mapping wetlands: a comparison of two different approaches for new ...

1 downloads 0 Views 1MB Size Report
1350 Regent Street, Fredericton. New Brunswick, E3C 2G6, Canada. Abstract: Wetlands have an important role in ecosystem function and biodiversity. Effective.
WETLANDS, Vol. 27, No. 4, December 2007, pp. 846–854 ’ 2007, The Society of Wetland Scientists

MAPPING WETLANDS: A COMPARISON OF TWO DIFFERENT APPROACHES FOR NEW BRUNSWICK, CANADA Paul N. C. Murphy1, Jae Ogilvie1, Kevin Connor2, and Paul A. Arp1 1 Nexfor-Bowater Forest Watershed Research Centre Faculty of Forestry and Environmental Management P.O. Box 44555, University of New Brunswick Fredericton, New Brunswick, E3B 6C2, Canada E-mail: [email protected] 2

Fish and Wildlife Branch New Brunswick Department of Natural Resources Hugh John Flemming Forestry Centre 1350 Regent Street, Fredericton New Brunswick, E3C 2G6, Canada Abstract: Wetlands have an important role in ecosystem function and biodiversity. Effective management of wetlands requires accurate and comprehensive spatial information on location, size, classification, and connectivity in the landscape. Using a GIS, two provincial wetland maps were compared with regard to their areal correspondence across different ecoregions of New Brunswick. The first consisted of discrete wetland units (vector data) derived from aerial photo interpretation. The second consisted of wet areas modeled by a newly developed depth-to-water index with continuous coverage across the landscape (raster data). This index was derived from a digital elevation model and hydrographic data. The relative advantages and disadvantages of the two approaches were assessed. The two maps were generally consistent with most discrete wetland areas (51%–67 %) embedded in the 0– 10 cm depth-to-water class, verifying the continuous modeling approach. The continuous model identified a larger wetland area. Much of this additional area consisted of riparian zones and numerous small wetlands (, 1 ha) that were not captured by aerial photo interpretation. Unlike the discrete map, the continuous model showed the hydrological connectivity of wetlands across the landscape. Both approaches revealed that topography was a major control on wetland distribution between ecoregions, with more wetland in ecoregions with flatter topography. Key Words: GIS, riparian zones, soil mapping, soil wetness index, topographic modeling, topography, wetland management

INTRODUCTION

2000). For example, this information is needed for setting policy targets for the protection and conservation of wetlands, formulating specific wetland management objectives, developing guidelines for best wetland management practices, and integrating scientific wetland information with socio-economic considerations. Spatially explicit wetland inventories are a very instructive source of information for planners. As part of a GIS, these data can be integrated with other data layers to allow more effective planning and management (Turner et al. 2000). In particular, digital elevation models (DEMs) and other hydrographic data can allow information on surface flow pathways and the hydrological connectivity of wetlands to be derived (Moore et al. 1991, Ryan et al. 2000, Murphy et al. 2006). Turner et al. (2000) noted that wetlands are lost or threatened because of information failures linked to the complexity and

Wetlands are critical elements in the landscape in terms of habitat and biodiversity, regulation of watershed hydrology, and mediation of biogeochemical cycles (Detenbeck et al. 1999, Haag and Kaupenjohann 2001, Bhatti and Preston 2006). They can have ecological, cultural, historical, and economic value to society (Mitsch and Gosselink 2000). If development is to be sustainable, it is important to manage the effects of human operations on wetlands in activities such as forestry, agriculture, recreation, and urban development (Christensen et al. 1996, Findlay and Bourdages 2000, Turner et al. 2000). A basic need for setting clear policy objectives and effective management with regard to wetlands is the development of knowledge inventories concerning the location, size, and type of wetlands (Turner et al. 846

