Monitoring Australian Rangeland Sites Using Landscape Function Indicators and Ground- and Remote-Based Techniques

If the goal for managing rangelands is to achieve a balance between production and conservation, then monitoring is essential to detect change and apply corrective action. In some range-land areas of northern Australia, monitoring has detected a tilt in the production-conservation balance towards excessive production. How big is this imbalance? Can it shift back? Robust monitoring is needed to answer these questions. The aim is to know what to monitor, and where. For example, to detect changes caused by livestock on rangeland forage production and soil erosion, indicators linking grazing disturbances to landscape function are needed, that is, indicators that signal how well landscapes are capturing, concentrating, and utilizing scarce water, nutrient, and organic resources. Studies in Australia and the USA document that simple vegetation and soil patch attributes can be measured as indicators of the 'state of health' of landscape function. For example, field and remote sensing-based grazing studies in Australia document that landscapes with a high cover of perennial plant patches function effectively to capture runoff water and nutrients in sediments, whereas landscapes with a low cover of these patches do not — they are dysfunctional — as indicated by large patches of bare soil. Aerial videography is proving to be a robust technique for measuring indicators of landscape function such as small patches of vegetation and the extent of bare soil. These indicators typically have a sigmoidal response to grazing impacts. We illustrate that if these indicators are measured on monitoring sites established near the sigmoidal 'point of inflection’ then small changes in these indicators can be detected.     

Fate of coliforms and pathogenic parasite in four full-scale sewage treatment systems in India

Global positioning systems (GPS) are increasingly being used for habitat mapping because they provide spatially referenced data that can be used to characterize habitat structure across the landscape and document habitat change over time. We evaluated the accuracy of using a GPS for determining the size and location of habitat patches in a riverine environment. We simulated error attributable to a mapping-grade GPS receiver capable of achieving sub-meter accuracy onto discrete macrophyte bed and wood habitat patches (2 to 177 m(2)) that were digitized from an aerial photograph of the Laramie River, Wyoming, USA in a way that emulated field mapping. Patches with simulated error were compared to the original digitized patches. The accuracy in measuring habitat patches was affected most by patch size and less by patch shape and complexity. Perimeter length was consistently overestimated but was less biased for large, elongate patches with complex shapes. Patch area was slightly overestimated for small patches but was unbiased for large patches. Precision of area estimates was highest for large (>100 m(2)), elongate patches. Percent spatial overlap, a measure of the spatial accuracy of patch location, was low and variable for the smallest patches (2 to 5 m(2)). Mean percent spatial overlap was not related to patch shape but the precision of overlap was lower for small, elongate, and complex patches. Mapping habitat patches with a mapping-grade GPS can yield useful data, but research objectives will determine the acceptable amount of error and the smallest habitats that can be reliably measured.     

Relationship Between Bird Abundances and Landscape Characteristics: The Influence of Scale

Scale is important to consider when investigating effects of the environment on a species. Breeding Bird Survey (BBS) data and landscape metrics derived from aerial photographs were evaluated to determine how relationships of bird abundances with landscape variables changed over a continuous range of 16 spatial scales. We analyzed the average number of birds per stop (1985–1994) for five songbird species (family Cardinalidae) for each of 50 stops on 198 BBS transects throughout six states in the Central Plains, USA. Land along each transect was categorized into six cover types, and landscape metrics of fractal dimension (a measure of shape complexity of habitat patches), edge density, patch density, and percent area were calculated, with principal components used to construct composite environmental variables. Associations of bird abundances and landscape variables changed in accordance with small scale changes. Abundances of three species were correlated with edge density and one with component I, which subsumes initial variables of patch density for urban, closed forest, open forest, and open country. Fractal dimension and component II (summarizing amount of closed forest versus open country) were associated with the most species. Correlation patterns of fractal dimension with northern cardinal (Cardinalis cardinalis) and painted bunting (Passerina ciris) abundances were similar, with highest correlations at intermediate to small scales, suggesting indirectly that these species thrive in areas where local habitat conditions are most important. Multiscale analysis can provide insight into the spatial scale(s) at which species respond, a topic of intrinsic scientific interest with applied implications for researchers establishing protocols to assess and monitor avian populations.     

