Human population growth and long-term climate change, among other factors, put pressure on water resources worldwide [1
]. Groundwater resources, which are typically thought of as being more resilient towards climatic changes, have specifically been under increasing pressure, as more than 1.5 billion people globally rely on groundwater resources to meet their water demands [3
]. This suggests the importance of having accurate information on aquifer processes and characteristics across scales (both locally and regionally) in order to not only understand but also sustainably develop and manage groundwater resources. Conventionally, information about groundwater resources and associated aquifer properties is often assessed through point observations such as pumping test techniques. This can, however, make it difficult to obtain information for groundwater aquifers in remote regions given the cost of these methods and the limited scale at which they are applicable.
Among the aquifer characteristics of interest for water resources, transmissivity is among the most widely considered, given its prime role in subsurface fluid flow. Aquifer transmissivity is a hydraulic property defined as the ability of an aquifer to transmit water through its saturated thickness with the prevailing kinematic viscosity [4
]. Mathematically, aquifer transmissivity (T
) can be expressed as the product of the hydraulic conductivity (kc
) and saturated thickness of the aquifer (b
]. It should be noted that while K
is typically reserved for hydraulic conductivity in the literature, we have opted for kc
here to avoid confusion with the recession coefficient, which also is denoted by K
. Field-based assessments of transmissivity via pumping tests have been standard practice in groundwater hydrology and engineering for a long period of time. For example, Theis [5
] used the type-curve matching method to estimate aquifer transmissivity from pumping test data. Cooper and Jacob [6
] used a straight-line method to estimate aquifer transmissivity from pumping test data. Such assessments are, however, costly [7
], as they require at least two boreholes—one for pumping and one for observing water table (for unconfined aquifer) and/or piezometric level (for confined aquifer) drawdowns. This is especially problematic in developing regions (including much of the “Global South”), where limited resources and infrastructure are available to facilitate such approaches. Further, these regions are often faced with data scarcity and quality issues that already potentially limit sustainable development and management of water resources [10
]. As such, we face the dilemma that the locations globally for which we most desperately need information about aquifer properties for resource management are often those that are the most difficult to assess and access.
Because of cost-based limitations associated with drilling bore holes and conducting pumping tests, there is interest (and fundamental need) for the development of alternative cost-effective approaches to estimate the aquifer hydraulic parameters such as transmissivity. To this end, hydrograph recession techniques that utilize streamflow records have been suggested for the estimation of aquifer properties [7
]. These techniques utilize recession flow (akin to the falling limb on a hydrograph) as part of their input data, assuming these periods of time are representative of streamflow originating from the groundwater or other delayed sources [16
]. Given that streamflow data are often more readily available than groundwater data, approaches that leverage recession flows might be particularly useful in data-limited and remote environments. For example, Lyon et al. [14
] utilized streamflow recession to determine characteristic drainage timescale variability and improve understanding of hydrological processes across the Kilombero Valley located in Tanzania. Mendoza et al. [7
] estimated the transmissivity and specific yield of a semi-arid mountainous watershed in rural Mexico by analyzing recession flow.
Of the various approaches that consider recession flows, the recession-curve-displacement method—originally developed by Rorabaugh [15
]—is one method of particular interest as a result of its potential applicability for the estimation of aquifer characteristics (specifically transmissivity and recharge) in remote regions. For example, Huang et al. [17
] used the recession-curve-displacement method to estimate the streamflow-derived aquifer transmissivity of Kaoping River basin, southern Taiwan. They demonstrated that the approach could be useful in catchments for which direct estimates of transmissivity through pumping tests are limited. Chen and Lee [18
] used the recession-curve-displacement method to estimate the groundwater recharge of the Cho-Shui River basin, Taiwan. Likewise, Abo and Merkel [19
] employed the recession-curve-displacement method to estimate the groundwater recharge of the Al Zerba region of Aleppo, Syria. These studies reported on the applicability of the method in real-world settings given enough information. The recession-curve-displacement method is however difficult to apply and interpret in catchments with (i) regulated flow [20
], (ii) diverted flows [20
], and (iii) evapotranspiration from the shallow aquifer [21
Historically, the recession-curve-displacement method is applied manually [13
], allowing for some subjectivity on the part of the analyst. This is a major concern with regard to its applicability and reliability (particularly in data-limited regions). However, Rutledge [24
] automated (computerized) the recession-curve-displacement method within the computer programs RECESS [24
] and RORA [24
] to facilitate the work flow. Coupling between these programs greatly increases the speed of analysis and reduces the potential for subjectivity inherent in manual analysis [20
]. Subsequently, the recession-curve-displacement method as applied within these programs has been purported to provide reasonable results [17
]. Despite such advances and positive developments, there are still questions surrounding the representative scale and regional applicability with regard to using the recession-curve-displacement method for deriving aquifer transmissivity. Thus, there is need for comparisons of the method with results of pumping tests across geological and climatic settings—particularly in regions for which information about groundwater aquifers is potentially limited while groundwater resources are simultaneously being targeted for development (which typifies much of the Global South).
In this study, we test the approach from Huang et al. [17
] to estimate streamflow-derived transmissivity from streamflow records using the recession-curve-displacement method. We compare these regional-scale estimates to local-scale estimates made from pumping tests across the Kilombero Valley of Tanzania. We selected this basin as a potentially representative case study of the Global South because it (1) has limited available data (both in extent and representativeness), while at the same time, it (2) is facing potentially rapid development, particularly through intensification of irrigation agriculture. Our central aim is to assess whether streamflow-derived transmissivity estimates are as useful as pumping-test estimated aquifer transmissivity values in this region. If they are, then there could be considerable potential to inform hydrologic model development and water resource management regionally, as streamflow records tend to be more available (in terms of both space and time) than pumping test data.