دانلود رایگان ترجمه مقاله اثر منبع DEM بر نسبت رتبه بندی هورتون – استرالر مبتنی بر GIUH (نشریه الزویر 2015)

اثر منبع DEM بر نسبت رتبه بندی هورتون – استرالر مبتنی بر GIUH برای آبخیزها در دو حوضه رودخانه هندوستان

Effect of DEM source on equivalent Horton–Strahler ratio based GIUH for catchments in two Indian river basins

Abstract

Horton–Strahler (H–S) concept has been extensively used for quantification of characteristics of a stream network since several decades. The quantified values are often sensitive to threshold area specified for initiation of streams to demarcate the network, and to the position of outlet of a catchment. This implies that inferences drawn based on derived characteristics for a stream network are likely to be inconsistent, which is undesirable. To address this, a strategy based on self-similarity properties of channel network was proposed recently by Moussa (2009), which involves estimation of equivalent H–S ratios using catchment shape descriptors that are independent of threshold area. This study investigates effectiveness of the strategy on 42 catchments of various sizes in two Indian river basins (Cauvery and Mahanadi). Effect of digital elevation model (DEM) source on estimates of equivalent H–S ratios and characteristics of Geomorphologic Instantaneous Unit Hydrograph (GIUH) derived based on the same are examined by considering SRTM and ASTER DEMs. Results indicate that self-similarity assumptions are valid for the Indian catchments. Comparison of equivalent GIUH derived for each of the catchments based on real channel network with that derived using different DEM sources indicated differences that could be attributed to DEM-based uncertainty associated with estimates of: (i) equivalent H–S ratios that are functions of the self-similarity properties of channel network, and (ii) equivalent length of highest order stream that depends on self-similarity properties and configuration/characteristics of stream network. This uncertainty cannot be ignored in hydrological studies.

1 Introduction

Distributed hydrological models are often used to model runoff generation mechanism in catchments. They require information on local hillslope profiles and stream network, as hillslopes control production of storm water runoff that is transported through the stream network towards the catchment outlet. Conventional practice is to assimilate information related to stream networks through qualitative classification on the basis of (i) basic patterns such as dendritic, parallel, trellis, rectangular, radial, annular, multi-basinal, contorted, or (ii) modified basic patterns such as subdendritic, pinnate, anastomotic, distributary, subparallel, fault trellis and recurved trellis (Zernitz, 1932; Howard, 1967). However, quantitative analysis of natural stream networks in hydrological systems (catchments) is quintessential for modeling their response. In this perspective, a few attempts were made to quantify stream networks. Among those, Horton (1945) was probably the first work that suggested a procedure involving ordering of streams and laws relating the number and length of streams of various orders. Later Strahler (1952, 1957, 1964) suggested modification to Horton’s stream ordering procedure for avoiding some ambiguities. The resulting procedure is widely referred to as Horton–Strahler (H–S) ordering scheme. Horton’s laws when implemented on a stream network quantified using H–S ordering scheme are referred to as H–S laws. The laws include (i) ‘bifurcation ratio’ relating stream numbers corresponding to streams of consecutive orders, (ii) ‘length ratio’ relating lengths of streams of successive orders, (iii) ‘area ratio’ relating areas drained by streams of successive orders. The ratios find use in establishing relations with the fractal nature of the channel network (e.g., Beer and Borgas, 1993; La Barbera and Rosso, 1989; Tarboton, 1996) and in modeling hydrological response from catchments using geomorphological concepts (e.g., Rodríguez-Iturbe and Valdés, 1979; Gupta et al., 1980). The concept of H–S laws has received some criticism (e.g., Scheidegger, 1965, 1968a,b; Moussa, 2009) owing to factors such as (i) inconsistency in classifying a river network into streams of various orders with change in scale of map and support/threshold area for initiation of first order streams and (ii) sensitivity to the position of outlet of catchment. The factor (i) implies that estimated H–S ratios (e.g., bifurcation ratio, area ratio and length ratio) and inferences drawn for a river network based on those ratios are conditional on support area, which is undesirable (Moussa and Bocquillon, 1996; Moussa, 2008a, 2009). To address this issue, researchers are devoting their efforts to arrive at effective strategies that alleviate the effect of support area on H–S ratios. Moussa and Bocquillon (1996) studied effect of threshold area on morphometric and scaling properties of channel network in three catchments (having areas in the range 75–16,250 km2 ) located in southern France, and developed a strategy based on self-similarity properties of channel network to define new catchment shape descriptors that are independent of threshold area. Those descriptors were subsequently used by Moussa (2009) to develop formulations for equivalent Horton–Strahler (H–S) ratios and GIUH that are independent of the threshold/support area chosen for extraction of stream network. Effectiveness of the equivalent H–S ratios and GIUH was demonstrated by Moussa (2009) through application to seven catchments in France whose areas ranged from 738 km2 to 5346 km2 . There is dearth of further attempts to examine potential of the strategy elsewhere in the world. Further, there is a need to examine: (i) potential of the strategy on catchments of relatively larger size, and (ii) sensitivity of estimates of equivalent H–S ratios to source of DEM. In this perspective, investigations are carried out in this study on a large number of catchments having much wider range in their areas located in Cauvery and Mahanadi river basins, India, with the following two objectives: (1) to test methodology of Moussa (2009) on the catchments, and consequently to verify whether hypothesis of ‘‘self-similarity’’ is valid for channel networks in those catchments, and (2) if the hypothesis of ‘‘self-similarity’’ is valid, then compare (i) morphometric properties of the channel networks, and (ii) equivalent H–S ratios and equivalent GIUHs for the catchments obtained corresponding to two different DEMs (ASTER and SRTM). The developed equivalent GIUHs could prove useful to derive unit hydrographs corresponding to desired durations for target sparsely gauged/ungauged locations in the Cauvery and Mahanadi river basins. The derived unit hydrographs could be used to simulate design flood events at the target locations that find use in hydrological design and risk assessment of water resources systems (e.g., Jain et al., 2000). The subsequent part of this paper is structured as follows. First, background information is provided on H–S laws, their scaling properties, procedure for assessment of self-similarity properties of a channel network, and estimation of equivalent H–S ratios. Following this, case study on catchments in Cauvery and Mahanadi river basins is presented. GIUHs constructed based on the conventional H–S ratios and equivalent GIUHs determined based on derived equivalent H–S ratios are compared for each of the catchments. Finally conclusions drawn based on the investigations are provided.