Proceedings of the 2007 SCGIS Annual Meeting
(June 25 - 27, 2007, Asilomar, Monterey, California, USA)
Track 2: Technologies and Methods
TUESDAY 11-12:30 Session 2A: Remote Sensing and Land-Cover Change
Application of GIS and Remote Sensing Techniques to Assess the Status of Forest Types of the Southern Eastern Ghats, India
This work was aimed to map the existing vegetation types of the region with focus on land use/land cover patterns using satellite imageries and GIS. This was particularly in and around the natural forest patches. The Eastern Ghats constitute an important biogeographic region of the Indian subcontinent. This has been identified as a major centre of unique plant diversity. Base maps with information such as forest patch, transportation network, village location, drainage, administrative boundaries were converted into digital formats using ArcView and ArcGIS software. Indian remote sensing satellite (IRS) with
Haqiq Rahman Rahmani
Eastern Forest Cover Change Detection
Eastern Forest in Afghanistan contains some of the remaining forest in the Himalayan mountain region and is critical to biodiversity. Because of the drought, the conflict in the country and the high rate of deforestation in the region, the wildlife in this area is under threat. The instability and absence of government has encouraged the logging enterprises to devastate the forest during the conflict period. Preliminary UNEP (2002) Landsat analysis found that forest cover in Nuristan has decreased by 53 percent and in Kunar by 29 percent between 1977 and 2002. Residents predict similar losses for the forested regions in the provinces of Paktia, Khost and Paktika. If this rate of deforestation continues, estimates suggest that most of the remaining forested valleys could be completely stripped of trees within five to ten years. So WCS is planning a forest cover change detection under its Eastern Forest Complex Program. This study will unite analysis of forest cover change over the past 3 decades with fieldwork on current wildlife distributions to prioritize conservation based on species threats.
Analysis of Factors of Forest Cover Change on Mt. Cameroon Using Remotely Sensed Data
The proximate causes of forest cover change on Mt. Cameroon are still poorly understood despite the fact that it is habitat for rare and endemic species. Information on driving forces of forest cover change is very vital in order to assist governments and conservation organizations in developing strategic plans.
The aim of this study was to investigate the forest cover dynamics of Mt. Cameroon in one subperiod (from 1986-2002) and to quantify some driving forces of deforestation. Digital data on driving forces of forest cover change, like roads, settlements, forest edges, rivers, and satellite imageries were employed in the study. Remote sensing and GIS tools were used to map, detect and quantify changes in land cover, inclusive forest cover and also to quantify the factors that are influencing forest cover change in the area.
The study mapped out the Mt. Cameroon region first through an unsupervised classification carried out prior to field survey. A supervised classification was then carried out and the field survey data was used to validate the results of the Landsat TM 2002. A supervised classification was carried out on Landsat MSS 1986 and archival data was used to validate the result. Change detection was based on UNION (a form of GIS overlay in ArcView 3.2a) of the two maps produced from the Landsat MSS of 1986 and Landsat TM of 2002. A spatial query of the output table from the UNION revealed the land cover change types and their direction of change which were further quantified.
The study showed a general increase in agricultural land and a decrease in forest cover. Further analyses were carried out to determine the factors that could be used to explain the forest cover loss (deforestation). A spatial model that was based on logistics regression analysis revealed some driving forces that influence deforestation in the study area. Distances to roads, settlement and forests edge showed statistical significance with deforestation while distances to rivers were not statistically significant with deforestation. The pattern of land cover showed that deforestation tended to cluster in some areas, occurred in a diffuse manner in other areas and occurred in a linear manner in some areas.
Land cover maps, the change results, the spatial models and the land cover pattern that have been developed in this study will permit the identification of areas where forest conversion is likely to occur in the near future. This information if used with information on the spatial distribution of species habitat will help planners in developing management plans that will prioritize conservation efforts.
Comparison of Satellite and Ground Vegetation Indexes to Assess Erosion in a Mediterranean-Climate Watershed
Miguel A. González-Botello*, Stephen H. Bullock and J. Mario Salazar-Ceseña
Centro de Investigaciones Científicas y de Estudios Superiores de Ensenada, Laboratorio de Ecología Terrestre, Departamento de Biología Experimental y Aplicada
*Corresponding author: firstname.lastname@example.org
Soil erosion across a 5,000 km2 watershed in northwest Baja California has recently been estimated using a GIS implementation of the Revised Universal Soil Loss Equation model "RUSLE" (Smith et al. in press), for the purpose of comparing lateral and vertical fluxes of carbon. The protective factor of cover (vegetation and litter) was estimated as a linear function of the maximum Normalized Difference Vegetation Index, based on a weak correlation between cover and NDVI in a small sample from southern France (DeJong 1994). Also, the satellite index provides complete coverage of the watershed.
