Classification of drainage network types in the arid and semi-arid regions of Arizona and California

doi:10.1016/j.jaridenv.2017.04.013

Journal of Arid Environments

Volume 144, September 2017, Pages 60-73

https://doi.org/10.1016/j.jaridenv.2017.04.013 Get rights and content

Highlights

•
Drainage networks in arid and semi-arid regions are examined for their network types.
•
Pinnate networks occur in arid and semi-arid areas with low preexisting slopes.
•
Pinnate networks are observed in the USA.

Abstract

The present study aims to classify channel network types in arid and semi-arid regions and to determine whether pinnate networks occur in these areas by using scaling invariance measures. The average slope of the preexisting topographic surface is also studied to identify whether the slope affects the development of drainage networks in such arid and semi-arid regions. Twenty channels in the states of Arizona and California, USA, were analyzed. In this study, the measures used to classify drainage networks and to obtain the Hurst exponent are the drainage area increments, the channel course irregularity, and the tributary junction angles. The average slope of the preexisting surfaces is calculated from sub-watershed areas that are small and topographic curvatures that are close to zero. For all channels, the Hurst exponent is less than 0.91 which indicates that networks seem to be self-affine instead of self-similar. These channels present pinnate or parallel networks although the associated preexisting slope is below 3%. Such observations suggest that the preexisting slope does not have a significant influence on the development of pinnate networks in these arid and semi-arid regions.

Section snippets

Introduction and review

The occurrence of drainage networks in different regions depends on local physiographic and climatic characteristics which produce specific attributes that define the network structure and impact the basin's hydrologic response (Berger and Entekhabi, 2001). Classification systems have been developed to distinguish various channel network types including dendritic, parallel, and pinnate networks (Zernitz, 1932, Parvis, 1950, Howard, 1967, Phillips and Schumm, 1987, Mejia and Niemann, 2008). A

Data set

Twenty washes represented by intermittent streams with erodible surfaces were randomly selected in the arid and semi-arid regions of the USA. Google Earth with a topomaps tool was used to identify the washes and their elevations. Fig. 2 illustrates the outlets of the 20 washes (diamonds) and the counties limits (lines) in Arizona and California.

The washes are located in Arizona and California, USA, and their locations and basin sizes are presented in Table 1. The basin sizes range from 5 to

Computation of the preexisting surface slope

The procedure for identification of the preexisting surfaces follows Jung et al. (2011) and is briefly described herein. Two main thresholds are required to estimate the slope of preexisting surfaces, which can be considered as regions relatively unaffected by modern fluvial processes (Jung et al., 2011). For the first threshold, the smallest drainage area altered by hillslope processes in the network is used for estimating channel head locations to avoid including large drainage areas which

Slope of preexisting surfaces

The slopes of preexisting surfaces for the 20 washes described in the data set section are estimated to identify the conditions for the occurrence of drainage networks in arid and semi-arid regions. The basins of the 20 washes of the region of study include a larger proportion of preexisting surfaces than was observed for river basins in non-arid regions of the USA (see Jung et al., 2011). For instance, the proportions of preexisting surfaces for a dendritic network and a parallel network in

Discussion

The analysis of 20 wash networks in the arid and semi-arid regions of Arizona and California is performed to identify the preexisting slopes of the networks. All the wash networks studied in the present paper show small preexisting slopes below 3%. This might be because arid environments affect the characterization of the slopes of preexisting surfaces in the washes featured by erodible surfaces.

We also examined the classification of drainage network types for the 20 wash networks based on

Conclusions

The main objectives of the present research are to identify channel network types in the arid and semi-arid regions of the USA and investigate the occurrence of pinnate networks by using three measures derived from scaling invariance. These measures are the drainage area increment, the channel course irregularity, and the tributary junction angle. The preexisting topographic slope is also examined to determine whether it influences the occurrence of different types of drainage networks in the

Acknowledgements

The authors are grateful to the Associate Editor and to the anonymous reviewer for their comments which greatly helped improve the quality of the manuscript. This research was supported by a grant(14AWMP-B082564-01) from Advanced Water Management Research Program funded by Ministry of Land, Infrastructure and Transport of Korean government.

