Lawrence W. Martz

Numerical definition of drainage network and subcatchment areas from digital elevation models

Abstract A set of ten algorithms to automate the determination of drainage network and subcatchment areas from Digital Elevation Models (DEMs) is presented. The algorithms perform such tasks as: DEM aggregation; depression identification and treatment; relief incrementation of flat areas; flow vector determination; watershed boundary delineation; drainage network and subcatchment area definition and systematic indexing; tabulation of channel and subcatchment area properties; and evaluation of drainage network composition. A computer program (written in FORTRAN 77) that integrates these algorithms for a full DEM evaluation also is discussed. The primary purpose of the algorithms and computer program is to parameterize rapidly drainage network and subcatchment properties from widely available DEMs for subsequent use in hydrologic surface runoff models, watershed discretizations, or statistical and topological evaluation of drainage networks. Selected results of a DEM evaluation are presented for illustration purposes.

The treatment of flat areas and depressions in automated drainage analysis of raster digital elevation models

Methods developed to process raster digital elevation models (DEM) automatically in order to delineate and measure the properties of drainage networks and drainage basins are being recognized as potentially valuable tools for the topographic parameterization of hydrological models. All of these methods ultimately rely on some form of overland flow simulation to define drainage courses and catchment areas and, therefore, have difficulty dealing with closed depressions and flat areas on digital land surface models. Some fundamental assumptions about the nature of these problem topographic features in DEM are implicit in the various techniques developed to deal with them in automated drainage analysis. The principal assumptions are: (1) that closed depressions and flat areas are spurious features that arise from data errors and limitations of DEM resolution; (2) that flow directions across flat areas are determined solely by adjacent cells of lower elevation; and (3) that closed depressions are caused exclusively by the underestimation of DEM elevations. It is argued that while the first of these assumptions is reasonable, given the quality of DEMs generally available for hydrological analysis, the others are not. Rather it seems more likely that depressions are caused by both under- and overestimation errors and that flow directions across flat areas are determined by the distribution of both higher and lower elevations surrounding flat areas. Two new algorithms are introduced that are based on more reasonable assumptions about the nature of flat areas and depressions, and produce more realistic results in application. These algorithms allow breaching of depression outlets and consider the distribution of both higher and lower elevations in assigning flow directions on flat areas. The results of applying these algorithms to some real and hypothetical landscapes are presented.

CATCH: a FORTRAN program for measuring catchment area from digital elevation models

Abstract This paper presents a computer program which analyses an elevation matrix to determine catchment area at every point represented by an element in the matrix. Catchment area is calculated both under the assumption that water flow over the surface will terminate in depressions and under the assumption that all depressions will overflow at the lowest available outlet. These are termed local and global catchment area respectively. A matrix of values is generated, with the same structure and dimensions as the elevation matrix, for each catchment area variable. A modified elevation matrix, produced as a by-product of the analysis, illustrates the topographic change caused by filling all depressions to the elevation of their outlets. Subtracting the original elevation matrix from the modified elevation matrix gives the maximum depth of water ponding on the land surface.

An outlet breaching algorithm for the treatment of closed depressions in a raster DEM