Texture Analysis for Mapping Tamarix parviflora Using Aerial Photographs along the Cache Creek, California

Natural color photographs were used to detect the coverage of saltcedar, Tamarix parviflora, along a 40 km portion of Cache Creek near Woodland, California. Historical aerial photographs from 2001 were retrospectively evaluated and compared with actual ground-based information to assess accuracy of the assessment process. The color aerial photos were sequentially digitized, georeferenced, classified using color and texture methods, and mosaiced into maps for field use. Eight types of ground cover (Tamarix, agricultural crops, roads, rocks, water bodies, evergreen trees, non-evergreen trees and shrubs (excluding Tamarix)) were selected from the digitized photos for separability analysis and supervised classification. Due to color similarities among the eight cover types, the average separability, based originally only on color, was very low. The separability was improved significantly through the inclusion of texture analysis. Six types of texture measures with various window sizes were evaluated. The best texture was used as an additional feature along with the color, for identifying Tamarix. A total of 29 color photographs were processed to detect Tamarix infestations using a combination of the original digital images and optimal texture features. It was found that the saltcedar covered a total of 3.96 km² (396 hectares) within the study area. For the accuracy assessment, 95 classified samples from the resulting map were checked in the field with a global position system (GPS) unit to verify Tamarix presence. The producer's accuracy was 77.89%. In addition, 157 independently located ground sites containing saltcedar were compared with the classified maps, producing a user's accuracy of 71.33%.     

Cover estimation versus density counting in species-rich pasture under different grazing intensities

Two methods for monitoring of grassland vegetation were compared: visual estimation of plant cover (C) and plant densities counting (D). C and D were performed in monthly intervals for three vegetation growing seasons after imposing different grazing regimes on abandoned grassland in 1998. Species scores obtained from paired redundancy analyses (RDA) of C and D data were compared and Spearman's rank correlations were used to show if the two methods give comparable results. Results of C and D were highly correlated in the first two growing seasons only. In the third season, correlation was substantially lower as the sward structure was more heterogeneous due to creation of differently defoliated patches especially under extensive grazing. Presence of the same plant species with different habit in frequently and in infrequently grazed patches, reduced significance of Spearman's rank correlations. Cover estimation can fully substitute plant density counting in grassland with lower proportion of frequently and infrequently grazed patches only, but caution should be used when comparing different management regimes in long term analyses.     

Assessing and Monitoring the Health of Western Rangeland Watersheds

The most important function of watersheds in the western U.S. is the capacity to retain soil and water, thereby providing stability in hydrologic head and minimizing stream sediment loads. Long-term soil and water retention varies directly with vegetation cover. Data on ground cover and plant species composition were collected from 129 sites in the Rio Grande drainage of south-central New Mexico. This area was previously assessed by classification of Advanced Very High Resolution Radiometry (AVHRR) imagery. The classification of irreversibly degraded sites failed to identify most of the severely degraded sites based on size of bare patches and 35% of the sites classified as degraded were healthy based on mean bare patch size and vegetation cover. Previous research showed that an index of unvegetated soil (bare patch size and percent of ground without vegetative cover) was the most robust indicator of the soil and water retention function. Although the regression of mean bare patch size on percent bare ground was significant (p < 0.001), percent bare ground accounted for only 11% of the variability in bare patch size. Therefore bare patch size cannot be estimated from data on percent bare ground derived from remote sensing. At sites with less than 25% grass cover, and on sites with more than 15% shrub cover, there were significant relationships between percent bare soil and mean bare patch size (p < 0.05). Several other indicators of ecosystem health were related to mean bare patch size: perennial plant species richness (r = 0.6, p < 0.0001), percent cover of increaser species (r = 0.5, p < 0.0001) and percent cover of forage useable by livestock (r = 0.62, p < 0.0001). There was no relationship between bare patch size and cover of species that are toxic to livestock. In order to assess the ability of western rangeland watersheds to retain soil and water using remote sensing, it will be necessary to detect and estimate sizes of bare patches ranging between at least 0.5 m in diameter to several meters in diameter.     

