio Negro and Rio Solimoes which join at Manaus to form the Amazon river) and the Biological Dynamics of Forest Fragments Project site, located approximately 90 km northwest of Manaus. Supervised classification using the maximum likelihood algorithm followed by change detection analysis allowed an accurate determination of the extent and pattern of deforestation of primary, upland terre firme tropical rainforest. Over 450 square kilometers of primary forest were converted to pasture within this 20 year time period, half of which was abandoned. Comparative NDVI and change detection using a forest biomass index calculated for the 20 year period confirmed results of change detection using classified images and added ecological insights into the biophysical nature of forest change. A detailed analysis of reflectance values for pixels located within forest fragments of 1, 10 and 100 ha was undertaken to determine the effect of forest fragmentation on biophysical forest properties, using NDVI and a forest biomass index. These values were then correlated with tree seedling species richness data collected within these fragments to assess the potential use of easily measured vegetation indices to predict biodiversity change with deforestation. Regression results indicate a correlation of 25-27% of these index values with tree seedling diversity suggesting their potential, in conjunction with other correlates of biodiversity, to predict losses in the biodiversity of forest regeneration with deforestation in areas where it is impossible to conduct detailed biological fieldwork.

 

Lucy Dunnington - Northern Russia boreal forest

This report is a description of remote sensing’s possible utility in evaluating the impact of climate change on Russia’s Boreal Forest. Studies have shown that the positive feedback mechanisms that work in Siberia’s forest has almost doubled the change in temperature in the area. With remote sensing techniques, such as measuring NDVIs, albedo, and emissive wavelengths with thermal bands, scientists are able to carefully and efficiently monitor the variation in climate and forest activity. In this study, there are a few examples of such measurements. This paper looks at a couple options for satellite imagery, and eventually uses MODIS bands to calculate NDVI of various land cover types at different times of year. It also allowed for measurement of surface temperature and albedo, to track the correlation between forest readings and global warming. The results are consistent with other research conducted in the area, though it does shed some light on the limitations of remote sensing in such investigations.

 

Roopa Krithivasan - Forest cover change in south India

My goal was to gain a better understanding of land cover change in the forests of south indai over the past two decades. I also wanted to test the hypothesis that increasing the number of bands used in a supervised classification would initially yeild increasingly superior results, then plateau and yeild no additional accuracy, or degrade the quality of results.

 

Eliot Logan-Hines - Land-use change in Coto Brus, Costa Rica 1987-2001

This paper uses various tools of remote sensing to analyze land use change in the county of Coto Brus, Costa Rica between 1987 and 2001. The goal of this project was to attempt to outline the foundation of baseline carbon stocks for the region for future use in regional carbon accounting. The results were far from satisfactory in this regard, but they shed light on the possible uses of remote sensing for such purposes and some of the challenges it represents. Landsat TM images from January, 1987, and January, 2001, were used in the analysis. A series of supervised classifications at various thresholds were conducted as well as a false color NDVI image. A discussion of each of these tools and their implications is discussed followed by recommendations for future analysis.

 

Pablo Reed - Deforestation and urbanization patterns around Lago Agrio Sucumbios, Ecuador

The area surrounding the city of Lago Agrio in the northeastern section of the Ecuadorian Amazon has been one that has suffered considerable and rapid land use and land cover changes in recent decades. Fueled by the building of roads first intended for oil exploration, the region quickly became a site for major migration where settlers took advantage of these new roads to clear large portions of tropical rainforest for cattle grazing and agriculture. Though colonization more or less subsided in the mid 1990’s, deforestation and agricultural expansion rates have continued to expand. This study uses remote sensing and satellite imagery to determine the extent and patterns with which this expansion into forested lands has continued, as well as determining what possibilities for conserving the few remnants of forest in the region may exist.

 

Giancarlo Raschio - Deforestation and land degradation detection using Landsat TM images

Illegal and small scale gold mining activities in the Manu region of the rain forest of Peru are the reason for deforestation and land degradation in the upper areas of the basin. This threaten the most biodiverse area of the country and put at risk the health of the people that live in the lower parts of the basin.