Murphy et al., COMPARISON OF WETLAND MAPPING APPROACHES ‘invisibility’ of spatial relations among groundwater, surface water, and wetland vegetation. The mapping and classification of wetlands is an abstraction of the reality of wetland distribution. In reality, wetlands form one part of the continuum of soil hydrological conditions across the landscape. There is some disagreement over what constitutes a wetland (Turner et al. 2000), but the Canadian Wetland Classification System (NWWG 1997) defines them as that part of the continuum that is ‘‘saturated with water long enough to promote wetland or aquatic processes as indicated by poorly drained soils, hydrophytic vegetation and various kinds of biological activity that are adapted to a wet environment.’’ Two distinct approaches to wetland mapping can be distinguished. As described by Grunwald (2006) with respect to soil mapping, the first, and more traditional approach, is to define crisp map units. These discrete mapping units are associated with representative attributes (binary membership functions) such as wetland type and are defined by abrupt changes at the border of the unit; in this case, from wetland to non-wetland. These units are typically mapped as vector (polygon) data in a GIS. Some of the drawbacks of this approach are that it ignores spatial variability within the units and does not represent ‘‘fuzzy’’ boundaries, where properties change gradually. The most common method for producing such discrete wetland maps is interpretation of remotely sensed data. Aerial photography is a widely used data source (Johnston and Meysembourg 2002, Kent and Mast 2005, U.S.F.W.S. 2007). The discrete wetland map used in this study was derived from aerial photo interpretation (DNR 2006). The second approach, which has been greatly aided by the development of GIS technologies, is the continuous field model in which soil properties are displayed as pixels with continuous coverage across the landscape. This spatial model can describe gradual change in properties, spatially, and is typically mapped as raster data in a GIS. A popular approach has been to derive topographic indices from a DEM and relate these to watershed properties that are of interest, such as soil drainage conditions or moisture content. Numerous studies have used a topographically derived soil wetness index (SWI) (Hornberger and Boyer 1995, Iverson et al. 1997, Boerner et al. 2000, Gu ¨ ntner et al. 2004). The SWI is an integral part of topographically based hydrologic models such as TOPMODEL and TOPOG (Hornberger and Boyer 1995). Tomer et al. (2003) used the SWI to map riparian zones and identify zones suitable for riparian buffers.

847

This study involved a comparison, using a GIS, of two wetland maps available to users in New Brunswick with provincial coverage. The first takes the discrete map unit approach, deriving wetland location, boundary, and classification from aerial photo interpretation. The second takes the continuous mapping approach, deriving a depth-to-water value for each pixel from a DEM and hydrographic data using a new GIS-based algorithm developed recently at the University of New Bunswick (UNB). This value serves as an indicator of the tendency of the soil to be saturated. The second data set has only recently become available, where as the first has been recently updated. The study was carried out to verify the ability of the continuous model to predict wetland location and extent by comparison with an independently derived wetland inventory. The relative advantages and disadvantages of the discrete versus continuous approach to wetland and soil mapping were also investigated. To the authors knowledge, such a comparative study of the discrete versus continuous approach to wetland mapping has not been carried out before. TWO WETLAND DATA SETS Discrete Wetland Map The New Brunswick Department of Natural Resources (DNR) wetland inventory was derived from interpretation of color ortho-rectified digital aerial photographs, updated most recently in 2006 (Table 1) (DNR 2006). Wetlands were mapped using photography from the period 1993–2002 at 1:12,500 scale. Wetlands were mapped in vector format as discrete units (polygons) with interpreted attributes associated with each discrete map unit. Aerial photos were interpreted at a resolution capable of identifying wetlands of approximately 1 ha or larger. However, the map also includes wetland units smaller than 1 ha, where these were distinct features on the landscape and were easily identified by the interpreter. As a result, it can be expected that the 10 m resolution continuous mapping process will identify more wetland areas of less than one ha in size. One of 16 wetland classes was assigned to each wetland unit by the interpreter based on characteristics visible in the aerial photos. Aerial photos were generally acquired in mid to late summer (leaf-on conditions). Continuous Wet Area Map Province-wide high resolution (10 m) maps depicting surface water flow channels and wet areas

848

Table 1.