Assessment Of Trampling Simulation Impacts On Native Vegetation In Mediterranean Sclerophyllous Forest

This study examines the effects of visitors on vegetation of natural semi-arid areas of Mediterranean basin using controlled experimental trampling. Impacts were assessed on days 1, 5, 15 and 30 examining presence of plant species, their cover and height in linear transects suffering a constant trampling intensity (25, 75, 100 and 200 passes/day). In addition, the study included the analysis of factors such as plant species, trampling intensity, the cumulative impact (day effect) and the type of footwear on plant variables. The results showed a decrease on plant cover and height influenced by cumulative (day) effect, an effect of trampling intensity and a different effect for each species. In contrast, shoe type affected only slightly the decrease in plant height. The evolution of plant cover revealed which species were most resistant to trampling. The disappearance of plant individuals during trampling appeared to be an impact index better than plant cover, because cover increased in some species when they were trampled thus introducing a bias.     

The effects of biotic and abiotic factors on the spatial heterogeneity of alpine grassland vegetation at a small scale on the Qinghai–Tibet Plateau (QTP), China

Understanding the complex effects of biotic and abiotic factors on the composition of vegetation is very important for developing and implementing strategies for promoting sustainable grassland development. The vegetation–disturbance–environment relationship was examined in degraded alpine grasslands in the headwater areas of three rivers on the Qinghai–Tibet Plateau in this study. The investigated hypotheses were that (1) the heterogeneity of the vegetation of the alpine grassland is due to a combination of biotic and abiotic factors and that (2) at a small scale, biotic factors are more important for the distribution of alpine vegetation. On this basis, four transects were set along altitudinal gradients from 3,770 to 3,890 m on a sunny slope, and four parallel transects were set along altitudinal gradients on a shady slope in alpine grasslands in Guoluo Prefecture of Qinghai Province, China. It was found that biological disturbances were the major forces driving the spatial heterogeneity of the alpine grassland vegetation and abiotic factors were of secondary importance. Heavy grazing and intensive rat activity resulted in increases in unpalatable and poisonous weeds and decreased fine forages in the form of sedges, forbs, and grasses in the vegetation composition. Habitat degradation associated with biological disturbances significantly affected the spatial variation of the alpine grassland vegetation, i.e., more pioneer plants of poisonous or unpalatable weed species, such as Ligularia virgaurea and Euphorbia fischeriana, were found in bare patches. Environmental/abiotic factors were less important than biological disturbances in affecting the spatial distribution of the alpine grassland vegetation at a small scale. It was concluded that rat control and light grazing should be applied first in implementing restoration strategies. The primary vegetation in lightly grazed and less rat-damaged sites should be regarded as a reference for devising vegetation restoration measures in alpine pastoral regions.     

The Influence of Linear Elements on Plant Species Diversity of Mediterranean Rural Landscapes: Assessment of Different Indices and Statistical Approaches