References (44)

A.D. Abrahams et al.
Drainage density in relation to precipitation intensity in the USA
J. Hydrol.
(1984)
K.P. Berger et al.
Basin hydrologic response relations to distributed physiographic descriptors and climate
J. Hydrol.
(2001)
S. El Adlouni et al.
On the tails of extreme event distributions in hydrology
J. Hydrol.
(2008)
E.J. Ijjasz-Vasquez et al.
Scaling regimes of local slope versus contributing area in digital elevation models
Geomorphology
(1995)
K. Jung et al.
Improved classification of drainage networks using junction angles and secondary tributary lengths
Geomorphology
(2015)
K. Jung et al.
Under what conditions do parallel river networks occur?
Geomorphology
(2011)
J.F. O'Callaghan et al.
The extraction of drainage networks from digital elevation data
Comput. Vis. Graph. Image Process
(1984)
J.A. Thompson et al.
Digital elevation model resolution: effects on terrain attribute calculation and quantitative soil-landscape modeling
Geoderma
(2001)
S. Wu et al.
Modeling the effects of elevation data resolution on the performance of topography-based watershed runoff simulation
Environ. Modell. Softw.
(2007)
A.D. Abrahams et al.
Geological controls on the topological properties of some trellis channel networks
Geol. Soc. Am. Bull.
(1983)

J.P. Akers

Geology and ground water in the central part of Apache County, Arizona

U.S. Geol. Surv. Water-Supply Pap.

(1964)

D. Argialas et al.

Quantitative description and classification of drainage patterns

Photogramm. Eng. Remote Sens.

(1988)

S. Castelltort et al.

Slope-control on the aspect ratio of river basins

Terra nova.

(2009)

K. Chang et al.

The effect of DEM resolution on slope and aspect mapping

Cartogr. Geogr. Inf.

(1991)

F. Chebana et al.

Multivariate L-moment homogeneity test

Water Resour. Res.

(2007)

F. Chebana et al.

Depth and homogeneity in regional flood frequency analysis

Water Resour. Res.

(2008)

W.E. Dietrich et al.

Geomorphic transport laws for predicting landscape form and dynamics

Predict. Geomorphol.

(2003)

I.S. Evans

General geomorphometry, derivatives of altitude, and descriptive statistics

Spatial Analysis Geomorphol.

(1972)

A.D. Howard

Drainage analysis in geologic interpretation: a summation

Am. Assoc. Pet. Geol. Bull.

(1967)

Y. Hundecha et al.

Regional estimation of parameters of a rainfall-runoff model at ungauged watersheds using the “spatial” structures of the parameters within a canonical physiographic-climatic space

Water Resour. Res.

(2008)

C. Ichoku et al.

A numerical approach to the analysis and classification of channel network patterns

Water Resour. Res.

(1994)

C.R. Longwell

Geology of the muddy mountains, Nevada, with a section to the grand wash cliffs in western Arizona

Am. J. Sci.

(1921)

Cited by (13)

