On the transition of base flow recession from early stage to late stage

doi:10.1016/j.advwatres.2015.11.015

Advances in Water Resources

Volume 88, February 2016, Pages 8-13

https://doi.org/10.1016/j.advwatres.2015.11.015 Get rights and content

Highlights

•
Identify the transition of base flow recession from early stage to late stage.
•
Reveal the linkage between recession transition and ephemeral stream contraction.
•
Demonstrate the role of ephemeral stream contraction in early stage recession.

Abstract

This paper is focused on the transition of base flow recession from early stage to late stage. The volume flow rate that takes place when such a transition occurs is identified for each of the twenty-three recession events observed at the Panola Mountain Research Watershed (PMRW) in Georgia, USA, using a newly developed cumulative regression analysis method. Meanwhile, the flow at the watershed outlet, which was recorded when the discharge at the perennial stream head diminishes to zero, is identified for each recession event. As evidenced by a correlation coefficient of 0.90, these two characteristic flows are found to be highly correlated, suggesting a fundamental linkage between the transition of base flow recession from early stage to late stage and the drying up of ephemeral streams. During the early stage, the contraction of ephemeral streams largely controls the recession behavior, whereas in the late stage when perennial streams dominate the flowing streams, the contraction of flowing streams is minimal and groundwater hydraulics governs the recession behavior.

Introduction

Base flow recession curve in the dry period is a distinct hydrologic signature of a watershed. It provides important information to water managers for decision-making on water supply, irrigation, and management of water quality [19] and aquatic ecosystem services [30]. Baseflow recession has been extensively studied in the last few decades. By eliminating post-rainfall event time reference in hydrograph recession analysis, Brutsaert and Nieber [7] proposed a classical method to analyze the time rate of change in discharge as a function of discharge itself (i.e., -dQ/dt=f(Q)). Logarithmic graphs of ln(−dQ/dt) versus ln(Q) based on observed discharge demonstrate approximately linear relationships, suggesting the following power relationship [7]: $- \frac{d Q}{d t} = a Q^{b}$

The value of exponent b estimated using observed discharge differs considerably between early stage of recession with high discharge and late stage of recession with low discharge. A sharp change of the exponent value from early recession to late recession has been observed for many watersheds (e.g., [[17], [21], [22], [39]]).

The factors controlling the recession exponent values are complex, including groundwater hydraulics [7], the interconnection of groundwater flow systems [31], spatial heterogeneity of watershed properties [15], stream contraction [2], [3], [4], and evapotranspiration [28]. In those studies which focused on the role of groundwater hydraulics on recession behavior, the total length of stream contributing to base flow is assumed to be constant [8], [32]. Based on the analytical solutions of the Boussinesq equation for a horizontal aquifer with a fully penetrating stream channel, the value of b equals 3 during early recession, and it is 1 for a linearized solution and 1.5 for a non-linearized solution during late recession [7]. Aquifer parameters at the watershed scale, including saturated hydraulic conductivity, drainable porosity and aquifer depth, were estimated based on analytical solutions for early and late recession [7], [8], [26], [29].

The role of flowing stream contraction on recession behavior has been investigated in recent years [2], [20]. The contraction of flowing stream networks is a result of geomorphological characteristics [18]. Perennial streams are active for most of the year, depending upon local climatology and basin characteristics. Ephemeral streams are intermittently active, in response to individual rainfall events [6], [13], [37], and gradually dry up during the recession period [10], [38]. The length of active channel network is highly correlated with streamflow, and power-law relationships between flowing channel length and streamflow are usually identified [12], [14]. However, the linkage between drying up of ephemeral streams and ceasing of early recession has not been fully explored in the existing literature.

We argue that the contraction rate of ephemeral streams is significant at the early stage of recession; however, at the late stage of recession when all ephemeral streams have dried up and perennial streams are the only sources for flowing streams, the contraction rate of perennial streams is negligible. The objective of this paper is to identify the transition of base flow from early recession stage to late recession stage for individual recession events, and investigate the potential linkage between the transition of base flow recession and the contraction of ephemeral streams.