Our project undertakes field sampling of the cover factor and evaluation of its relation to NDVI, EVI and terrain variables (exposure, slope and elevation) and the relation of NDVI to different components of cover (canopy cover, vegetation height, crown overlap and species diversity or composition).
Location of the field sites involved seven steps with intensive use of GIS and satellite imagery to map present shrublands and accessible areas and to locate potential sites. This process involved integrating a 30 m digital elevation model (from INEGI, Mexico), imagery from GoogleEarth 4 (Digital Globe, 2002-2006), orthophotos (INEGI's Digital Orthophoto Web Service), existing digital land use maps, paper road maps and Landsat imagery. The GIS was created using ArcGIS 9.2, (Spatial Analyst, 3D Analyst). With KMLer v1.2 extension, extensive interaction between Google Earth v.3.0 and ArcGIS 9.2 was possible. Also we incorporated IDRISI Andes v.15 in order to process the satellite image and generate NDVI.
Field measurements were based on 30 m line transects within which we sampled drip height, surface cover and aerial cover. Landsat data from late April to early May 2007 will be processed for NDVI (and EVI), and variance among years of contrasting rainfall will also be analyzed (2001, 2003, 2005). Terrain variables will be calculated from the digital elevation model.
The field work (~70 sites) is being completed in late May. Preliminary results, using NDVI from early May 2005, show that field values of the RUSLE C are far below the NDVI estimates (i.e., there is much more cover than estimated by NDVI), and the difference is clearly due to a high coverage of the soil surface not to interception by branches and leaves.
Application of GIS for Conservation and Sustainable Management of Wetland Ecosystem in Nepal
Main objective of this paper is to show role of GIS, RS and GPS technology for the sustainable management of natural resources and biodiversity conservation. This is based on the application of GIS introduced by WBC Nepal for conservation and sustainable management of wetland and Ecosystem in Nepal. To depict the application of GIS, this paper is divided into three parts: Role of GIS, RS and GPS for Wetland inventory, Detail mapping of critical Wetland areas and GIS analysis for their sustainable management. The first part describes the application of GIS, RS and GPS to perform wetland inventory of the Terai Region (physiographically Nepal is divided into three parts: Terai, Hill and Mountain, see map no. 1 for physiographic division of Nepal and no. 2 for wetlands of Terai region) of Nepal. For this purpose vector data of land use pattern was used as a reference information and field verification was done with Satellite Imageries and GPS. Among these Terai wetlands only four are listed in Ramsar site (see map no. 3 for the location of Ramsar site in Nepal). For the sustainable management of Ramsar site, WBC Nepal realized the need for detailed mapping of these wetlands. The second part of this paper describes the application of GIS for detailed mapping of Ramsar sites of Nepal (see map no. 4, 5, 6 & 7) for individual wetland listed in Ramsar site). Though the wetlands are known as the biological supermarket, awareness of the dependent community and local stockholders plays the vital role for their sustainable management and conservation. For this, WBC Nepal has prepared an Action Plan on Conservation and Sustainable Management of Beeshazar Wetland System, Nepal. Last part of this paper describes the application of GIS tools to perform geographical analysis for different natural and socioeconomic component of wetland area and surrounding dependent community (see map no. 8, 9, 10, 11,12 &13).
TUESDAY 2-3:30 Session 2B: Wildlife Methods
Using GIS for Wildlife Corridor Design in Arizona
Authors: Emily Garding, Paul Beier, and Daniel Majka
Habitat loss and fragmentation are the most important threats to maintaining populations of many of Arizona's wildlife species. We have developed Linkage Designs for 8 areas within Arizona where wildlife habitat connectivity is threatened by transportation projects and other landscape alterations. For each Linkage Zone, we identify focal species, including species that are closely related to ecosystem function or sensitive to fragmentation. We perform Least Cost Corridor Analysis to estimate the optimal location of a landscape linkage between core protected areas based on estimated relationships between the focal species and 4 landscape features, namely: vegetation/land use, topographic features, elevation, and distance to roads. The final Linkage Design consists of the union of all the habitat patches identified as important for movement of each focal species. Each Linkage Design includes detailed recommendations for locations for highway-crossing structures, optimal types of structures, and other opportunities to enhance wildlife movement across potential barriers. These Designs can help agencies, planners, private landowners and other stakeholders work together with a coordinated approach toward conservation and highway safety goals. The linkages assessment is a step forward in protecting Arizona's wildlife as the state addresses the challenges associated with accommodating for the growth of Arizona's population and economy.