WETLANDS, Volume 27, No. 4, 2007

Summary of the two wetland inventories used in this study. Provincial wetland inventory

Type of data Wetlands mapped as

Vector Discrete units (polygons)

Origin of data

Aerial photo interpretation

Resolution Classification

1 ha (minimum wetland size) Wetland type: aquatic bed, beach, bog, dune, emergent wetland, fen, forested wetland, non-productive forest, rock outcrop, salt marsh, shrub wetland, salt lake, shrub wetland, tidal flat, vegetated dune, wetland

were derived from a DEM and hydrographic data using a newly developed GIS process (Murphy et al. 2006). In these maps, wetlands are indicated by an index of soil wetness, expressed as depth-to-water, with continuous raster coverage across the landscape. Most similar studies have made wide use of a compound topographic index called the Soil Wetness Index (SWI) (Moore et al. 1991). There are several forms of this topographic index, and it is given different names, but the basic formula is given as SWI ~ ln(A = S) where A is the area draining to the point in the landscape, and S is the slope at the point. The wet area map examined in this study was also derived from a DEM, but using a new and different approach. A provincial DEM was acquired from New Brunswick DNR. Digital elevation data was derived photogrammetrically from 1:35,000 digital aerial photographs to produce a near regularly spaced DEM grid with north-south sampling grid lines of roughly 70 m spacing (0.1 m vertical resolution) and a density of 2.84 points ha21. A resampled DEM, at approximately twice the original point density (6.22 points ha21), was derived from the original DEM using a process known as ‘‘TIN Random Densification’’, developed at the Department of Geodesy and Geomatics at UNB (Pegler et al. 2000). This process minimizes problems of ‘‘ridging’’ associated with the use of regular photogrammetric sampling lines, improving the performance of the DEM for hydrologic applications. The DEM was then geo-spatially interpolated to produce a 10 m raster using an Inverse Distance Weighted (IDW) technique involving 12 nearest neighbor points and a power value of 2 (Watson and Philip

Soil wet-area map Raster Continuous depth-to-water as indicator of tendency to be saturated Derived from DEM and hydrographic data 10 m (raster resolution) Predicted depth-to-water class: 0–10 cm 10–25 cm 25–50 cm 50–100 cm 100–200 cm . 200 cm

1985). All of this was done with ArcView GIS software (ArcView GIS 3.2, ESRI Inc., 1992–1999). It should be noted that this DEM processing does not incorporate any new topographic information into the DEM, but makes it more suitable for deriving hydrologic information at the scale required (, 20 m resolution). Pre-existing hydrographic information was incorporated into the DEM by lowering the elevation of grid cells corresponding to surface water features to produce a ‘‘hydrologically corrected’’ DEM for which the derived flow conforms with all the known flow paths (Olivera 1996, Saunders and Maidment 1996, Simley 2004). The FILL function of ArcView was used to create a depressionless DEM. The flow direction algorithm D8 (deterministic-8 node algorithm) (Hornberger and Boyer 1995) was then used to derive a flow accumulation network. The resulting flow network was merged with the already mapped hydrographic coverage to produce a single improved hydrographic layer. Large depressions were derived from the DEM and the improved hydrographic layer was combined with these depressions to produce a hydrographic source layer. A wetness index value, expressed as approximate depth-to-water, was determined for each cell in the landscape using an iterative function. This function finds the least slope path from each cell in the landscape to a source cell, based on the cumulative value of slopes along the possible paths. It then assigns that landscape cell to the particular source cell (surface water feature) and assigns the cumulative slope value to the landscape cell. All surface water features (lakes, streams, rivers) are assumed to have a value of 0. The depth-to-water value approximates the elevation difference between the cell in the landscape and the assigned hydrologic