Nowadays, the Ministry of the Environment in Peru is working to develop regulations to legalize the activities of low scale/illegal gold miners aiming to improve their environmental practices and to help them develop this activity under a sustainable and controlled framework. However, as this area present high biodiversity and is close to national protected areas, the time it’s taking regulations to take place means tropical forest loss and the opportunity to generate economic and cultural benefits for local indigenous communities.

 

Troy Hill - Salt marsh vegetation change at Sherwood Island State Park: 1974-present

This study assesses changes in vegetation at Sherwood Island State Park, Westport, CT, over the time period 1974-2005. False-color infrared aerial photos were georeferenced and, following a site visit, supervised classification was used to divide the marsh into botanically distinct zones. The Sherwood Island salt marsh declined in area by 21% over the three decades studied, amounting to a net loss of 4.4 hectares of vegetated area. The most affected vegetation class was the ecologically critical Spartina patens-dominated high marsh, which declined in areal coverage by 75%, accounting for 90% of the loss in total vegetated area. These results, which contribute to an understanding of the nature of marsh drowning in Long Island Sound, are contextualized with a discussion of previous research at Sherwood Island and other data regarding anthropogenic stress on salt marshes. This study illustrates the potential value of freely available aerial photos in assessing vegetation change in coastal wetlands.

 

Eliza Little - Detectng environmental drivers of 2 important Dengue Fever vectors

Aedes aegypti is the main dengue vector worldwide because of its close association with humans in tropical and sub-tropical urbanized areas. Given its role in the emergence of epidemic dengue, vector control methods have focused on controlling Ae. aegypti by eliminating artificial water containers that function as breeding sites. However, the efficacy of vector control methods has been limited in part by the presence of alternative natural container breeding species. In Puerto Rico, Ae. mediovittatus, a native mosquito species, is a competent
dengue vector with a high rate of vertical transmission for all dengue virus serotypes. Its distribution is thought to overlap with that of Ae. aegypti in areas with both low-density housing and the presence of trees. The primary goal of this research is to use geospatial analytical tools to describe the environmental drivers that shape the habitat of important vectors of dengue viruses in Patillas, Puerto Rico. BG-Sentinel traps will be used to collect and measure density of adult mosquitoes in Patillas. Using a GIS, the distribution of both Ae. aegypti and Ae. mediovittatus will be mapped and special attention will be given to areas of overlap. The second goal of this project is to explore the use of remotely sensed imagery to detect environmental features linked to presence of Ae. aegypti, Ae. mediovittatus, and co-occurrence. Environmental variables will be assessed at different spatial scales using multiple remote sensed images at different spatial resolutions. Using these geospatial tools to define environmental features linked to co-occurrence provides the opportunity to focus intervention efforts to areas of high risk, especially important in resource poor locations.

 

Dan Summers - Characterize the dengue vector in Ribeirão Preto, Brazil using several ASTER images

Among the countries of South America, Brazil in particular has suffered from a series of major outbreaks of dengue fever in the past two decades. Rapid, dense urbanization of cities, combined with insufficient health infrastructure and a tropical climate have created a perfect environment for the urban vector of dengue, the Aedes aegypti mosquito. In Ribeirao Preto, located in the southeast of Brazil, efforts have been made to map areas at risk for disease transmission. As part of a larger study of mosquito wintering populations as predictors for rainy-season dengue outbreaks, remote sensing of ASTER images was utilized to attempt to identify parts of the city as potential case-predictor locations. Calculations of temperature, vegetation index, and land cover change were performed on three images depicting winter 2000, winter 2009, and summer 2009. It was then determined that change has indeed occurred in several areas around Ribeirao Preto, creating new potential new mosquito habitat. The role that these locations will play as dengue predictors has yet to be determined.