Section snippets

Panola Mountain Research Watershed and data

To investigate the linkage between the transition of base flow recession from early stage to late stage and the contraction of ephemeral streams, at the minimum, we need streamflow data simultaneously recorded at both the outlet and the perennial stream heads of a watershed. Such streamflow observations are rather rare to find since streamflow gauges are usually located on perennial streams. Fortunately, for the Panola Mountain Research Watershed (PMRW) streamflow observations are available at

Contraction of flowing stream to perennial stream head

The streamflow observations at the upstream gauge in the PMRW provide an opportunity to identify the time when flowing stream contracts to the perennial stream head. Using discharge time series of the upstream gauge, we identified the moment when the ephemeral streams dry up (Q_e → 0) and the corresponding discharge at the watershed outlet Q_p0. Another major ephemeral tributary is located in the southeastern side of the watershed. The drainage areas for both perennial stream heads are about

Flow at the outlet when the ephemeral stream dries up

As shown in Table 1, twenty-three recession events during the 1985–2007 period were selected for analysis. For instance, Fig. 2 shows the observed hydrographs at the outlet and the perennial stream head during the recession of January, 2007. After 57 mm of rainfall, the peak discharge at the outlet reached 5.85 mm day⁻¹; while the peak discharge at the perennial stream head is 7.53 mm day⁻¹. The discharge in volume at the outlet was higher than that at the perennial stream head. The discharge

Conclusions

This paper is focused on the transition of base flow recession from early stage to late stage. A cumulative regression analysis method is proposed to identify the transition flow quantitatively for individual recession events observed in the Panola Mountain Research Watershed. The identified transition flows are then compared with the discharge rates observed when the flowing stream contracts to the perennial stream head. It is found that these two characteristic flows are strongly correlated.

Acknowledgment

The research conducted in this paper is partially supported by the CGIAR Research Program on Water, Land and Ecosystems (WLE) through the project “Enhancing groundwater simulation in the IMPACT-Water model for assessment of groundwater irrigation sustainability and food production impacts” led by the International Food Policy Research Institute (IFPRI). The authors thank Jake Peters for providing the discharge and rainfall data for the Panola Mountain Research Watershed.

References (39)

BiswalB. et al.
Estimation of ‘drainable’ storage – a geomorphological approach
Adv Water Resour
(2015)
BiswalB et al.
‘Universal’ recession curves and their geomorphological interpretation
Adv Water Resour
(2014)
InamdarS.P. et al.
Contributions of riparian and hillslope waters to storm runoff across multiple catchments and storm events in a glaciated forested watershed
J Hydrol
(2007)
RuppD.E. et al.
Analytical assessment and parameter estimation of a low-dimensional groundwater model
J Hydrol
(2009)
ShawS.B. et al.
Examining individual recession events instead of a data cloud: using a modified interpretation of dQ/dt–Q streamflow recession in glaciated watersheds to better inform models of low flow
J Hydrol
(2012)
BevenK.J. et al.
A physically based variable contributing area model of basin hydrology
Hydrol Sci Bull
(1979)
BiswalB. et al.
Geomorphological origin of recession curves
Geophys Res Lett
(2010)
BiswalB. et al.
A general geomorphological recession flow model for river basins
Water Resour Res
(2013)
BlythK. et al.
A stream length study
Water Resour Res
(1973)
BrutsaertW. et al.
Regionalized drought flow hydrographs from a mature glaciated plateau
Water Resour Res
(1977)

BrutsaertW. et al.

Basin-scale geohydrologic drought flow features of riparian aquifers in the Southern Great Plains

Water Resour Res

(1998)

ClarkM.P. et al.

Consistency between hydrological models and field observations: linking processes at the hillslope scale to hydrological responses at the watershed scale

Hydrol Process

(2009)

DayD.

Drainage density changes during rainfall

Earth Surf Processes

(1978)

FreerJ. et al.

The role of bedrock topography on subsurface storm flow

Water Resour Res

(2002)

GodseyS.E. et al.

Dynamic, discontinuous stream networks: hydrologically driven variations in active drainage density, flowing channels and stream order

Hydrol Process

(2014)

GregoryK.J.

Drainage networks and climate

GregoryK.J. et al.

The variation of drainage density within a catchment

Hydrol Sci J

(1968)

HarmanC.J. et al.

Power law catchment-scale recessions arising from heterogeneous linear small-scale dynamics

Water Resour Res

(2009)

KrakauerN.Y. et al.