Abstract Automated drainage analysis of raster DEMs typically begins with the simulated filling of all closed depressions and the imposition of a drainage pattern on the resulting flat areas. The elimination of closed depressions by filling implicitly assumes that all depressions are caused by elevation underestimation. This assumption is difficult to support, as depressions can be produced by overestimation as well as by underestimation of DEM values.This paper presents a new algorithm that is applied in conjunction with conventional depression filling to provide a more realistic treatment of those depressions that are likely due to overestimation errors. The algorithm lowers the elevation of selected cells on the edge of closed depressions to simulate breaching of the depression outlets. Application of this breaching algorithm prior to depression filling can substantially reduce the number and size of depressions that need to be filled, especially in low relief terrain.Removing or reducing the size of a depression by breaching implicitly assumes that the depression is due to a spurious flow blockage caused by elevation overestimation. Removing a depression by filling, on the other hand, implicitly assumes that the depression is a direct artifact of elevation underestimation. Although the breaching algorithm cannot distinguish between overestimation and underestimation errors in a DEM, a constraining parameter for breaching length can be used to restrict breaching to closed depressions caused by narrow blockages along well-defined drainage courses. These are considered the depressions most likely to have arisen from overestimation errors. Applying the constrained breaching algorithm prior to a conventional depression-filling algorithm allows both positive and negative elevation adjustments to be used to remove depressions.The breaching algorithm was incorporated into the DEM pre-processing operations of the TOPAZ software system. The effect of the algorithm is illustrated by the application of TOPAZ to a DEM of a low-relief landscape. The use of the breaching algorithm during DEM pre-processing substantially reduced the number of cells that needed to be subsequently raised in elevation to remove depressions. The number and kind of depression cells that were eliminated by the breaching algorithm suggested that the algorithm effectively targeted those topographic situations for which it was intended. A detailed inspection of a portion of the DEM that was processed using breaching algorithm in conjunction with depression-filling also suggested the effects of the algorithm were as intended.The breaching algorithm provides an empirically satisfactory and robust approach to treating closed depressions in a raster DEM. It recognises that depressions in certain topographic settings are as likely to be due to elevation overestimation as to elevation underestimation errors. The algorithm allows a more realistic treatment of depressions in these situations than conventional methods that rely solely on depression-filling.

Multisensor Hydrologic Assessment of a Freshwater Wetland

This article evaluates the use of synthetic aperture radar (SAR) and visible/infrared (VIR) satellite imagery for mapping the extent of standing water in the Peace-Athabasca Delta during spring and summer of 1998. SAR images contain data about the geometric and electrical characteristics of the objects, while VIR images contain information about the reflectivity of objects. Radar pulses can also penetrate vegetation to some degree depending on the wavelength and vegetation thickness. It is hypothesized that since Radarsat and SPOT images contain complementary information, flood mapping will be more efficient when the two image types are used in combination. Radarsat SAR and SPOT multispectral imagery from May 1998 and July 1998 were used for the flood mapping. A Radarsat S2 image (27.5° incidence angle) was obtained for May 1998. To evaluate the incidence angle effect on flood mapping, a Radarsat S1 image (27.5° incidence angle) and a Radarsat S7 image (47.0° incidence angle) were obtained for July 1998. The Radarsat scenes were calibrated and filtered, and all imagery were orthorectified to minimize geometric distortion. A Mahalanobis distance algorithm was used to classify the SPOT scenes, the Radarsat scenes, and a combination of the two into open water, flooded vegetation, and nonflooded land. The results indicate that flood mapping in both spring and summer conditions has significantly higher accuracy when Radarsat and SPOT imagery are used in combination, rather than separately. However, it is important to use Radarsat imagery acquired at low incidence angles. Classification of the SPOT scene combined with the Radarsat S1 scene achieved significantly better results than those obtained when the SPOT scene was classified in combination with the Radarsat S7 scene.

Using cesium-137 to assess the variability of net soil erosion and its association with topography in a Canadian Prairie landscape

Summary Cesium-137 in the soil was used to estimate net erosion in a small (178 ha), cultivated drainage basin in central Saskatchewan. Net erosion over the past 20 to 25 years was estimated at 174 sites and ranged from a loss of 121 kg m −2 to a gain of 1197 kg m −2 . Soil loss was most severe on slope crests and in swales while soil gain was greatest in upland depressions and along the main outlet channel. In a representative sample of 71 sites spaced regularly over the basin, 58% of the sites showed soil loss and 17% showed soil gain. Computer programs were developed to calculate topographic variables from a digital terrain model of the basin. The relationship between net erosion and topographic variables was examined by grouping sites into landform categories. Both the group means and standard deviations of net erosion, and the nature of within-group correlations of topographic variables with net erosion, differed between landform categories.