Using Phone Call Reports to Assess the Relative Abundance of Urban Mesocarnivores
The focus of the present study was to determine the relative abundance of raccoons, opossums, and skunks using nuisance phone call reports from the public. I used verified reports made over four years (2000, 2003, 2004, and 2005) to Brevard Animal Services in Brevard County, Florida, as indicators of mesocarnivore presence. I included the calls in ArcMap 9.1 and created nonadjacent grid cells to perform quadrat sampling. I combined land cover and human demographic information within quadrats and used stepwise multiple linear regression analyses to determine which land cover categories best predicted the relative abundance of raccoons, opossums, and skunks in urban areas. I tested areas of land covers as the independent variables and verified locations of animals per capita as the dependent variables. Raccoon abundance was high in medium and high density residential and industrial areas but low in wetland areas. Opossum abundance was high in medium density and high density residential areas but low in areas with barren land. Skunk abundance was high in commercial, medium density residential, and upland nonforested areas but low in high density residential areas. The results indicate that these animals are using habitats based upon availability as opposed to selecting specific land covers. Results also show techniques to remove bias associated with phone call reports appear successful. This is a low cost method that shows promise to determine the relative abundance of mesocarnivores in urban areas. However, data from phone call reports (including road-kill locations) are often not recorded in a format that is ready for analysis. Increased interagency communication is needed to ensure data are recorded in a standardized format that is compatible with a GIS.
Using GIS and Road Kill Data to Identify Habitat Characteristics Associated with North American Badger Movement Patterns and to Evaluate Effectiveness of Corridor Models
Small populations of North American badgers, Taxidea taxus, live in fragmented habitats within the San Francisco Bay Area and Monterey Bay Counties in Northern California. Due to increasing fragmentation of badger habitats, it is important to identify potential corridors that badgers may utilize to facilitate movement between the habitat patches. Using GIS, a habitat suitability model for badgers was created using soil, vegetation, slope, urban areas, and road layers. We have found that the choices made in ranking and weighing habitat characteristics need to be quantified when developing a habitat suitability model. We often assume that species will choose to travel through habitat characteristics that are typically found in their home ranges, which may not be the case. Badger road kill data was used to identify habitat characteristics selected by badgers while traveling outside of their home range. A Least-Cost Corridor analysis was performed on the cost surface layers to identify potential corridors within the region. Road kill data was then compared with the delineated corridors to test the model and identify the primary areas currently utilized by badgers for implementing future corridor plans. Road kill data was also used to evaluate alternative models by comparing the distance from road kill locations to the location of a road predicted by various other models. Road kill data was obtained from County Animal Control agencies and by County and State park field officers. In cooperation with data sharing, maps of the resulting model will be provided to the contributors of the road kill data for their particular parks and regions of concern. As an example of how this model could be used, five badgers have been killed by vehicles along Reservation road in Monterey County, which runs through existing badger habitat. Comparison with the corridor model shows this is one of the primary areas to implement a future corridor to facilitate badger movement among the fragmented habitats.
Integrating Microsensing Technologies with Agent-Based Simulations for the Study of Individualized Diversity in Ecological Communities
Agent-based and individual-based models that simulate the dynamics of individual members of a community often reveal remarkable patterns of emergent organization at collective scales. Such modeling results are difficult to calibrate, update, and validate on the basis of empirical data since observations of the natural world are seldom recorded with a sufficient degree of spatiotemporal detail to properly inform the models. However, constraints on the empirical study of many interacting individuals and their environments are beginning to erode as distributed electronic sensing and communications devices become capable of continuously monitoring their locations and attributes. This report discusses the integration of spatiotemporal data from distributed sensors and remote sensing devices with agent-based simulations for the study of mobile individuals and their collective dynamics.
Characterizing Savanna Herbivore Dynamics in Northern Kenya: Systematic Surveys and GIS Applications
Ecosystem processes in African savannas can be better conserved if management is based on sound understanding of wildlife dynamics in livestock-dominated landscapes.
We used simple GIS techniques to characterize factors influencing the dynamics of large herbivores on three land-use types in Laikipia District, a nonprotected savanna region in northern Kenya. The three land use types were commercial ranches that favor wildlife, communal "group ranches" practicing pastoralism, and the remainder ("transitional" properties). For 21-year time series of nine wild and two domestic species, linear model selection was used to ascribe between 45% (Grant's gazelle) and 95% (plains zebra) of observed variation in biomass density to land use, rainfall-dependence, density-dependence, and trends over time.
Strongly opposing patterns of variation across the landscape in wildlife and livestock densities affirmed the primacy of land use among factors influencing wildlife abundance in nonprotected areas. Rainfall limited densities of only the dominant grazing species throughout the monitoring period (plains zebra and cattle) and of most other species while their densities were high. Regulating effects of density were detected only for the dominant wild grazing and browsing species (zebra and giraffe). All but two wild species (zebra and Grant's gazelle) declined on at least one land-use type for reasons that varied among land uses.
