An Improved Method for Watershed Delineation and ...€¦ · performing puddle delineation, executing the computation of depression storages, determining flow directions and accumusummary

1

An Improved Method for Watershed Delineation and Computation of Surface Depression Storage

Xuefeng Chu, A.M.ASCE1,*, Jianli Zhang1, Yaping Chi1, Jun Yang1

1Department of Civil Engineering (Dept 2470), North Dakota State University,

PO Box 6050, Fargo, ND 58108-6050, United States *PH (701) 231-9758; FAX (701) 231-6185; Email: [email protected]

Abstract

Surface microtopography plays an important role in overland flow generation and soil erosion processes. Characterization of surface depressions and delineation of the entire watershed are critical to watershed modeling and management. In most hydrologic models, however, surface depression storage is estimated indirectly and inputted as a known value. In addition, it is often assumed that overland flow initiates after all surface depressions are fully filled. In reality, surface microtopography may control overland flow generation, surface runoff, soil erosion, and other hydrologic processes in a dynamic manner. The role of depressions that have various spatial scales and distribution characteristics is far beyond the functions of storing and detaining/retaining surface runoff. In this study, an improved surface delineation method was proposed to identify surface depressions and their relationships, precisely quantify the surface microtopography, and compute the maximum depression storage based on high-resolution DEM data obtained by using an instantaneous-profile laser scanner. Furthermore, a user-friendly, Windows-based software package was developed to facilitate the associated computations and visualization. The delineation method and the related software were tested using various scale DEM data. It is demonstrated that the new delineation approach is effective and efficient.

Introduction

Soil surface microtopography plays an important role in overland flow generation and soil erosion processes (Huang and Bradford 1990; Kamphorst et al. 2000; Hansen 2000; Kamphorst and Duval 2001; Darboux and Huang 2003, 2005; Abedini et al. 2006). Surface depressions affect infiltration, surface runoff, and other hydrologic processes in a dynamic manner. In the current watershed modeling practice, however, surface depression storage is often estimated indirectly and treated as a constant. It is also assumed that overland flow initiates after all surface depressions are fully filled. In reality, the role of depressions of various spatial scales can be far beyond the functions of storing and detaining/retaining surface runoff (Chu 2010). Characterization of surface topography and quantification of depression storage are critical to watershed hydrologic and water quality modeling, and the associated environmental assessment. The maximum depression storage (MDS) can be estimated by surface roughness indices (e.g., Onstad 1984; Mwendera and Feyen 1992; Hansen et al. 1999; Kamphorst et al. 2000). Methods also have been developed to determine

1113Watershed Management 2010 © ASCE 2011

2

surface depression storage by directly using DEMs (Ullah and Dickinson 1979; Planchon and Darboux 2002). In addition, a variety of methods and techniques have been developed for digital terrain analysis, extraction of drainage networks, and delineation of watershed boundaries from DEMs (e.g., Marks et al. 1984; O’Callaghan and Mark 1984; Jenson and Domingue 1988; Jenson 1991; Martz and Garbrecht 1993; Garbrecht and Martz 2000).

In the current study, an automated algorithm is developed to characterize surface microtopography and precisely compute surface depression storages based on high-resolution DEM data obtained by an instantaneous-profile laser scanner (Huang and Bradford 1992; Darboux and Huang 2003). To facilitate the relevant computation and visualization, a user-friendly, Windows-based software package is developed. Testing of the algorithm and the associated program demonstrates its efficiency and effectiveness. The major unique features of this new surface delineation algorithm include: (1) it takes into account the real puddle filling-merging-spilling process and eliminates the artificial depression-filling pre-processing of original DEMs, which is implemented in existing watershed delineation tools; and (2) a series of schemes have been developed to deal with special topographic features related to complex land surfaces, such as flats, flow over flat surfaces, and combination of multiple puddles.