This paper mainly aims to study the linear element influence on the estimation of vascular plant species diversity in five Mediterranean landscapes modeled as land cover patch mosaics. These landscapes have several core habitats and a different set of linear elements -habitat edges or ecotones, roads or railways, rivers, streams and hedgerows on farm land- whose plant composition were examined. Secondly, it aims to check plant diversity estimation in Mediterranean landscapes using parametric and non-parametric procedures, with two indices: Species richness and Shannon index.Land cover types and landscape linear elements were identified from aerial photographs. Their spatial information was processed using GIS techniques. Field plots were selected using a stratified sampling design according to relieve and tree density of each habitat type. A 50×20 m² multi-scale sampling plot was designed for the core habitats and across the main landscape linear elements. Richness and diversity of plant species were estimated by comparing the observed field data to ICE (Incidence-based Coverage Estimator) and ACE (Abundance-based Coverage Estimator) non-parametric estimators.The species density, percentage of unique species, and alpha diversity per plot were significantly higher (p < 0.05) in linear elements than in core habitats. ICE estimate of number of species was 32% higher than of ACE estimate, which did not differ significantly from the observed values. Accumulated species richness in core habitats together with linear elements, were significantly higher than those recorded only in the core habitats in all the landscapes. Conversely, Shannon diversity index did not show significant differences.     

Estimation of Plant Diversity at Landscape Level: A Methodological Approach Applied to Three Spanish Rural Areas

Approaches linking biodiversity assessment with landscape structure are necessary in the framework of sustainable rural development. The present paper describes a methodology to estimate plant diversity involving landscape structure as a proportional weight associated with different plant communities found in the landscape mosaic. The area occupied by a plant community, its patch number or its spatial distribution of patches are variables that could be expressed in gamma plant diversity of a territory. The methodology applies (1) remote sensing information, to identify land cover and land use types; (2) aspect, to discriminate composition of plant communities in each land cover type; (3) multi-scale field techniques, to asses plant diversity; (4) affinity analysis of plant community composition, to validate the stratified random sampling design and (5) the additive model that partitions gamma diversity into its alpha and beta components. The method was applied to three Spanish rural areas and was able to record 150-260 species per ha. Species richness, Shannon information index and Simpson concentration index were used to measure diversity in each area. The estimation using Shannon diversity index and the product of patch number and patch interspersion as weighting of plant community diversity was found to be the most appropriate method of measuring plant diversity at the landscape level.     

Precision Measurements from Very-Large Scale Aerial Digital Imagery

Managers need measurements and resource managers need the length/width of a variety of items including that of animals, logs, streams, plant canopies, man-made objects, riparian habitat, vegetation patches and other things important in resource monitoring and land inspection. These types of measurements can now be easily and accurately obtained from very large scale aerial (VLSA) imagery having spatial resolutions as fine as 1 millimeter per pixel by using the three new software programs described here. VLSA images have small fields of view and are used for intermittent sampling across extensive landscapes. Pixel-coverage among images is influenced by small changes in airplane altitude above ground level (AGL) and orientation relative to the ground, as well as by changes in topography. These factors affect the object-to-camera distance used for image-resolution calculations. ‘ImageMeasurement’ offers a user-friendly interface for accounting for pixel-coverage variation among images by utilizing a database. ‘LaserLOG’ records and displays airplane altitude AGL measured from a high frequency laser rangefinder, and displays the vertical velocity. ‘Merge’ sorts through large amounts of data generated by LaserLOG and matches precise airplane altitudes with camera trigger times for input to the ImageMeasurement database. We discuss application of these tools, including error estimates. We found measurements from aerial images (collection resolution: 5–26 mm/pixel as projected on the ground) using ImageMeasurement, LaserLOG, and Merge, were accurate to centimeters with an error less than 10%. We recommend these software packages as a means for expanding the utility of aerial image data.     