Stream recession curves and storage variability in small watersheds

Hydrol Earth Syst Sci

(2011)

Cited by (27)

A proposed composite Boussinesq equation for estimating baseflow recessions and storage-outflow relationship
2023, Journal of Hydrology
Baseflow recession analysis characterizes the interactions between ground and surface water, and many studies have focused on estimating the mathematical form of the relationship and its parameters. The Boussinesq equations can estimate baseflow recession and storage-outflow relationship. However, these equations assume either linear or nonlinear behavior, depending on the streamflow conditions. The linear equation cannot represent the nonlinear behavior of the baseflow, and the nonlinear equation cannot fit the lowest discharge values. In addition, the nonlinear equation is oversimplified and does not include baseflow from groundwater storage below the stream’s water level. Therefore, we propose a composite Boussinesq equation that includes the linear portion of baseflow below the stream’s water level, includes a nonlinear storage, and fits low discharge values. The performance of the linear, nonlinear, and composite methods are assessed using 1561 USGS stream gaging stations across the contiguous US (CONUS). The regression constants, which represent the aquifer characteristics, are estimated for all gages using the three methods. Results show that the composite equation results in smaller errors for estimating baseflow recession events and the storage-outflow relationship. Finally, four different variables, precipitation, evapotranspiration, average discharge, and drainage area, were analyzed to find the relationship between the watershed conditions and baseflow analysis in terms of the regression constants and the relative errors of the best fit curves. Results show that the nonlinearity increases in drier regions, and the relative error for all three methods increases in wet climate regions with high precipitation and evapotranspiration. The composite equation is a well-performing alternative to linear or nonlinear equations across all basin characteristics, allowing for estimating the full range of baseflow behavior.
A probabilistic framework for robust master recession curve parameterization
2023, Journal of Hydrology
Master recession curve (MRC) representing the long-term catchment streamflow recession is widely used to estimate catchment storage-discharge relationship and predict low flows. Recession analysis of the streamflow functional form ( $- d Q / d t \sim Q$ ) is an effective way to construct MRC. However, the great variability of recession processes among events makes it difficult to parameterize the recession processes by deterministic methods and indicates the recession rate could be thought of as a random variable. In this study, a probabilistic approach (Parameterized Binning-percentile Method, or PBM) is proposed to construct MRCs based on $- d Q / d t \sim Q$ analysis. The probabilistic PBM is introduced by describing the distribution of recession rate at partitioned $- d Q / d t$ intervals by a Gamma distribution. MRCs at any percentile can be obtained by fitting regenerated data points of $- d Q / d t \sim Q$ in each interval. The PBM is validated by both numerically generated and observed hydrographs. It shows that the PBM is robust in capturing the distribution of recession rate and can adapt to various observed hydrographs with different recession characteristics. It is more accurate to generate probabilistic MRCs compared to the individual recession method in $Q \sim t$ form and quantile regression in $- d Q / d t \sim Q$ form. Further, MRCs from traditional deterministic methods can viewed as special cases of the probabilistic framework. Our newly proposed probabilistic framework can be used to quantify the statistical distributions of low flows, which can be helpful in regional water resources management during dry seasons.
A new method for effective parameterization of catchment-scale aquifer through event-scale recession analysis
2023, Advances in Water Resources
Accurate estimation of aquifer properties is of critical importance in representing the interactions between hydraulically connected surface-subsurface systems. The variability in recession characteristics (i.e., decline rates and slope nonlinearity) shows that each recession event (RE) occurs under different hydraulic diffusivity conditions between the stream-hillslope. Therefore, the streamflow recession characteristics should be considered in the effective parameterization of the catchment-scale aquifer. This study presents a new method for estimating catchment-scale aquifer properties while accounting for distinct recession characteristics through event-scale recession analysis. In this work, we used the recently developed BE3S (Hong et al., 2020) for (1) explicit incorporation of river stage observations as forcing data to account for the distinct recession characteristics and (2) numerical prediction of outflow from a Boussinesq aquifer to comparatively evaluate analytically derived effective groundwater parameters. We also addressed the relevance of the net subsurface discharge (NSD) observations in event-scale recession analysis for the catchment-scale aquifer parameterization. Specifically, we found that the catchment-scale NSD data plays a role in determining the initial saturated aquifer thickness (D_ini) and the size of contributing aquifer (A_c), which varies in individual REs. The applicability of the proposed parameterization scheme was successfully validated for individual REs with distinct recession slopes, demonstrating that hydraulic diffusivity parameters can be estimated by explicitly incorporating recession characteristics. As a catchment-scale application of hydraulic groundwater theory, we expect that the proposed scheme for effective groundwater parameterization will contribute to the accurate representation of the catchment-scale stream-hillslope hydrologic continuum.
Dynamic coevolution of baseflow and multiscale groundwater flow system during prolonged droughts
2022, Journal of Hydrology
Citation Excerpt :
This hillslope-centric conceptualization greatly simplifies real world landscape conditions, such as complex topography and geology (Zecharias and Brutsaert, 1988; Tague and Grant, 2004; McGuire et al., 2005; Gleeson and Manning, 2008; Yao et al., 2017; Rapp et al., 2020; Condon et al., 2020), and it cannot represent hydrological processes occurring at multiple spatial scales beyond that of hillslopes. It cannot be used to explain watershed-scale emergent baseflow generation behavior with active contribution from deep bedrock groundwater exfiltration, such as the spatial scaling of baseflow proportion and solute concentration (Frisbee et al., 2011; Frisbee et al., 2017; Asano et al., 2020; Iwasaki et al., 2021) and the connection between dynamically active drainage density and baseflow recession characteristics at the watershed scale (Biswal and Marani, 2010; Mutzner et al., 2013; Ghosh et al., 2016; Prancevic and Kirchner, 2019). Baseflow can be sustained by a groundwater flow system (GWFS) characterized by flow paths of different scales, such as those within riparian zones and hillslopes, and watershed-scale deep groundwater flow paths connecting headwaters and main streams (Winter et al., 1998; Jiang et al., 2014; Rumsey et al., 2015).