Automatic segmentation of parallel drainage patterns supported by a graph convolution neural network
2023, Expert Systems with Applications
Citation Excerpt :
The DP classification is to classify the overall structure of the drainage. Some researchers have preliminarily achieved it through empirical threshold methods based on drainage features (Bouramtane, Yameogo et al., 2020; Burr, Drummond, Cartwright, Black, & Perron, 2013; Pereira-Claren et al., 2019) and the self-similarity index method (Jung & Ouarda, 2017; Mejia & Niemann, 2008). However, drainage with certain patterns is often embedded in the complex drainage in a watershed.
Drainage pattern (DP) recognition is critical in hydrographic analysis, topography identification, and drainage characteristic detection. The traditional method is based on rule computation and self-similarity idea preliminarily performing the DP classification. However, DP segmentation is an uncertain spatial cognitive problem affected by enormous factors. To settle such a multi-conditions decision question, this study takes the segmentation of parallel drainage pattern (SPDP) as an example presenting a deep learning method, namely the graph convolution neural network (GCNN) based on Graph SAmple and aggreGatE (GraphSAGE). First, a directed graph and dual graph were used to construct a dual drainage graph recording spatial-cognition features of drainage. Second, nine drainage features were built to define the graph description from three perspectives: topological connectivity, meandering equilibrium, and directional unity. Finally, the GraphSAGE model was designed for SPDP and trained by typical samples to finish the segmentation works. The experiment examined the optimal feature combination and hyperparameter sensitivity, which can provide sufficient information for SPDP supported by GraphSAGE. Besides, our model outperformed other machine learning methods and GCNNs driven by a fixed quantity sampling mechanism and hydrological knowledge. This work provides a vital reference for hydrology research supported by combing hydrological knowledge with GCNNs.
Characterization and classification of river networks in South Korea
2022, Environmental Modelling and Software
Citation Excerpt :
Li et al. (2020) improved a D8-compatible high-efficient approach to determine channel heads which is identified as the geomorphological change point by calculating the geomorphologic parameter series along with flow paths. To analyze the conditions affecting river network development, natural river basins were also investigated in nonarid regions such as Colorado, Tennessee, Utah, and Wyoming and arid regions such as Arizona and California (Jung et al., 2011; Jung and Ouarda, 2017). For the river networks in nonarid regions, a transition between dendritic and parallel networks was identified, and the parallel network type was observed when the preexisting slope exceeded approximately 3%.
The aims of this study are to characterize river network patterns and identify whether the slope of preexisting surfaces relatively unaffected by modern landscape processes influences the formation of networks on the peninsula. The distributional properties of the tributary junction angles by the beta distribution are examined, and support vector machines based on the parameter estimates of the distribution are used to distinguish the different river network types. The classified river networks are validated using three measures: drainage area increment, stream course irregularity, and tributary junction angle derived from scaling invariance. 41 river networks are analyzed for the characterization and classification of networks in the region. Among the river networks, 12 networks represent the parallel network type and 29 networks represent the rectangular network type. These results imply that the preexisting slope does not seem to have a main effect on the development of the networks on South Korea.
Quantification of river network types based on hierarchical structures
2022, Catena
Citation Excerpt :
The parallel drainages of the Piceance Creek Basin and Mesa Verde in Colorado developed on surfaces formed on sedimentary rocks that dipped between 3° and 5° (Phillips and Schumm, 1987). The slope of the surface that remains unaffected by the current generation of topography, namely, the preexisting surface, seems important for parallel networks (Jung et al., 2011; Jung and Ouarda, 2017). Pinnate patterns tend to occur in steep terrain controlled by tectonics (Zernitz, 1932) and represent a prominent type of drainage in Moldova and southwestern Ukraine.
The development of river networks is closely related to landscape evolution processes and terrain analysis. Several methods for classifying river networks based on qualitative evaluation or quantitative indices have been developed. However, these methods pay little attention to the hierarchical structures of river networks and omit some generalized and detailed information about basins. In this work, a classification method based on the hierarchical structure of river networks is established. Drainage texture, flow direction and aspect ratios are employed to describe river network characteristics. The specific attributes calculated include the drainage density, the stream frequency of river networks and the maximum frequency, and the distribution uniformity of the flow direction of river reaches on each order. Classification trees are established using these metrics. The method is applied to classify 83 river networks in the Yellow River source zone ranging in size from 17 to 241 km². A previous classification method using basin invariance attributes was contrasted with our network structure approach, and cross-validation (CV) accuracies were obtained of 66.8% and 82.0% respectively. To verify the applicability of the proposed method, another 45 river networks classified in previous studies are considered. A CV accuracy of 76.7% is achieved with the proposed method. These results demonstrate that the proposed method based on the hierarchical structures of river networks is more accurate than methods that ignore the hierarchical structure.
Beyond the Long Profile
2022, Treatise on Geomorphology
The intricate dissection of mountainous landscapes by rivers and water divides has fascinated geomorphologists for generations. Early workers focused on qualitative descriptions of network patterns but without necessarily relating them to specific environmental drivers. Following a dramatic increase in the availability of high-resolution topographic data and computational power in recent decades, we are now able to quantify metrics related to the linkages between tectonics, rock properties, and fluvial processes as revealed in the geometry of fluvial networks. These links are a consequence of river networks being imprinted on the top boundary of the upper crust. This means that in tectonically active landscapes, surface landforms can advect together with shallow crustal rocks in response to deformation induced by tectonic stresses and to reorganize when deformation is intense. In this contribution, we explore factors that drive changes in channel network patterns and review quantifiable metrics for identifying such changes. We focus on three major tectono-lithologic drivers of change. First, in landscapes featuring transpression, rotation of channels can reveal competition between horizontal motions and erosion, and in landscapes with pure shortening, basin shape and aspect ratio may change. Second, in extensional settings, network patterns evolve when fault arrays nucleate, ramify, and then coalesce, leading to different patterns at each stage. Tectonic tilting could manifest in widespread reorganization with migrating drainage divides. Third, the lithology through which rivers carve their valleys appears to have a first-order control on the fluvial network geometry and fractal dimension. The juxtaposition of lithologies along tectonic contacts can also leave a revealing imprint on river networks.
Statistical approach of factors controlling drainage network patterns in arid areas. Application to the Eastern Anti Atlas (Morocco)
2020, Journal of African Earth Sciences
Several studies have revealed that the complexity in the distribution of drainage network patterns is not random and controlled by major parameters, variable in space but also throughout geological time. Drainage networks in the Eastern Anti-Atlas of Morocco consist of complex spatial arrangements with various types of patterns, such as trellis, angular, dendritic and parallel. The objective was to distinguish, quantify and rank the relationship that may exist between the different drainage networks patterns, geology and geomorphology. A total of 230 basins were extracted from the ASTER-GDEM Elevation Data (USGS), which were assigned 16 parameters reflecting their topography, morphometry, slope and geology. The statistical treatment of the dataset (16 variables x 230 observations) was carried out through principal component analysis (PCA), linear discriminant analysis (LDA) and agglomerative hierarchical clustering (AHC), in order to investigate the complexity of drainage network patterns and their distribution. The PCA showed that the topographical, slope and geological parameters, i.e. primarily the parameter associated with structural control, best explains the variation in the type of the drainage pattern. The LDA made it possible to distinguish between the four types of drainage patterns with a success rate of 90%, using 3 discriminant functions that were better correlated with geological and slope parameters. LDA and AHC statistical treatments show confusion between the parallel, trellis and angular patterns, on the one hand, due to similar factors responsible for their formation, and on the other because of transitions phenomenon from one drainage pattern to another over time or space. Such possible drainage network shifting may be explained by the geological events that have occurred in the Eastern Anti Atlas from Lower Mesozoic to the Quaternary.
Planform geometry and relief characterization of drainage networks in high-relief environments: An analysis of Chilean Andean basins
2019, Geomorphology
Citation Excerpt :
These basins have areas ranging between 60 and 2100 km2, and were classified by Mejía and Niemann (2008). The networks from low-relief regions include five types (Burr et al., 2009, 2013; Poore, 2010; Cartwright et al., 2011; Donadio et al., 2015; Jung et al., 2017; Jung and Ouarda, 2017): dendritic, parallel, pinnate, rectangular and trellis (Fig. 1). Those types were established in previous studies (Mejía and Niemann, 2008; Jung et al., 2015, 2019).
Drainage network types give indications of the landscape evolution processes and the lithologic and structural conditions that produced them. Moreover, drainage network attributes affect the hydrologic response of basins. Several network types have been identified through visual inspection and, more recently, through quantitative methods developed and applied primarily on relatively low-relief regions. These methods have not been tested in high-relief environments where other factors can control network evolution and other network types may exist. In this work, we study the occurrence of network types in high-relief environments and their potential identification through quantitative classification methods. We analyze 29 high-relief basins in the Chilean Andean region. After exploring the contextual properties (geological, climatic and landcover) of these networks, we calculate several metrics describing their horizontal and vertical attributes and benchmark them against those for network types in low-relief regions. Results show that the high-relief networks have three different types (herringbone, barbed, and high-relief pinnate), which differ in terms of their contextual properties as well as their horizontal and vertical attributes. Classification trees built using only vertical attributes have accuracies of 93% and 84% when differentiating among the high- and low-relief networks, respectively. Furthermore, an overall accuracy of 91% is reached when classifying the entire set of high- and low-relief networks with a combination of vertical and horizontal attributes. Overall, we conclude that classification of drainage patterns through vertical and horizontal attributes is feasible and more accurate than with horizontal attributes alone.