Using digital terrain analysis modeling techniques for the parameterization of a hydrologic model

Abstract This paper discusses the application of digital terrain analysis modeling techniques to the parameterization of a semi-distributed hydrologic model. Most current techniques for deriving physiographic parameters in watershed analyses, including those using commercial geographic information systems (GIS), are tedious, costly and time consuming. The demands of these techniques result in them usually being limited in practical application to deriving parameters at only one level of detail or for only one set of sub-basins. This paper presents a computerized interface (SLURPAZ) that was developed to combine the output of an established digital terrain analysis model (TOPAZ) with digital land cover data to derive all the necessary physiographic parameters required as input by a widely used semi-distributed hydrological model (SLURP). This interface makes it possible to derive physiographic parameters rapidly and accurately, at several different levels of detail and for varying numbers of sub-basins. This paper describes the methods by which the SLURPAZ interface integrates land cover data with the topographic parameters derived by TOPAZ from a digital elevation model (DEM). It also presents an example application of the interface to an intermediate-sized alpine basin in Yukon, Canada. Hydrological model outputs obtained using the computerized interface are compared with those obtained using manual techniques.

Using cesium-137 and landform classification to develop a net soil erosion budget for a small Canadian Prairie watershed

Summary The primary aim of this study was to determine if the relationship between objectively defined landform classes and soil erosion identified by Martz & De Jong (1987) for a small, agricultural Prairie basin, could be exploited in the calculation of soil erosion budgets and sediment yields. A secondary aim was to assess the degree to which the relationship described in that earlier study might apply to other small agricultural basins in the region. Following the earlier study, computer-assisted analysis of a digital elevation model was used to subdivide a 65 ha cultivated basin near Saskatoon, Saskatchewan into landform classes, and137Cs measurements at a sample of sites over the basin were used to determine the mean net erosion of each class. Mean net erosion differed significantly between the classes, but the relationship of net erosion to landform class in the study basin differed from that reported for the basin where the classification was originally applied. This was thought to be due largely to differences in soil properties between the basins. The mean net erosion and area of the landform classes were used to calculate a net erosion budget for the basin. The total soil loss from the basin given by the erosion budget agreed closely with the sediment accumulation, also measured using137Cs, behind a dam at the basin mouth, and was consistent with the few sediment yield data available for other small Saskatchewan basins. It was concluded that the landform classes provided a reliable basis for the empirical calculation of the net erosion budget and sediment yield.

An investigation of the spatial association between snow depth and topography in a Prairie agricultural landscape using digital terrain analysis

Abstract The association of snowcover depth with topography and land cover was examined for a 1.5 km 2 agricultural site in the Canadian Prairies. Topography was quantified by a suite of general geomorphometric variables that were measured directly by the analysis of a raster digital elevation model (DEM) with a 10 m grid cell size. These variables expressed local surface morphology and relative topographic position. The major land covers represented in the study area were wheat stubble and summerfallow. Point observations of snow depth at the site were obtained near the time of peak snow accumulation and during melt. Initial analysis was made of scatterplots and correlations between snow depth and individual topographic variables. These correlations were weak but illustrated that, at the local scale, snow depth patterns cannot be adequately modelled through simple bivariate relationships with topographic variables. The correlations also showed that snow depth is more strongly related to relative topographic position variables than to variables expressing local surface morphology. Subsequent analysis involved a divisive subdivision of the study site into six terrain units defined in terms of several relative topographic position and land cover variables. The units were used to model the spatial patterns of snow depth that occurred during the observed snowcover situations. Although the terrain classification only reduced the total snow depth variance by approximately 30%, it delineated the major patterns of snow depth observed over the study area and elucidated the spatial relationships between snowcover depth and landscape attributes.

Energy and water cycles in a high-latitude, north-flowing river system: Summary of results from the Mackenzie GEWEX Study-phase I

Abstract The MacKenzie Global Energy and Water Cycle Experiment (GEWEX) Study, Phase 1, seeks to improve understanding of energy and water cycling in the Mackenzie River basin (MRB) and to initiate and test atmospheric, hydrologic, and coupled models that will project the sensitivity of these cycles to climate change and to human activities. Major findings from the study are outlined in this paper. Absorbed solar radiation is a primary driving force of energy and water, and shows dramatic temporal and spatial variability. Cloud amounts feature large diurnal, seasonal, and interannual fluctuations. Seasonality in moisture inputs and outputs is pronounced. Winter in the northern MRB features deep thermal inversions. Snow hydrological processes are very significant in this high-latitude environment and are being successfully modeled for various landscapes. Runoff processes are distinctive in the major terrain units, which is important to overall water cycling. Lakes and wetlands compose much of MRB and are p...