Murphy et al., COMPARISON OF WETLAND MAPPING APPROACHES source cell. The index value reflects both the distance from a source and the slope of the land surface between the landscape cell and the source. Lower depth-to-water values indicate wetter soils. Values tend to increase away from the source into the landscape, indicating drier soils. It increases more rapidly in steeper terrain (higher slope values) and more slowly in flatter terrain (lower slope values). The depth-to-water value can be interpreted as a relative measure of the tendency of the soil to be saturated: cells with a low value can be expected to have water at or near the surface for a significant part of the year. In this way, the depth-to-water can be used to map wetlands continuously across the landscape. It should be noted that the water table depth in upland (non-wetland) soils of any watershed would depend on factors that may vary widely, such as the sediment and soil hydraulic properties, heterogeneities and isotropies in those properties, climate, and vegetation. Of course, the water table may also rise and fall seasonally and over longer time scales. Therefore, relating the depth-to-water index to actual water table depth in upland soils would need to be verified by region. As a first approximation, it is likely to be closest to actual water table depths in watersheds with free-draining and highly conductive soils and sediments. ANALYSIS The areal correspondence of depth-to-water classes used to identify wetlands in the continuous map and wetland units in the discrete map was compared for the entire province and by ecoregion. This served as a verification of the DEM-derived wet areas by comparison with a wetland data set derived independently from aerial photographs. The areal correspondence of depth-to-water classes with different wetland types was similarly compared to ascertain how the relationship between depth-towater classification and discrete wetland units might vary with wetland type. The distribution of wetlands and depth-to-water values was compared by eco-region. The province, based on the provincial ecological site classification, is divided into seven distinct eco-regions (Figure 1). The areal extent and proportion of wetlands in the discrete wetland inventory and depth-to-water classes were calculated for each ecoregion. Hypsometric curves were derived from the DEM for each ecoregion to characterize topography. Zonal summaries of wetlands and depth-to-water classes were performed using raster-based statistics algorithms, at a grid-cell resolution of 10 m. For

849

that purpose, all vector wetland features (the discrete wetland inventory) were converted to raster data with 10 m resolution. In the case of cells that corresponded to more than one wetland class, the cell was assigned to the classification that made up most of the cell. The new raster data was ‘snapped’ to the depth-to-water raster coverage to ensure pixels were coincident with each other. The two coverages were then compared. RESULTS AND DISCUSSION Table 2 shows the areal extent of wetlands and depth-to-water classes for the two provincial maps. By visual inspection, the two layers were found to be generally consistent with one another. Wetlands identified in the discrete map were generally found to fall within depth-to-water classes of 0 to 1 m (Figure 2). Most of these (51%–67%, by ecoregion) were fully embedded in the 0–10 cm class. An illustrative example of this is given for the Comeau Point area of eastern New Brunswick (Figure 3). This serves as a verification of the continuous maps derived from topographic and hydrologic data. Embedding in the 0–10 cm class was strongest with the coastal wetlands (SL, DU, TF, BC, SA), and with emergent and shrub wetlands (EW, SB) (Figure 4). Embedding was weakest for bogs (BO). Ombrogenous wetlands such as bogs, particularly blanket bogs or domed bogs, can have a wetland surface and water table raised above the surrounding terrain due to organic matter (Sphagnum peat) build-up. Therefore, the location and areal extent of bogs may not be as tightly controlled by local topography and surface flow regime as in other wetlands. The raised surface of such bogs might also lead to an over-estimation of depth-to-water for these areas. As a result, the DEM may not be as effective at predicting such wetlands. The discrete wetland map likely captured most of the larger (. 1 ha) wetland areas across the province. The continuous map, however, identified a much greater area as wet soils (Table 2). Total area with depth-to-water of 0–10 cm was approximately three times that of wetlands identified in the discrete map. Field investigations have found these predicted wet areas to be reliable for a small (1,457 ha) study area in New Brunswick (Murphy et al. 2006), and it is likely that these additional areas represent small wetlands and wet riparian zones. These small wetlands (, 1 ha) are too small to be captured by the aerial photo interpretation procedures used to produce the discrete wetland map. However, such small wetlands and wet riparian areas are important in terms of watershed and

850

WETLANDS, Volume 27, No. 4, 2007

Figure 1.