 

Fleur Porter - Urban expansion of Dhaka, Bangladesh, with implications for disease ecology and health management

Supervised classifications are traditionally conducted using ground data in order to generate regions of interest that guide the classification process, by defining a range of spectral signatures that characterize particular land‐cover classes. In the absence of such field data, unsupervised classification algorithms are traditionally used, as they generate classes based on statistical computations rather than any training regions representing the area. However, unsupervised classifications are limited in their descriptive capacity, and are not ideal for generating robust information about a study area. The purpose of this study was to devise a multifactorial method for conducting supervised classifications in the absence of field data. Spectral data was combined with NDVI, brightness, greenness, and wetness indices, and digital elevation data in order to produce new “spectral signatures” defining each land‐cover class. These regions of interest were then used to carry out a change detection analysis to assess the degree of urban expansion and land cover change that has occurred in and around Dhaka, Bangladesh from 1989 to 2010. The results indicate that the multifactorial method for conducting supervised classification in the
absence of field data is a promising alternative to unsupervised classification using the Kmeans or ISODATA algorithm, and calls for continued experimentation and refinement of the protocol.

 

Tyler Griswold - Examining the ecologic determinants of risk for yellow fever and dengue in Santa Cruz, Bolivia

Yellow fever virus is vector-borne pathogen that puts a large population at risk for life threatening illness in the tropical regions of South America every year. Yellow fever is transmitted to humans through contact with Aedes Aegypti and Aedes Haemogogus mosquitoes who serve as virus reservoirs. Between 1997 and 1999 Santa Cruz, Bolivia experienced the first acute outbreak of urban yellow fever virus in Latin America in over 40 years. Political, economic and social reforms aimed at land use policy can affect the natural environment, shift epidemiological patterns, and alter the risk parameters for yellow fever by encouraging deforestation, and agricultural and human expansion. In 1996, the Bolivian government pushed through a series of land reforms (PLUS/INRA) that opened up once private, estate-owned land areas for development. This created rapid land development and a population shift to the lowland forested regions around Santa Cruz. Remote sensing techniques provide a suitable platform to assess land use change and examine the ecological determinants that characterize the environments for mosquito breeding habitat and human expansion.  The objective of this study is to estimate land use change from LANDSAT TM 4-5 images in 1996 and 2005 in the Santa Cruz region to examine the effects of the INRA/PLUS Reforms. The purpose of estimating land use change is to identify high risk areas that showed human expansion into suitable mosquito breeding habitat. Several multi-spectral analysis techniques are compared to assess land cover classification and land use change detection.

 

Ian Starr - Using ASTER and Landsat to detect small illegal road networks within the Sete de Setembro Indigenous Reserve

Illegal logging operations occur throughout Brazil, although surprisingly, the forests on indigenous reserves in the country have remained remarkably intact. One such example is the Sete de Setembro Indigenous Reserve located in Rondônia, Brazil, which boasts a demonstrated interest by the communities to monitor and safeguard their territory from illegal uses.  Despite the relatively limited scope of illegal logging operations in the reserve, indigenous community leaders have sought the help of outside organizations to assist them in identify and monitor this activity across tens of thousands of acres. Remote sensing holds great promise for its potential to assist organizations in the management of tropical forests resources at a lower cost.  The goal of this project therefore is to attempt to use low-cost images such as ASTER and LandSat to detect small illegal road networks within the Sete de Setembro Indigenous Reserve in southwestern Brazil. Ultimately such efforts could serve to direct land vigilance efforts in the reserve in a more cost-effective way.

 

Alexandra Andrews - Rabbit Creek Gold Mine: a study of land use change in Winnemucca, NV

Remote sensing creates a unique method for quantitatively measuring land use changes over extended periods of time.  Three Landsat 4-5 TM images, dated April 5, 1985, September 5, 1994, and July 9, 2008 were used to assess land use changes before, immediately after, and long after the Rabbit Creek Gold Mine was operational in Winnemucca, Nevada. The area was classified using Isodata and K-Means unsupervised classifications. However, even the best of these classifications (K-Means, 5 classes) was not able to successfully classify the area, specifically failing in the 1994 image.  Then, supervised classification was used with 11 classes, including 6 classes to differentiate specific sections of the mine.  This was more successful and proved to cover most of the pixels in the range of study as the threshold was increased from no threshold to 0.99.  To compare land cover changes, masks were applied to two specific classes. This method failed, but a third ROI mask using an outline of the mine from the 2008 image allowed the areas to be compared in a quantitative way.  The regions of interest used in the supervised classification limited the results of the study as the 1985 image classified areas as parts of the mine when it had not yet been built.  If the supervised classification were revised, this method could prove useful for determining the environmental effect abandoned mines have on the surrounding area.