View all citing articles on Scopus

View full text

Classification of drainage network types in the arid and semi-arid regions of Arizona and California

Highlights

Abstract

Section snippets

Introduction and review

Data set

Computation of the preexisting surface slope

Slope of preexisting surfaces

Discussion

Conclusions

Acknowledgements

J. Hydrol.

J. Hydrol.

J. Hydrol.

Geomorphology

Geomorphology

Geomorphology

Comput. Vis. Graph. Image Process

Geoderma

Environ. Modell. Softw.

Geological controls on the topological properties of some trellis channel networks

Geol. Soc. Am. Bull.

Geology and ground water in the central part of Apache County, Arizona

U.S. Geol. Surv. Water-Supply Pap.

Quantitative description and classification of drainage patterns

Photogramm. Eng. Remote Sens.

Slope-control on the aspect ratio of river basins

Terra nova.

The effect of DEM resolution on slope and aspect mapping

Cartogr. Geogr. Inf.

Multivariate L-moment homogeneity test

Water Resour. Res.

Depth and homogeneity in regional flood frequency analysis

Water Resour. Res.

Geomorphic transport laws for predicting landscape form and dynamics

Predict. Geomorphol.

General geomorphometry, derivatives of altitude, and descriptive statistics

Spatial Analysis Geomorphol.

Drainage analysis in geologic interpretation: a summation

Am. Assoc. Pet. Geol. Bull.

Regional estimation of parameters of a rainfall-runoff model at ungauged watersheds using the “spatial” structures of the parameters within a canonical physiographic-climatic space

Water Resour. Res.

A numerical approach to the analysis and classification of channel network patterns

Water Resour. Res.

Geology of the muddy mountains, Nevada, with a section to the grand wash cliffs in western Arizona

Am. J. Sci.