Ecoregions of New Brunswick.

habitat protection and land management in forestry and agriculture. This additional information may help improve planning and management. This observation highlights a disadvantage of the discrete mapping approach; wetlands appear as isolated units rather than as a part of the continuous landscape hydrologic system that connects wetlands in reality. In contrast, the depth-to-water mapping approach, based on topographic and hydrologic data, produces a continuous coverage where wetlands are nested within the surface hydrologic regime. This adds considerably to the information available to land planners and managers as the pattern of hydrologic flow and wetland connectivity becomes apparent, which is important for watershed

Table 2. Ecoregions 1 2 3 4 5 6 7 Total (ha) Total (%)

management, and identifying surface pollutant transport pathways or wildlife migration routes. The continuous model also offers a systematic means to explore to what extent wetlands may be affected by road and highway construction, and by encroaching rural and urban developments. Changes to the local topography (infills, excavations) and drainage network (road ditches, culverts), as a result of these land disturbances, can be incorporated into the model by altering the DEM accordingly. The likely changes in surface flow regime and wetland distribution can then be derived from this altered topographic surface. Figure 5 presents the frequency distribution of all the discrete wetland features, classified by area. A

Areal extent (ha) of wetlands and depth-to-water classes for each of the wetland maps. Wetland inventory

Depth-to-water (cm) 0–10

10–25

25–50

50–100

100–200

. 200

9,911 35,026 12,631 20,215 33,629 50,240 339,102 24,437 93,684 50,910 70,582 106,791 139,597 412,953 27,419 104,684 49,062 72,871 113,259 160,884 698,508 22,068 43,682 13,416 17,491 25,607 34,432 91,223 93,989 300,861 132,974 182,313 268,581 338,791 792,455 183,630 442,642 280,688 340,233 420,290 361,222 245,710 256,16 104,800 33,268 42,628 60,055 67,218 69,973 387,070 1,125,380 572,949 746,334 1,028,213 1,152,383 2,649,924 5.3 15.5 7.9 10.3 14.1 15.8 36.4

Proportion as wetland (%) Total area 490,843 874,517 1,199,269 225,852 2,015,975 2,090,786 377,942 7,275,184 100.0

2.02 2.79 2.90 9.75 4.66 8.78 6.78

Murphy et al., COMPARISON OF WETLAND MAPPING APPROACHES

Figure 2. The composition of wetlands in the discrete wetland inventory by depth-to-water class for each ecoregion.

851

marked increase in number of wetlands as the wetland areas get smaller is apparent. Such a trend suggests that there are likely to be many additional wetlands that are even smaller and that the number of wetlands per area will continue to increase with decreasing wetland size. However, these wetlands are too small to be captured by the aerial photo interpretation process. These additional wetlands are likely to be the small wet-areas mapped by the continuous model, as discussed above. The aerial proportion of the 0–10 cm depth-towater class and the discrete wetlands for each ecoregion is shown in Figure 6. Strong variations by ecoregion are evident. Areas with a depth-towater of less than 10 cm are less prevalent in ecoregions 1–3, indicating better drained soils, and most prevalent in ecoregions 6 and 7, indicating a greater proportion of poorly drained soils. The

Figure 3. Detail of Comeau Point, New Brunswick, showing that the discrete wetland units (various blue-shaded features) typically fall into the inner portion of the wet-area zonation (pink to red indicates depth-to-water from 0–1 m, respectively). Background: mosaic of air photos. Wetland type: Salt Marsh (SA), Shrub Wetland (SW), Emergent Wetland (EW), Aquatic Bed (AB), Tidal Flat (TF), Beach (BC), Rocky Shore (RK).