 

Manuel Mavila - Infiltration and runoff processes across three main land uses in the Peruvian Andean Amazon

The Andean Amazon maintains exceptional diverse cloud forests that provide essential hydrological services. Rural communities in these mountainous headwaters rely on streams for drinking water, generating electricity, and obtaining fish -their main protein source. With high deforestation rates surrounding the cloud forests, and global warming elevating the fog-formation altitude, the hydrological services these forests provide downstream are at risk. Once the forest is lost, the alteration of the physical and hydraulic properties of soils becomes a key component in the degradation of the hydrological services in terms of water quality and quantity.

The study will provide scientific understanding of water, forests and people dynamics in an under researched and fragile ecosystem. In terms of practical application, this study will support the construction of hydrologic models beneficial for the catchment’s water system and energy supplier, contribute to the baseline and monitoring of the hydrological services, and facilitate the implementation of a pilot program in Payment for Ecosystem Services planned by the local government and National Park’s authority for this catchment.

My main goal in the OEFS course project is to use remote sensing for the first stage of the study to: delimit the watershed, characterize its geomorphology, classify land use patterns, and identify a representative sub-area of the watershed where to conduct my field sampling.

 

Pam Labib - Egypt land cover changes that impact ecosystem primary productivity

Despite rapid land-use and land cover change on the Eastern Desert’s Red Sea coast over the last two decades, very little information is available on the nature, extent and impact of the land-use changes in that region over-all. Although the largest scale changes have primarily been through the tourism industry and the development of large resorts that have brought along with them towns, roads and infrastructure. The landscape is variegated by developments, such as military bases, government housing projects, extractive industries and protected parks. The Egyptian government planned to transform the coast into a Red Sea Riviera extending from Hurghada to Sudan by 2005.

This project is a component of larger project titled, “Environmental Identity, Ecosystem Productivity and Land-Cover Change in The Eastern Desert Red Sea Coast of Egypt” This component of the project will be measuring the extent of coastal developments and its impact on the above ground biomass of the vegetation in that region. The main objective of this project is to explore the different tools that are available through remote sensing that can be applied in assessing built infrastructural land-cover changes and changes in sparse vegetation cover during the dry summer months for the application to my thesis research.  I  also conducted a small analysis on change detection in coral reefs to give myself the opportunity to explore the capacity of the ENVI change detection tools.

 

Chris Mackey - Evaluating Chicago’s Urban Heat Island policy with remote sensing

Since the great Chicago heat wave of 1995, the city government has enacted a large number of policies under the pretense that it is combating its urban heat island. While much investigation on the subject thus far has focused on either the local benefits of policies or broad descriptions of the entire region, few studies have evaluated whether the small-scale benefits of policies have begun to collectively
impact the larger heat island. This study addresses this issue by using LANDSAT imagery to observe changes in Chicago’s vegetated and reflective surfaces in relation to changes in surface temperature since 1995. This is accomplished using two images from early June of 1995 and 2009 with similar air temperature and atmospheric conditions. Results indicate that reflective policies (especially reflective roofs) were very successful and increased the overall albedo of the city from 0.150 to 0.173 in addition to cooling certain neighborhoods by 2°C – 3°C. Vegetation policies produced noticeable results but their overall impact was slightly less impressive, generating a weaker correlation to temperature decrease and often affecting small areas instead of the larger region. Also, the fact that the number of vegetated pixels in Chicago decreased by 9,727 since 1995 indicates that an optimal current urban heat strategy would probably involve a restraint of development over vegetated areas in conjunction with good reflective policies.