Vertical gradients of mineral elements in Pinus sylvestris crown in alkalised soil

Alkalisation of soil has been assumed to be the principal cause of changes in vertical gradients of nutrients in Pinus sylvestris crown. The long-term influence of alkaline dust pollution ( ${\rm {pH}}_{{\rm H}_{2}{\rm O}}$  12.3–12.6) emitted from a cement plant on the element composition of soil and needles of Scots pine in different canopy layers was studied. In the polluted area, the pH of soils was >7, and high amounts of Ca, K and Mg were measured in the upper layers of soil (0–30 cm), while the mobility and solubility of some contaminants have decreased, nutrition processes have become complicated, and imbalance of mineral composition of trees was revealed. Reduced N and increased K, Ca and Mg concentrations in needles were observed in the heavily polluted area. Vertical gradients of elements and their ratios in canopies varied depending on the alkalisation level of soil. Needles on the upper-crown shoots had higher concentrations of N, C, Ca and Mg and lower concentrations of P and K compared to the lower layer of the crown. In the unpolluted area, higher concentrations of N, P, K and Ca were found in lower-crown needles and of C and Mg in needles at the top of the canopy. The P/N ratio below 0.125 indicated P deficiency in pines. The ratios N/Ca, N/Mg and N/K had significantly decreased, while the ratios Ca/Mg, K/Mg and K/Ca had a tendency to increase in heavily polluted sample plots. Magnitude of changes of element ratios indicates on the disbalances of availability and translocation of nutrients in the crown of trees.     

Quantifying landscape pattern and connectivity in a Mediterranean coastal settlement: the case of the Urla district,Turkey

This study was aimed at analyzing and interpreting changes in landscape pattern and connectivity in the Urla district, Turkey using core landscape metrics based on a 42-year data derived from 1963 CORONA and 2005 ASTER satellite images and ten 1/25,000 topographical maps (1963–2005). The district represents a distinctive example of re-emerged suburbanization in the Izmir metropolitan area. In order to explore landscape characteristics of the study area, nine landscape composition and configuration metrics were chosen as follows: class area, percentage of landscape, number of patches, patch density, largest patch index, landscape shape index, mean patch size, perimeter area fractal dimension, and connectance index. The landscape configurations in the Urla district changed significantly by 2005 in that the process of (sub-)urbanization in the study area evolved from a rural, monocentric urban typology to a more suburban, polycentric morphology. Agricultural, maquis-phrygana, and forest areas decreased, while the built-up, olive plantation and phrygana areas increased. There was nearly a fivefold increase in the built-up areas during the study period, and the connectivity of the natural landscape declined. To prevent further fragmentation, it is important to keep the existing natural land cover types and agricultural areas intact. More importantly, a sustainable development scenario is required that contains a green infrastructure, or an ecological network planning for conservation and rehabilitation of the vital natural resources in the study area.     

Characterizing the forest fragmentation of Canada’s national parks

Characterizing the amount and configuration of forests can provide insights into habitat quality, biodiversity, and land use. The establishment of protected areas can be a mechanism for maintaining large, contiguous areas of forests, and the loss and fragmentation of forest habitat is a potential threat to Canada’s national park system. Using the Earth Observation for Sustainable Development of Forests (EOSD) land cover product (EOSD LC 2000), we characterize the circa 2000 forest patterns in 26 of Canada’s national parks and compare these to forest patterns in the ecological units surrounding these parks, referred to as the greater park ecosystem (GPE). Five landscape pattern metrics were analyzed: number of forest patches, mean forest patch size (hectare), standard deviation of forest patch size (hectare), mean forest patch perimeter-to-area ratio (meters per hectare), and edge density of forest patches (meters per hectare). An assumption is often made that forests within park boundaries are less fragmented than the surrounding GPE, as indicated by fewer forest patches, a larger mean forest patch size, less variability in forest patch size, a lower perimeter-to-area ratio, and lower forest edge density. Of the 26 national parks we analyzed, 58% had significantly fewer patches, 46% had a significantly larger mean forest patch size (23% were not significantly different), and 46% had a significantly smaller standard deviation of forest patch size (31% were not significantly different), relative to their GPEs. For forest patch perimeter-to-area ratio and forest edge density, equal proportions of parks had values that were significantly larger or smaller than their respective GPEs and no clear trend emerged. In summary, all the national parks we analyzed, with the exception of the Georgian Bay Islands, were found to be significantly different from their corresponding GPE for at least one of the five metrics assessed, and 50% of the 26 parks were significantly different from their respective GPEs for all of the metrics assessed. The EOSD LC 2000 provides a heretofore unavailable dataset for characterizing broad trends in forest fragmentation in Canada’s national parks and in their surrounding GPEs. The interpretation of forest fragmentation metrics must be guided by the underlying land cover context, as many forested ecosystems in Canada are naturally fragmented due to wetlands and topography. Furthermore, interpretation must also consider the management context, as some parks are designed to preserve fragmented habitats. An analysis of forest pattern such as that described herein provides a baseline, from which changes in fragmentation patterns over time could be monitored, enabled by earth observation data.     