852

Figure 4. The composition of each wetland type from the discrete inventory by depth-to-water class. Wetland type: Aquatic Bed (AB), Beach (BC), Bog (BO), Dune (DU), Emergent Wetland (EW), Forested Wetland (FW), Non-productive Forest (NP), Rocky Shore (RK), Salt Marsh (SA), Shrub Wetland (SB, SW), Salt Lake (SL), Tidal Flat (TF), Vegetated Dune (VD), Wetland (WL).

proportion of wetlands in the discrete inventory reflects the same pattern. These trends are likely a reflection of the differing topography of the ecoregions, as characterized by the hypsometric curves in Figure 6. The prevalence of the 0–10 cm depth-to-water class increases as the hypsometric curves become flatter. Ecoregions 1–3 are upland and highland ecoregions with rolling to mountainous topography (200 to . 800 m elevation) developed on folded sedimentary, igneous, and metamorphic Palaeozoic rocks dissected by river valleys. Ecoregions 6 and 7 are lowland ecoregions with flat to undulating topography developed on flat to gently dipping Carboniferous sandstones, shales, and conglomerates, rising from 0–200 m elevation. This latter terrain leads to poorer drainage conditions in the landscape and more prevalent wetland development than in the upland ecoregions. Currently, the continuous mapping process is in part limited by the coarse-gridded nature of the provincial digital elevation grid (average grid spacing is about 70 m). Field studies have shown

Figure 5. Frequency distribution (logarithmic scale) of wetland size for the discrete wetland inventory.

WETLANDS, Volume 27, No. 4, 2007

Figure 6. The areal proportion of each ecoregion covered by the 0–10 cm depth-to-water class and photointerpreted wetlands. Hypsometric curves for each ecoregion characterize topography.

that the reliability of location of DEM-derived flow channels is better than 50 m, on average (Murphy et al. 2006). However, further improvements are achievable. DEM resolution can be enhanced. Ideally, this might be done through acquisition of LiDAR (Light Detecting and Ranging) data (Gomes Pereira and Wicherson 1999, Wehr and Lohr 1999, Iverson et al. 2004). Paine et al. (2004) explored the use of a LiDAR DEM to improve more traditional photo-interpreted wetland maps in Texas. However, LiDAR acquisition is still costly. Automation of the process for hydrographic features identification and classification, based on supervised and un-supervised analysis of surface images (orthorectified aerial photos, Landsat, RADARSAT), using image classification software might also improve the hydrographic data used to derive the depth-to-water maps. These approaches can also be used to add to the wetland mapping effort, including the use of multi-temporal images (Lunetta and Balogh 1999, Parmuchi et al. 2002, Toeyrae et al. 2002). Interestingly, a number of studies have combined image interpretation and topographic data and this approach shows good potential. Baker et al. (2006) applied multitemporal Landsat TM imagery, combined with topographic and soils data to map wetlands and riparian systems in Montana. Li and Chen (2005) combined slope data from a DEM with Landsat TM and multitemporal Radarsat data to improve wetland classification for selected sites across eastern Canada. Ensuring data quality is important to ensuring reliable output from the continuous mapping pro-