Field and numerical studies suggest that baseflow is composed of waters from a spectrum of groundwater flow paths termed the Groundwater Flow System (GWFS) – from shallow hillslope contributions to watershed-scale deep circulation originating in headwaters and discharging into lowland rivers. Here, we explore the evolution of the GWFS under prolonged droughts to understand its dynamics and multiscale nature, and to elucidate its role in baseflow generation and recession at the watershed scale. We consider three drought scenarios of varying severity and simulate groundwater flow in a 2-D cross-section of an idealized watershed with deep permeable bedrock, tracking the evolution of flow paths, baseflow, and residence times during the recession process. We find that baseflow generation at different drainage stages, and within different subwatersheds, is influenced distinctly by flow paths of different scales, depending on the relative strength of the flow paths and the position of the subwatersheds relative to the recharge/discharge zones of the deeper watershed-scale groundwater circulation. Despite having the same local relief, geology, and climate, baseflow from each subwatershed has a distinct recession behavior and time-dependent residence time distribution. Also, the hydraulic and transport characteristics of baseflow generation co-evolve and are strongly affected by the connection state of the water table to subwatersheds. These findings suggest that asynchrony and dissimilarity of baseflow generation from hillslopes under the impact of the watershed-scale groundwater flow, and interactions with local-scale and intermediate-scale groundwater flow, must be taken into account when interpreting baseflow recession data and building conceptual baseflow models at the watershed scale.
An improved method to estimate the rate of change of streamflow recession and basin synthetic recession parameters from hydrographs
2022, Journal of Hydrology
Citation Excerpt :
The recession slope b is assumed to be a constant value in a catchment in this study, but several catchments show that the recession slopes tend to increase from high to low flow parts (for example, HNZ and NLD in Fig. 9b-c). The phenomenon is also found by other several studies (Ghosh et al., 2016; Gao et al, 2017; Tashie et al., 2020a). Thus, further study should put more attention on the variation of recession slope with magnitude of flow.
Recession analysis of -dQ/dt ∼ Q is popularly used for determining catchment storage-discharge relationship, estimating the aquifer hydraulic properties, and predicting low flow processes. However, there are uncertainties arising from the numerical approximation of -dQ⁄dt by the finite difference (ΔQ/Δt) for individual recession and the fitting cloud-like data of -dQ⁄dt ∼ Q for many recession events in a catchment. In this study, we proposed an improved variable time step increment method (IVTS) to minimize the effects of the noisy observations on estimating -dQ⁄dt, and a fitting method in terms of the binning-average of -dQ⁄dt ∼ Q data points to determine the basin synthetic recession parameters. The proposed methods are validated by using numerical generation of recessions with different noises and recession behaviors and the observed hydrographs from twenty catchments in Huai River Basin in China. The results from our proposed method are compared with those from the other popularly used methods. The IVTS method is proven to be robust for approximation of -dQ⁄dt, which can significantly reduce errors of the estimated -dQ⁄dt and the recession slope (b). Combining with the IVTS method, the fitting method in terms of the binning average by splitting range of -dQ⁄dt, instead of Q, could overcome the shortage of traditional recession analysis methods that underestimate the recession slope. Our study indicates that the more accurate and reliable calculation of -dQ⁄dt ∼ Q data and estimation of basin synthetic recession parameters could increase the objectivity in interpretation of recession behaviors and accuracy in perdition of low flow processes.
Incorporating channel network information in hydrologic response modelling: Development of a model and inter-model comparison
2017, Advances in Water Resources
Citation Excerpt :
Hence, subsurface flow is a distinct and complex phenomenon, and it needs to be treated differently from surface flow in a modeling framework. Modelling of hydrological response thus requires understanding of how water is redistributed among different flow paths and how the channel network morphology can help in developing conceptual models (Biswal and Marani, 2010; Ghosh et al., 2016; Gupta et al., 2010; Rodríguez-Iturbe and Rinaldo, 2001). Early hydrological models, based largely on the work of Horton (1933), partitioned total flow into two components: fast surface flow or surface flow, dominating during floods, and slow subsurface flow, dominating during recession periods.
Incorporation of channel network information in streamflow modelling is a well-accepted scientific practice now. In particular, channel network morphology based instantaneous unit hydrographs (IUHs) are widely used for modelling of flood response. However, very few attempts have been made so far to use channel networks for modelling total flow, not just flood flow. In this study, total flow is partitioned into pure surface flow (PSF) and mixed surface-subsurface flow (MSSF), which are then modelled separately by constructing channel network morphology based IUHs. For modelling total flow, the combined IUH is then obtained by introducing a splitting parameter that represents the relative proportions of PSF and MSSF. We compare the performance of the proposed geomorphology based routing structure and a variant with a commonly used routing structure – two linear reservoirs in parallel. The three routing structures are then integrated with a well-known water balance model to perform continuous streamflow modeling. We perform inter-model comparison quantitatively by considering eight performance metrics within a multi-objective framework as well as qualitatively by observing the simulated storage-discharge relationships. By performing the inter-model comparison for 71 catchments across the US, we find that the geomorphology based models perform better than the linear model for low flow related metrics. They are also better at capturing non-linear and dynamic relationship between catchment water storage and discharge. The geomorphology based models perform particularly well in northeastern and midwestern US, while no such region of dominance emerges for the linear routing based model. Results also indicate the possibility of using the proposed models to capture the dominant flow generation processes in a basin.