Assessing implications of land use and land cover changes in forest ecosystems of NE Turkey

Monitoring land use and land cover change (LUCC) and understanding forest cover dynamics is extremely important in sustainable development and management of forest ecosystems. This study analyzed the spatial and temporal pattern of LUCC in the Yaln?zçam and U?urlu forest planning units which are located in the northeast corner of Turkey. The investigation also evaluates the temporal changes of the spatial structure of forest conditions through the spatial analysis of forest-cover type maps from 1972 and 2005 using geographical information systems and FRAGSTATS^TM. As an overall change between 1972 and 2005, there was a net increase of 1,823 ha in forested areas, and cumulative forest improvement accounted for 2.06 %. In terms of spatial configuration, the landscape structure in the study area changed substantially over the 33-year study period, resulting in fragmentation of the landscape as indicated by large patch numbers and smaller mean patch sizes, owing to heavy grazing, illegal cutting, and uncontrolled stand treatments.     

Patchy distribution fields: A spiral survey design and reconstruction adequacy

A mathematical model was used to examine the effects of a spiral survey design on the adequacy of reconstructing patchy distribution fields. The model simulates fish or plankton patches (or gaps) of different shapes and spatial orientations, and acoustic surveys by a spiral of Archimedes; for comparison, surveys by parallel or zigzag transects are imitated. Adequacy of the reconstructed fields to those originally generated was evaluated by calculating their correlations (r). The mathematical experiments conducted showed that spiral surveys ensure, practically speaking, the same adequacy of field reconstruction (both in cases of immovable or movable patches) as do surveys by parallel or zigzag transects with greater sampling effort (overall path). In the case of a spiral survey, a patchy field can be reconstructed properly (r² = 0.70) if the overall survey path is not less than S/R_av = 20–30, where R_av is the autocorrelation radius averaged for various directions. Thus, the results obtained allow us to conclude that a spiral survey design is expedient in cases that the minimal duration of a survey is a decisive factor for its conduction and there is a priori information that no onshore-offshore gradients of fish density exist in a region under study.     

Sampling for landscape elements—a case study from Lower Saxony,Germany

The estimation of coverage, i.e., the proportion of the total area in a study region covered by a given target class, is essential to many aspects of environmental monitoring. We analyze and compare the efficiency of different sample-based approaches for the estimation of coverage of different land cover classes from aerial imagery in a case study in Lower Saxony, Germany on the basis of the estimated standard errors. A complete delineation of vegetation classes in n?=?279 aerial photo plots of 400 × 400 m thrown onto the study region of 1,117.7 km² in accordance with a systematic grid is compared to different configurations of line intercept sampling and clusters of points. The observation designs under study are characterized by different complexity and total size of the observation units and therefore also to the efforts related to yield a single observation. Especially for those classes that cover a relatively large proportion of the sampling frame, our results show that difference in performance between the different designs are negligible. A cluster of four transects of 200 m each allows estimating the area of land cover classes with high coverage with nearly similar precision as a complete mapping of fixed area plots of 16 ha each. Clusters of points show unexpected high precision for the estimated coverage of land cover classes with relatively high coverage.     