Murphy et al., COMPARISON OF WETLAND MAPPING APPROACHES cess. Data quality may be improved by careful inspection and correction of the DEM to ensure consistency with the hydrographic data, and of the hydrographic features, to ensure they are accurate and continuous across the landscape. In conclusion, excellent agreement was found between the discrete photo-interpreted wetland maps and the continuous depth-to-water map derived from a DEM and hydrographic data. This served as a verification of the GIS process used to produce the latter. The continuous map identified much greater areas of wet soils. These likely include many small wetlands (, 1 ha) and riparian zones that are too small to be identified through photo interpretation. The continuous map also has the advantage of locating wetlands within the continuum of water flow in the landscape, adding to the information available to planners. Topography was found to be an important factor in determining the prevalence of wetlands and their distribution in each ecoregion of New Brunswick. ACKNOWLEDGMENTS The authors wish to acknowledge the assistance of the New Brunswick Department of Natural Resources in providing data and other help for this work. This work was financially supported by the Nexfor-Bowater Forest Watershed Research Centre, with supplementary support from Wildlife Habitat Canada (c/o Fish and Wildlife Branch, New Brunswick Department of Natural Resources), the Sustainable Forest Management Network and an NSERC Discovery grant to P.A.A. LITERATURE CITED Baker, C., R. Lawrence, C. Montagne, and D. Patten. 2006. Mapping wetlands and riparian areas using landsat ETM+ imagery and decision-tree-based models. Wetlands 26:465–74. Bhatti, J. S. and C. M. Preston. 2006. Carbon dynamics in forest and peatland ecosystems: preface. Canadian Journal of Soil Science 86:155–58. Boerner, R. E. J., S. J. Morris, E. K. Sutherland, and T. F. Hutchinson. 2000. Spatial variability in soil nitrogen dynamics after prescribed burning in Ohio mixed-oak forests. Landscape Ecology 15:425–39. Christensen, N. L., A. M. Bartuska, J. H. Brown, S. Carpenter, C. D’Antonio, R. Francis, J. F. Franklin, J. A. MacMahon, R. F. Noss, D. J. Parsons, C. H. Peterson, M. G. Turner, and R. G. Woodmansee. 1996. The report of the Ecological Society of America committee on the scientific basis for ecosystem management. Ecological Applications 6:665–91. Detenbeck, N. E., S. M. Galatowitch, J. Atkinson, and H. Ball. 1999. Evaluating perturbations and developing restoration strategies for inland wetlands in the Great Lakes Basin. Wetlands 19:789–820. DNR. 2006. New Brunswick wetland classification for 2003–2012 photo cycle. Department of Natural Resources New Brunswick, Fredericton, NB, Canada.

853

Findlay, C. S. and J. Bourdages. 2000. Response time of wetland biodiversity to road construction on adjacent lands. Conservation Biology 14:86–94. Gomes Pereira, L. M. and R. J. Wicherson. 1999. Suitability of laser data for deriving geographical information: a case study in the context of management of fluvial zones. ISPRS Journal of Photogrammetry and Remote Sensing 54:105–14. Grunwald, S. 2006. What do we really know about the space-time continuum of soil-landscapes? p. 3–36. In S. Grunwald (ed.) Environmental Soil-landscape Modeling: Geographic Information Technologies and Pedometrics. CRC Press, New York, NY, USA. Gu ¨ ntner, A., J. Seibert, and S. Uhlenbrook. 2004. Modeling spatial patterns of saturated areas: an evaluation of different terrain indices. Water Resources Research 40:1–19. Haag, D. and M. Kaupenjohann. 2001. Landscape fate of nitrate fluxes and emissions in Central Europe: a critical review of concepts, data, and models for transport and retention. Agriculture, Ecosystems and Environment 86:1–21. Hornberger, G. M. and E. W. Boyer. 1995. Recent advances in watershed modelling. Review of Geophysics 33:949–57. Iverson, L. R., M. E. Dale, C. T. Scott, and A. Prasad. 1997. A GIS-derived integrated moisture index to predict forest composition and productivity of Ohio forests (U.S.A.). Landscape Hydrology 12:331–48. Iverson, L. R., A. M. Prasad, and J. Rebbeck. 2004. A comparison of the integrated moisture index and the topographic wetness index as related to two years of soil moisture monitoring in Zaleski State Forest, Ohio. 14th Central Hardwoods Forest Conference. Wooster, OH, USA. Johnston, C. A. and P. Meysembourg. 2002. Comparison of the Wisconsin and National Wetlands Inventories. Wetlands 22:386–405. Kent, B. J. and J. N. Mast. 2005. Wetland change analysis of San Dieguito Lagoon, California, USA: 1928–1994. Wetlands 25:780–87. Li, J. and W. Chen. 2005. A rule-based method for mapping Canada’s wetlands using optical, radar and DEM data. International Journal of Remote Sensing 26:5051–69. Lunetta, R. S. and M. E. Balogh. 1999. Application of multitemporal landsat 5 TM imagery for wetland identification. Photogrammetric Engineering and Remote Sensing 65:1303– 10. Mitsch, W. J. and J. G. Gosselink. 2000. Wetlands, third edition. John Wiley & Sons, Inc., New York, NY, USA. Moore, I. D., R. B. Grayson, and A. R. Ladson. 1991. Digital terrain modelling: a review of hydrological, geomorphological, and biological applications. Hydrological Processes 5:3–30. Murphy, P. N. C., J. Ogilvie, M. Castonguay, T. Connors, F.-R. Meng, and P. A. Arp. 2006. DEM-derived flow channel and wet area mapping: a new tool for forest operations planning. Sustainable Forest Management Network, Fourth International Conference. Edmonton, AB, Canada. NWWG. 1997. The Canadian Wetland Classification System. National Wetlands Working Group, Wetlands Research Centre, University of Waterloo, Waterloo, ON, Canada. Olivera, F. 1996. Spatial hydrology of the Urubamba river system in Peru using geographic information systems (GIS). Center for Research in Water Resources, The University of Texas at Austin, TX, USA. (www.ce.utexas.edu/prof/maidment/CE397/ urubamba/peru.htm). Paine, J. G., W. A. White, R. C. Smyth, J. R. Andrews, and J. C. Gibeaut. 2004. Mapping coastal environments with lidar and EM on Mustang Island, Texas, U.S. Leading Edge 23:894–98. Parmuchi, M. G., H. Karszenbaum, and P. Kandus. 2002. Mapping wetlands using multi-temporal RADARSAT-1 data and a decision-based classifier. Canadian Journal of Remote Sensing 28:175–86. Pegler, K. H., D. J. Coleman, and H. T. T. Nguyen. 2000. Comparing TIN random densification with the mean profile filter to minimize the ridging phenomenon in Service New Brunswick digital terrain models. Geomatica 54:433–40.