Methodology Land surfaces are characterized with depressions, peaks, ridges, and channels. A depression/puddle has at least one center and one threshold. If the elevation of a cell is lower than those of its eight adjacent cells, the cell is identified as the center of a puddle. The threshold of a puddle has the highest elevation for water ponding, beyond which the ponded water spills. If two puddles (1st level) share a common threshold, they will be merged and form a 2nd level puddle. Similarly, further combination of such a puddle with others forms higher level puddles. The algorithm developed in this study is used for characterizing soil surface microtopography based on DEM data. It involves identifying puddles, including their centers, thresholds, and all other cells included in the puddles, computing their storages [up to the threshold(s) of a puddle] and the MDS of the entire surface, computing the maximum ponding area (MPA), and determining the levels and the hierarchical relationships of the puddles. Theoretically, the developed method for puddle identification and delineation is scale independent. Different DEMs (both lab and watershed scales) have been used to test this puddle delineation method. For small scale surfaces, the instantaneous-profile laser scanner is utilized to quantify the surface microtopography with a horizontal resolution of 0.98 mm and a vertical resolution of 0.5 mm. Identification of Puddles: Based on the DEM data, the center of each puddle is first determined (Fig. 1a). Following a set of criteria, a searching process is performed to identify all cells to be included in the puddle. Eight neighboring cells of the puddle center are first added to a searching list (Fig. 1a). Then, the cell with the lowest elevation in the searching list is located (Fig. 1b) and added to the puddle (Fig. 1c). The searching list is updated by adding more neighboring cells of the puddle (Fig. 1c).


3

Such an expansion process continues (Figs. 1d and 1e) until a threshold (overflow point) is found (Fig. 1f). The puddle, including its center, threshold, and all other cells is identified (Fig. 1g). In such a way, all first-level puddles are determined.

A puddle may have multiple centers (a flat) and multiple thresholds. A flat may form an individual puddle of zero depression storage. Special rules and criteria are also introduced to account for these special conditions in the searching process. In addition, a puddle may have various relationships with its adjacent puddles. For example, two puddles with a common threshold will be combined during the filling process (Fig. 2). Thus, a new searching process starts to identify all cells included in this combined puddle at the second level. Such a procedure is repeated until all higher-level puddles are determined. Chu et al. (2010) detailed the procedure for puddle delineation and the methods for dealing with special topographic conditions, such as flats, multiple thresholds, and combination of multiple puddles.

Fig. 2. Identification of higher level puddles

■ Center ▲ Threshold ■■ Two 1st level puddles ■ 2nd level puddle

(f) (e)(d)(c) (b)

Fig. 1. Puddle identification and the searching process

(a)

Center Cells in the searching listLowest cell in the searching listCells included in a puddle

Threshold

(g)


4

Computation of Depression Storages: After completion of the identification of puddles, the storages of puddles are computed for all levels. The MDS for any level can be expressed as:

( )∑∑= =

ΔΔ−=l in

i

m

jjiitl yxzzMDS

1 1,, (1)

where MDSl = maximum depression storage of the entire area at level l; nl = number of puddles at level l; mi = number of cells in puddle i; zt,i = elevation of the threshold of puddle i; zi,j = elevation of cell j in puddle i; Δx = size of a cell along x direction; and Δy = size of a cell along y direction. Windows Interface A user-friendly Windows interface (Fig. 3) has been developed with C# computer programming and Windows GDI+ (Graphics Device Interface). The Windows-based software facilitates data loading, puddle delineation, computation of flow directions and accumulations, computation of depression storages, and visualization of the puddle delineation process and the computation results. The main interface includes a map view for visualizing the results and a control panel for importing DEM data, performing puddle delineation, executing the computation of depression storages, determining flow directions and accumulations, displaying summary results, and accessing more detailed results.

The procedure for puddle identification and computation of the MDS can be summarized as the following three steps:

Fig. 3. Main interface of the Windows-based software


5

Step 1: Selection of the format and unit of the DEM data and loading of the DEM. Step 2: Delineation of puddles: After importing the DEM data, the puddle searching

process can be initiated by clicking the button “Puddle Delineation.” The map area can display the searching process that follows a hierarchical order (from the lowest to highest levels). Colors represent different levels. The status bar at the bottom of the interface shows the current level, current puddle, and the number of cells within the current puddle (Fig. 3).

Step 3: Visualization of results: After completion of the puddle searching and

identification process, the control panel displays a summary of puddle delineation, including puddle levels, and MDS and MPA for all levels. In addition, the detailed results can be visualized by a new map view or displayed in a series of summary tables (Fig. 4). The new map view supports basic visualization functions, such as zooming and moving. Users can select and view the distribution of puddles at different levels. The summary tables include basic puddle delineation information and all details on puddles, their levels, centers, thresholds, storages, water-covered areas, as well as cells of any puddles and their coordinates (x, y, z).