Effects of land-use changes on landslides in a landslide-prone area (Ardesen, Rize, NE Turkey)

Various natural hazards such as landslides, avalanches, floods and debris flows can result in enormous property damages and human casualties in Eastern Black Sea region of Turkey. Mountainous topographic character and high frequency of heavy rain are the main factors for landslide occurrence in Ardesen, Rize. For this reason, the main target of the present study is to evaluate the landslide hazards using a sequence of historical aerial photographs in Ardesen (Rize), Turkey, by Photogrammetry and Geographical Information System (GIS). Landslide locations in the study area were identified by interpretation of aerial photographs dated in 1973 and 2002, and by field surveys. In the study, the selected factors conditioning landslides are lithology, slope gradient, slope aspect, vegetation cover, land class, climate, rainfall and proximity to roads. These factors were considered as effective on the occurrence of landslides. The areas under landslide threat were analyzed and mapped considering the landslide conditioning factors. Some of the conditioning factors were investigated and estimated by employing visual interpretation of aerial photos and topographic data. The results showed that the slope, lithology, terrain roughness, proximity to roads, and the cover type played important roles on landslide occurrence. The results also showed that degree of landslides was affected by the number of houses constructed in the region. As a consequence, the method employed in the study provides important benefits for landslide hazard mitigation efforts, because a combination of both photogrammetric techniques and GIS is presented.     

Mapping the spatio-temporal distribution of key vegetation cover properties in lowland river reaches,using digital photography

The presence of vegetation in stream ecosystems is highly dynamic in both space and time. A digital photography technique is developed to map aquatic vegetation cover at species level, which has a very high spatial and a flexible temporal resolution. A digital single-lens reflex (DSLR) camera mounted on a handheld telescopic pole is used. The low-altitude (5 m) orthogonal aerial images have a low spectral resolution (red-green-blue), high spatial resolution (～1.9 pixels cm^?2, ～1.3 cm length) and flexible temporal resolution (monthly). The method is successfully applied in two lowland rivers to quantify four key properties of vegetated rivers: vegetation cover, patch size distribution, biomass and hydraulic resistance. The main advantages are that the method is (i) suitable for continuous and discontinuous vegetation covers, (ii) of very high spatial and flexible temporal resolution, (iii) relatively fast compared to conventional ground survey methods, (iv) non-destructive and (v) relatively cheap and easy to use, and (vi) the software is widely available and similar open source alternatives exist. The study area should be less than 10 m wide, and the prevailing light conditions and water turbidity levels should be sufficient to look into the water. Further improvements of the image processing are expected in the automatic delineation and classification of the vegetation patches.     

Vegetation Mapping for Change Detection on an Arid-Zone River

A vegetation mapping system for change detection was tested at the Havasu National Wildlife Refuge (HNWR) on the Lower Colorado River. A low-cost, aerial photomosaic of the 4200 ha, study area was constructed utilizing an automated digital camera system, supplemented with oblique photographs to aid in determining species composition and plant heights. Ground-truth plots showed high accuracy in distinguishing native cottonwood (Populus fremontii) and willow (Salix gooddingii) trees from other vegetation on aerial photos. Marsh vegetation (mainly cattails, Typha domengensis) was also easily identified. However, shrubby terrestrial vegetation, consisting of saltcedar (Tamarix ramosissima), arrowweed (Pluchea sericea), and mesquite trees (Prosopis spp.), could not be accurately distinguished from each other and were combined into a single shrub layer on the final vegetation map. The final map took the form of a base, shrub and marsh layer, which was displayed as a Normalized Difference Vegetation Index map from a Landsat Enhanced Thematic Mapper (ETM+) image to show vegetation intensity. Native willow and cottonwood trees were digitized manually on the photomosaic and overlain on the shrub layer in a GIS. By contrast to present, qualitative mapping systems used on the Lower Colorado River, this mapping system provides quantitative information that can be used for accurate change detection. However, better methods to distinguish between saltcedar, mesquite, and arrowweed are needed to map the shrub layer.