854 Ryan, P. J., N. J. McKenzie, D. O’Connell, A. N. Loughhead, P. M. Leppert, D. Jacquier, and L. Ashton. 2000. Integrating forest soils information across scales: spatial prediction of soil properties under Australian forests. Forest Ecology and Management 138:139–57. Saunders, W. K. and D. R. Maidment. 1996. A GIS assessment of nonpoint source pollution in the San Antonio-Nueces coastal basin. Center for Research in Water Resources, The University of Texas at Austin, TX, USA. (http://www.ce.utexas.edu/ centers/crwr/reports/online.html). Simley, J. 2004. The geodatabase conversion. USGS National Hydrography Newsletter 3(4). USGS. Toeyrae, J., A. Pietroniro, L. W. Martz, and T. D. Prowse. 2002. A multi-sensor approach to wetland flood monitoring. Hydrological Processes 16:1569–81. Tomer, M. D., D. E. James, and T. M. Isenhart. 2003. Optimizing the placement of riparian practices in a watershed using terrain analysis. Journal of Soil and Water Conservation 58:198–206.

WETLANDS, Volume 27, No. 4, 2007 Turner, R. K., J. C. J. M. van den Bergh, T. Soderqvist, A. Barendregt, J. van der Straaten, E. Maltby, and E. C. van Ierland. 2000. Ecological-economic analysis of wetlands: scientific integration for management and policy. Ecological Economics 35:7–23. U.S.F.W.S. 2007. National Wetlands Inventory Metadata. United States Fish and Wildlife Service. (http://www.fws.gov/nwi/downloads/metadata/nwi_meta.txt). Watson, D. F. and G. M. Philip. 1985. A refinement of inverse distance weighted interpolation. Geoprocessing 2:315–27. Wehr, A. and U. Lohr. 1999. Airborne laser scanning- an introduction and overview. ISPRS Journal of Photogrammetry and Remote Sensing 54:68–82.

Manuscript received 13 December 2006; accepted 30 May 2007.