 

Ziyan Chu - Urbanization and land cover in Connecticut

This project focuses on Connecticut in Northeastern US, using a Landsat MRLC image in July 16, 1991, May 8, 2001 and July 27, 2001 without cloud cover. We also downloaded the Landsat MRLC images in April and November 2001 to help us to classify the land cover types through seasonal effects. The major goal in this project: one is comparing different classification methods (i.e. supervised classification, multi-temporal and multi-spectral unsupervised classification) based on the image of Connecticut with a USGS land-cover classification of this region in 2001.  Such a comparison could help to test the accuracy of the land-cover classification methods according to the degree of agreement with USGS classification result. Another minor goal is to analysis the urbanization development in Connecticut using the win-out method on a Landsat MRLC image in July 16, 1991. The urbanization analysis is based on the land use change between 1991-2001. We will use the change detection to calculate the change of urban areas/density in 10 years and compare our results with USGS’s land cover change product between 1992 and 2001

 

Eric Fournier - Changes in urban areas of California expecting significant precipitation shortages

For this project I studied how anticipated changes in annual precipitation associated with global climate change will impact the availability of water for rapidly developing urban areas. The study site which I have selected is the State of California. A region of the country which is associated with both rapid rates of urbanization as well as a significant degree of existing water stress.

 

Kirsty Escalante - Exploring the effects of the 1997-1998 El Niño on the Moche Valley in Peru

My project aims to explore the effects of the 1997-1998 El Niño, one of the most devastating El Niño events in recorded history, on the Moche Valley in Peru, specifically its effect on agricultural land. I used a variety of different classification methods in order to gauge the level of change in agriculture between images taken in June 1996, prior to the event; May 1999, immediately after the El Niño; and February 2007, ten years after the event.

 

Matt Ramlow - Evaluating the effects of thermokarst lakes on climate in the Bering Land Bridge

This project used remote sensing techniques to analyze the climate effects of changes in thermokarst lakes throughout the Bering Land Bridge National Preserve in western Alaska. Thermokarst lakes are geological features formed by the melting of permafrost and are hotspots for methane emissions. This study detected the expansion of thermokarst lakes through thermokarst erosion and the disappearance of the lake through thermokarst failure. Landsat images were obtained for near anniversary dates in July and preprocess using various indices to mask cloud cover. Lakes were detected using the normalized difference water index (NDWI), normalized difference moisture index (NDMI), tasseled cap wetness function, tasseled cap brightness function, and albedo equation. Both an unsupervised and supervised classification with the same training regions across the two images will be performed to evaluate the pixels that have changed between 2002 and 2009. The supervised classification found that 28.81 km2 underwent thermokarst erosion while 19.55 km2 of lakes disappeared through thermokarst failure. This resulted in a net decrease of albedo across the lakes for the summer months and a net emission of 3.19 Gg CH4.

 

Chelsea Willett - Investigation of elevation and topography’s influence on N. Patagonian ice sheet

The Pleistocene orogeny of the Andes Mountains along the western edge of South America has been greatly influenced by the erosional and massive presence of the North and South Patagonian Ice Sheets. Erosion and mass wasting at the top of the mountain belt interacts with the tectonic isostatic balance and partially governs the rate of mountain uplift where the Nazca Plate is subducting below the South American Plate. In addition to its geologic time scale effects, however, ice sheet dynamics are also sensitive to shorter-term forcings, which have the potential to be carefully monitored by means of remote sensing.

Paleo-measurements from Greenland and Antarctic ice and sediment cores as well as modern weather station measurements have shown that the effects of modern global warming are amplified at the high latitudes. Melting ice not only contributes to a rising global sea level but also lowers the percentage of global landcover with a high albedo, potentially driving a positive feedback loop of further warming. Therefore, understanding the sensitivity of ice sheets and glaciers at the high latitudes is not just interesting science, it is pertinent as policies for emissions and global warming are being created and put into action.

For this project, I intend to use my new knowledge of remote sensing to investigate the current state of the North Patagonian Ice Sheet. Though just one of many choices, Landsat satellites detect in the visible, near-, mid- and thermal infrared and can be used to calculate important factors in the heat budget in and around this mountainous, glacial area. This project may be simple, but it is meant to explore a few of the applications of remote sensing in the southern Chilean Andes and form a basis for future research.