Testing and Analysis Tests of the puddle delineation algorithm and the associated software have been conducted for various scales. Small-scale soil surfaces were created and scanned by

Fig. 4. Visualization of the detailed delineation results


6

using the instantaneous-profile laser scanner, from which high-resolution DEM data were obtained. Then, surface microtopography was characterized, and MDS and MPA were computed. As one of the testing examples, a small mold with an impervious rough surface was created and scanned by the laser scanner (Fig. 5). The surface includes twelve first-level puddles of varying sizes.

The DEM was used for delineation of puddles and computation of their MDS values. In addition, colored water was used to fill depressions and the amount of water for fully filling each depression was recorded. Fig. 6a shows the distribution of puddles under the fully-filled condition (highest level). Note that some puddles have been combined. The developed software was applied to identify depressions, simulate the puddle filling process, and compute the storage volume of each identified puddle and the MDS of the entire area (Fig. 6b). Then, the simulated depression storages were compared against the water volumes measured for the puddles and the differences were evaluated (Table 1). It can be observed that the depression storages computed by the program are very close to the measured ones. The overall relative error for the ten puddles (Fig. 6b) is 2.01%. For the six larger puddles (Puddles 1 - 6) with a storage volume greater than 5.0 mL, the relative error is only 0.90%. Thus, the algorithm is able to provide accurate estimation of depression storages for both individual puddles and the entire area.

(b) (a)

Fig. 6. Spatial distributions of the measured (a) and simulated (b) puddles

300 350 400 450 500 550 600 650

250

300

350

400

450

500

550

0

1

2

3

4

5

6

7

89

10

11

1 2

3

5

4

8

7

10 9

6

Fig. 5. 3-D puddle distribution of the small mold surface


7

Table 1. Comparison of the Measured and Simulated Depression Storages (mL)

Puddle Measured Simulated Difference Relative Error (%) 1 59.00 60.87 1.87 3.17 2 26.00 24.21 1.79 6.88 3 22.00 21.39 0.61 2.77 4 19.00 18.80 0.20 1.05 5 12.00 11.35 0.65 5.42 6 7.00 7.07 0.07 1.00 7 4.00 3.52 0.48 12.00 8 3.00 2.34 0.66 22.00 9 1.00 0.73 0.27 27.00

10 1.00 0.62 0.38 38.00 Sum 154.00 150.90 3.10 2.01

Further testing of the puddle delineation program was conducted for a lab-scale rough soil surface (Fig. 7a) and a real watershed-scale land surface (Fig. 8a). A soil box was used to create the lab-scale soil surface. The surface was scanned by using the laser scanner. The acquired DEM was imported into the program and delineation was performed. Fig. 7b shows the distribution of the delineated puddles, which match the real ones (Fig. 7a). The computed MDS and MPA of the soil surface are 1,425 cm3 and 1,564 cm2, respectively. The watershed-scale 30-m DEM (Fig. 8a) was obtained from the USGS. Similarly, puddle delineation was performed. Nine puddle levels and 231 1st level puddles were identified. The computed MDS and MPA of the land surface are 25.125 ×106 m3 and 4.514 ×106 m2, respectively. Fig. 8b indicates that the new delineation program is able to identify all scale puddles and handle special topographical conditions, such as flats (the areas in yellow). Note that some small depressions/puddles in Fig. 8b may result from the DEM data processing. Unlike real depressions, such artefacts should be removed before the delineation results can be used for further watershed hydrologic or environmental modeling.

Fig. 7. Lab-scale soil surface (a) and the delineated puddles (b)

(a) (b)


8

Conclusions An algorithm was presented in this paper for delineation of surface depressions/puddles and computation of maximum depression storage and maximum ponding area. A user-friendly, Windows-based software package also was developed to facilitate the computation and visualize the puddle delineation process and results. Tests of the algorithm and the associated software for various surfaces demonstrated that the program can be used as an efficient tool for characterizing surface microtopography, handling special surface conditions (e.g., flats), and precisely computing the depression storage of any individual puddle as well as the maximum depression storage of the entire surface based on raster DEM data.

Acknowledgments

This material is based upon the work supported by the National Science Foundation under Grant No. EAR-0907588 and ND NASA EPSCoR through NASA grant #NNXO7AK91A.