View all citing articles on Scopus

View full text

On the transition of base flow recession from early stage to late stage

Highlights

Abstract

Introduction

Section snippets

Panola Mountain Research Watershed and data

Contraction of flowing stream to perennial stream head

Flow at the outlet when the ephemeral stream dries up

Conclusions

Acknowledgment

Adv Water Resour

Adv Water Resour

J Hydrol

J Hydrol

J Hydrol

A physically based variable contributing area model of basin hydrology

Hydrol Sci Bull

Geomorphological origin of recession curves

Geophys Res Lett

A general geomorphological recession flow model for river basins

Water Resour Res

A stream length study

Water Resour Res

Regionalized drought flow hydrographs from a mature glaciated plateau

Water Resour Res

Basin-scale geohydrologic drought flow features of riparian aquifers in the Southern Great Plains

Water Resour Res

Consistency between hydrological models and field observations: linking processes at the hillslope scale to hydrological responses at the watershed scale

Hydrol Process

Drainage density changes during rainfall

Earth Surf Processes

The role of bedrock topography on subsurface storm flow

Water Resour Res

Dynamic, discontinuous stream networks: hydrologically driven variations in active drainage density, flowing channels and stream order

Hydrol Process

Drainage networks and climate

The variation of drainage density within a catchment

Hydrol Sci J

Power law catchment-scale recessions arising from heterogeneous linear small-scale dynamics

Water Resour Res

Stream recession curves and storage variability in small watersheds

Hydrol Earth Syst Sci