 

Wade Campbell - Remote sensing for archaeological prospecting in the Méma, Mali

Remote sensing is a technique of great utility and promise for archaeologists, enabling the exploration of regions that might be otherwise inaccessible to field expeditions while simultaneously encouraging the understanding of local landscapes beyond the limits of the excavation trench. Yet while archaeologists are aware of the benefits of remotely sensed data, the actual use of this data has historically been limited to a few specific techniques. The most prevalent of these has been the use of air – and now satellite – imagery to gain an overhead perspective that is particularly valuable for those interested in questions of survey and archaeological mapping. Through this simple and overwhelmingly visual approach to remote sensing much has been learned, the identification of moated sites in northeast Thailand through the analysis of World War II aerial reconnaissance photos or the use of declassified high-resolution CORONA images to identify ancient Middle Eastern canals being excellent examples of this method.

My project sought to utilize two basic tools of multispectral satellite imagery analysis – unsupervised and supervised classification – in an attempt to identify archaeological sites.  Two separate attempts at analysis were made, one focusing on sites in Peru and another on sites in Mali, West Africa. The specific regions were chosen due to the nature of the type of site in question and it was hypothesized that in each case, there would be sufficient differences between the sites and the surrounding landscape to permit the successful application of different classification schemes and the identification of specific sites.

 

Diana Connett - Using satellite data to estimate ground-level CO2 emissions

Because of the impacts of climate change, carbon dioxide is one of the most important gasses of our generation to measure.  Half of all anthropogenic CO2 emissions remain in the atmosphere for up to hundreds of years, no matter where they are emitted.  In order to facilitate national and international agreements on greenhouse gasses, emissions must be measured on a global scale.  On-the-ground technologies, like continuous emissions monitors and ambient air sensors, are used in many developed nations, but installing this expensive hardware on smoke stacks and in cities around the world may not be feasible or economically desirable.  The need for accurate and reliable measurements has inspired countries to look to remote sensing of CO2 and other greenhouse gases.  To date, four satellite instruments currently measure or have associated post-processing algorithms for carbon dioxide: GOSAT, SCIAMACHY, AIRS, and IASI, but none of these instruments are yet able to detect where emissions sources and sinks are prevalent at the ground level.   

 

Michele Dufault - Characterizing seasonal change in Russell Crater Dune System

In this project, I explore remote sensing methods for studying frost distribution and change over the course of the spring on a dune system in the southern hemisphere of Mars. The Russell Crater dunes are a particularly relevant site because some of the pole-facing slopes contain thin, roughly 10 m wide and deep ‘gullies’ that run down the length of the dune slope. There is reason to suspect that change in these gullies is driven by seasonal variation. Five hyperspectral IR CRISM images spanning mid-winter to mid-spring are obtained and analyzed. A supervised ‘maximum likelihood’ classification using user-defined regions of interest is compared with an unsupervised IsoData classification with inconclusive results. A frost index is developed to identify CO2 frost distribution over the scene. While the simple index tested here has generated some reasonable results, the index can certainly be developed further to serve as a much more robust diagnostic of frost presence. Finally, terrain parameters are estimated as a proof-of-concept for future frost studies that might study the correlation between frost distribution and physical terrain characteristics such as elevation, slope, and slope orientation.

 

Sarah Bahan - Characterization of the oilbird habitat near Cueva del Guácharo, Venezuela

Mt. Kilimanjaro and the Greater Amboseli Ecosystem, which stretches to the north of the iconic volcano’s slopes, represent one of the most biodiverse regions in East Africa. The system is also one of the few where large wildlife can still roam relatively free, and thus represents a critical conservation zone. The region is currently experiencing environmental change due to a synergistic combination of shifting ecological and social patterns that stretch back centuries, if not millennia. While local Maasai tribesmen are becoming more sedentary and shaping the landscape in new ways via changes in land use, the area is also undergoing many ecological changes – the most famous being the dramatic disappearance of Mt. Kilimanjaro’s glaciers in response to changing climatic conditions. In light of a rapidly changing global climate, elucidating the patterns and processes of past change in the region can aid our ability to predict how the ecosystem and its inhabitants might be affected in the future. To do so, I analyzed Landsat images taken from a period of 1987 to 2002, focusing primarily on changes in the surface area of Kilimanjaro’s glaciers, ecological classification of the mountain’s vegetation zones, and the extent of critical wildlife habitat in Amboseli National Park.