This paper also demonstrates the utility of sample counts using systematic reconnaissance surveys as a relatively cheap yet effective means of wildlife surveys, hence, management. Maintaining higher wild species diversity in the landscape will depend on the creation of a network of unfenced conservation areas in which livestock densities are persistently low or zero, which are sufficiently large to act as "sources" of wild species that are prone to displacement by humans and livestock and which generate benefits to community members that exceed opportunity costs.
TUESDAY 4-5:30 Session 2C: Landscape Analysis
Landscape Species and Threat Distributions in the Core Area of the Western Forest Complex, Thailand
The Western Forest Complex (WEFCOM), an area of 18,000 km2, is one of the largest protected area systems in Southeast Asia. The WEFCOM's core area is also a natural world heritage site named Huai Kha Khaeng Wildlife Sanctuary. Since 2005 WCS Thailand has worked with the Thai government to use a wildlife-based approach called the Living Landscape Program to measure the success of conservation through status of landscape species, threats, and types and degrees of interventions. The landscape species for the area are elephants, tigers, rufous-necked hornbills, and otters. The key intervention is to establish information-based patrolling system that equips rangers with GPS, digital camera, and standardized data forms. The main data coming into the database includes patrol tracks, GPS points of landscape species, and threats. From one year of data recording and preliminary analysis we have found that the approach has helped the sanctuary manager to plan for better protection and management. More comprehensive analysis approach is needed to estimate threat intensity and landscape species distribution.
Assessing the Effect of Roads on Landscape Connectivity
For decades, ecologists and land managers have recognized the importance of maintaining ecosystem connectivity in order to sustain native wildlife populations. However, evaluating connectivity among habitat patches remains a difficult task, in part because several complex questions must be answered in order to adequately frame a realistic assessment of connectivity. For example, what type of habitat are we interested in connecting? Seasonal migrations between summer and winter habitat or daily movements between suitable forage and denning locations? How robust are the data that characterize these habitat patches? Do they reflect actual animal locations or detailed resource selection functions? Or are they merely subjective assignments of habitat quality to land cover types? How strongly does human disturbance influence animal movement? How connected or fragmented was the natural landscape before human disturbance? Clearly, there are so many different scenarios to consider and different types of connectivity to evaluate that any one of these questions could overwhelm and stall land managers with a dizzying array of potentially never-ending analyses.
Nevertheless, these challenges do not exonerate land managers from their responsibility to evaluate habitat connectivity within timeframes determined by planning processes and decision-making. The purpose of this document is to demonstrate an innovative approach for evaluating habitat connectivity in the absence of answers to the questions above. It is not our intent to accurately model any individual species specifically, but rather, by relying on several basic assumptions, to demonstrate how ecosystem connectivity can be assessed within practical timeframes. We expect that public land managers will either have more data available to them for any particular species of conservation concern, or, through such analysis, they will identify knowledge gaps necessary to fill to assess impacts on connectivity. Armed with more specific data, this model can allow public land managers to better evaluate impacts of various land uses on habitat connectivity. For the purposes of this demonstration, we focus on the Kenai National Wildlife Refuge (KNWR) in Alaska and, specifically, the impact of oil and gas infrastructure on habitat connectivity for a hypothetical species with mappable habitat affinities and aversions.
Savrina Flora Carrizo
Trajectories to Extinction
There are several open questions regarding the spatial patterns of population declines, in particular, how does the geographic range and abundance structure change with a declining population? We aim to find empirical evidence for these spatial patterns. (A flip side of the same question can provide insights into the nature of invasions based on increasing populations.) We also aim to test the efficacy with which different monitoring strategies are able to detect changes in both abundance and occupancy. An understanding of the spatial patterns of decline can also tell us where conservation efforts would be best focused, for example, whether a landscape approach would be necessary. The aims of the study will be discussed in the context of the data collected from the U.S. Breeding Bird Survey.
Gergely Torda, Zsolt Molnár, Ferenc Horváth, Eszter Lellei-Kovács, and Bálint Czúcz
Adaptation Capacity of Pannonian Grassland Habitat Types
Adaptation capacity of two typical and widespread Pannonian grassland habitat types, Mesotrophic meadows and Artemisia salt steppes to predicted climate change was assessed based on the Habitat Database of Hungary (MÉTA database, http://www.novenyzetiterkep.hu/meta/en/index.shtml). For both habitat types, an index was calculated to describe the potential of the species of the habitats to adapt to climate change via migration, i.e., the connectedness with surrounding potentially colonized sites. As a metric of spatial coherence, we used a pixel-wise proximity index (PWP) based on the distance-dependent area of adjacent habitats with similar species pools. The index assumes that the potential migratory stepping-stone character of surrounding habitat patches decreases with the reciprocal of the square distance measured from the examined habitat and increases with the dimensions of the surrounding patches. Results were compared with the coverage of the National Ecological Network, which forms part of the Pan-European Ecological Network, to reveal the extent to which these legally protected sites support potential adaptive migration for the examined habitat types.