References Abedini, M. J., Dickinson, W. T., and Rudra, R. P. (2006). “On depressional storages:

The effect of DEM spatial resolution.” J. Hydrol., 318, 138-150.

Chu, X. (2010). “Characterization of microtopography and its hydrologic significance” (Chapter 1), p1-20. In: Modeling Hydrologic Effects of Microtopographic Features, edited by X. Wang. Nova Science Publishers, Inc.

Chu, X., Chi, Y., and Yang, J. (2010). Development of a new algorithm for puddle delineation of rough surfaces. J. Hydrol. (in submission).

Fig. 8. Watershed-scale land surface (a) and the delineated puddles (b)

(a) (b)

Puddle threshold Puddle center Puddle


9

Darboux, F., and Huang, C. (2003). “An instantaneous-profile laser scanner to measure soil surface microtopography” Soil Sci. Soc. Am. J., 67, 92-99.

Darboux, F., and Huang, C. (2005). “Does soil roughness increase or decrease water and particle transfer?” Soil Sci. Soc. Am. J., 69, 748-756.

Garbrecht, J., and Martz, L. W. (2000). TOPAZ: An Automated Digital Landscape Analysis Tool for Topographic Evaluation, Drainage Identification, Watershed Segmentation and Subcatchment Parameterization: TOPAZ User Manual. Grazinglands Research Laboratory, USDA Agricultural Research Services, El Reno, OK. ARS Pub No. GRL 2-00, 144pp.

Hansen, B. (2000). “Estimation of surface runoff and water-covered area during filling of surface microrelief depressions.” Hydrol. Processes, 14, 1235-1243.

Hansen, B., Schjonning, P., and Sibbesen, E. (1999). “Roughness indices for estimation of depression storage capacity of tilled soil surfaces.” Soil & Tillage Research, 52, 103-111.

Huang, C., and Bradford, J. M. (1990). “Digressional storage for Markov – Gaussian surfaces.” Water Resour. Res., 26(9), 2235-2242.

Huang, C., and Bradford, J, M. (1992). “Application of a laser scanner to quantify soil microtopography.” Soil Sci. Soc. Am. J., 56, 14-21.

Jenson, S. K. (1991). “Applications of hydrologic information automatically extracted from digital elevation models.” Hydrological Processes, 5(1), 31-44.

Jenson, S. K., and Domingue, J. O. (1988). “Extracting topographic structure from digital elevation data for geographic information system analysis.” Photogrammetric Engineering and Remote Sensing, 54(11), 1593-1600.

Kamphorst, E. C., and Duval, Y. (2001). “Validation of a numerical method to quantify depression storage by direct measurements on moulded surfaces.” Catena, 43, 1-14.

Kamphorst, E. C., Jetten, V., Guerif, J., Pitkanen, J., Iversen, B. V., Douglas, J. T., and Paz, A. (2000). “Predicting depression storage from soil surface roughness.” Soil Sci. Soc. Am. J., 64, 1749-1758.

Marks, D., Dozier, J., and Frew, J. (1984). “Automated basin delineation from digital elevation data.” Geo-Processing, 2, 299-311.

Martz, L. M., and Garbrecht, J. (1993). “Automated extraction of drainage network and watershed data from digital elevation models.” Water Resources Bulletin, 29(6), 901-908.

Mwendera, E. J., and Feyen, J. (1992). “Estimation of depression storage and Manning’s resistance coefficient from random roughness measurements.” Geoderma, 52, 235-250.

O’Callaghan, J. F., and Mark, D. M. (1984). “The extraction of drainage networks from digital elevation data.” Computer Vision, Graphics and Image Processing, 28, 323-344.


10

Onstad, C. A. (1984). “Depression storage on tilled soil surfaces.” Transactions of the ASAE, 27(3), 729-732.

Planchon, O., and Darboux, F. (2002). “A fast, simple and versatile algorithm to fill the depressions of digital elevation models.” Catena, 46, 159-176.

Ullah, W., and Dickinson, W. T. (1979). “Quantitative description of depression storage using a digital surface model: I. determination of depression storage.” J. Hydrol., 42, 63-75.


Documents

An Improved Method for Watershed Delineation and ...€¦ · performing puddle delineation, executing the computation of depression storages, determining flow directions and accumusummary