Suitability of satellite-based hydro-climate variables and machine learning for streamflow modeling at various scale watersheds

ABSTRACT

Streamflow modeling is essential to investigate processes in the hydrologic cycle and important for water resource management application. However, in-situ hydrologic data paucity, because of various factors such as economic, political, instrument malfunctioning, and poor spatial distribution, makes the modeling process challenging. To overcome this limitation, we introduced a satellite remote sensing-based machine learning approach – boosted regression tree (BRT) – that integrates spatial land surface and climate variables that describe the sub-units, and applied it in three variable size watersheds in the Upper Mississippi River Basin (UMRB), USA. The model simulation results were tested using an independent dataset and showed Nash–Sutcliffe efficiency values of 0.80, 0.76, and 0.69 for the UMRB, Illinois River Watershed, and Raccoon River Watershed, respectively. In addition, we compared the performance of the machine learning models with existing process-based modeling results. Overall performance is comparable with the process-based approaches, but with significantly less modeling effort and resources.     

Multi‐site evaluation of hydrology component of SWAT in the coastal plain of southwest Georgia

Simulation of watershed scale hydrologic and water quality processes is important for watershed assessments. Proper characterization of the accuracy of these simulations, particularly in cases with limited observed data, is critical. The Soil & Water Assessment Tool (SWAT) is frequently used for watershed scale simulation. The accuracy of the model was assessed by extrapolating calibration results from a well studied Coastal Plain watershed in Southwest Georgia, USA, to watersheds within the same geographic region without further calibration. SWAT was calibrated and validated on a 16.7‐km² subwatershed within the Little River Experimental Watershed by varying six model parameters. The optimized parameter set was then applied to a watershed of similar land use and soils, a smaller watershed with different land use and soils and three larger watersheds within the same drainage system without further calibration. Simulation results with percent bias (PB) ±15% ≤ PB < ±25% and Nash–Sutcliffe efficiency (NSE) 0.50 < NSE ≤ 0.65 were considered to be satisfactory, whereas those with PB < ±10% and 0.75 < NSE ≤ 1.00 were considered very good. With these criteria, simulation results for the five non‐calibration watersheds were satisfactory to very good. Differences across watersheds were attributed to differences in soils, land use, and surficial aquifer characteristics. These results indicate that SWAT can be a useful tool for predicting streamflow for ungauged watersheds with similar physical characteristics to the calibration watershed studied here and provide an indication of the accuracy of hydrologic simulations for ungauged watersheds. Copyright © 2012 John Wiley & Sons, Ltd.     

Initial abstraction and curve numbers for semiarid watersheds in Southeastern Arizona

The Soil Conservation Service (SCS) curve number (CN) estimates of direct runoff from rainfall for semiarid catchments can be inaccurate. Investigation of the Walnut Gulch Experimental Watershed (WGEW) (Southeastern Arizona) and its ten nested catchments determined that the inaccuracy is due to the original SCS ratio (λ) of 0.2 between initial abstraction and maximum potential retention. Sensitivity analyses indicate that runoff estimation can be very sensitive to the initial abstraction ratio, especially for relatively low rainfall amount and for watersheds covered by deep, coarse, and porous soil, conditions that dominate many semiarid watersheds worldwide. Changing the ratio of initial abstraction to the maximum potential retention to optimal values ranging from 0.01 to 0.53 for different Walnut Gulch catchments improved runoff estimates. The greater the channel area and the finer the soil, the smaller the initial abstraction ratio is. The variation of the initial abstraction ratio for the WGEW is due to the variation of maximum potential retention and initial abstraction, which are channel area and soil‐dependent parameters. The greater the channel area, the higher the maximum potential retention S is, and the coarser the soil, the larger the initial abstraction I_a is. In addition, the effect of initial abstraction ratio on runoff estimation increases with decreasing CN. Thus, impacts of initial abstraction ratio on runoff estimation should be considered, especially for semiarid watersheds where the CN is usually low. Copyright © 2012 John Wiley & Sons, Ltd.     

Little River Experimental Watershed,a keystone in understanding of coastal plain watersheds

The US Department of Agriculture-Agricultural Research Service Southeast Watershed Research Laboratory (SEWRL) initiated a hydrologic research program on the Little River Experimental Watershed (LREW) in 1967. Long-term (52 years) streamflow data are available for nine sites, including rainfall-runoff relationships and hydrograph characteristics regularly used in research on interactive effects of climate, vegetation, soils, and land-use in low-gradient streams of the US EPA Level III Southeastern Plains ecoregion. A summary of prior research on the LREW illustrates the impact of the watershed on building a regional understanding of hydrology and water quality. Climatic and streamflow data were used to make comparisons of scale across the nine nested LREW watersheds (LRB, LRF, LRI, LRJ, LRK, LRO, LRN, LRM, and LRO3) and two regional watersheds (Alapaha and Little River at Adel). Annual rainfall for the largest LREW, LRB, was 1200 mm while average annual streamflow was 320 mm. Annual rainfall, streamflow, and the ratio between annual streamflow and rainfall (Sratio) were similar (α = 0.05) across LREWs LRB, LRF, LRI, LRJ, LRK, and LRO. While annual rainfall within the 275 ha LRO3 was found to be similar to LRO and LRM (α = 0.05), annual streamflow and Sratio were significantly different (α = 0.05). Comparisons of annual rainfall, streamflow, and Sratio between LRB and the regional watersheds indicated no differences (α = 0.05). Based upon this analysis, most regional watersheds shared similar hydrologic characteristics. LRO3 was an exception, where increases in row crops and decreases in forest coverage resulted in increased streamflow. LREW data have been instrumental in building considerable scientific understanding of flow and transport processes for these stream systems. Continued operation of the LREW hydrologic network will support hydrologic research as well as environmental quality and riparian research programs that address emerging and high priority natural resource and environmental issues.     

A process-based,distributed hydrologic model based on a large-scale method for surface–subsurface coupling

Process-based watershed models are useful tools for understanding the impacts of natural and anthropogenic influences on water resources and for predicting water and solute fluxes exported from watersheds to receiving water bodies. The applicability of process-based hydrologic models has been previously limited to small catchments and short time frames. Computational demands, especially the solution to the three-dimensional subsurface flow domain, continue to pose significant constraints. This paper documents the mathematical development, numerical testing and the initial application of a new distributed hydrologic model PAWS (Process-based Adaptive Watershed Simulator). The model solves the governing equations for the major hydrologic processes efficiently so that large scale applications become relevant. PAWS evaluates the integrated hydrologic response of the surface–subsurface system using a novel non-iterative method that couples runoff and groundwater flow to vadose zone processes approximating the 3D Richards equation. The method is computationally efficient and produces physically consistent solutions. All flow components have been independently verified using analytical solutions and experimental data where applicable. The model is applied to a medium-sized watershed in Michigan (1169 km²) achieving high performance metrics in terms of streamflow prediction at two gages during the calibration and verification periods. PAWS uses public databases as input and possesses full capability to interact with GIS datasets. Future papers will describe applications to other watersheds and the development and application of fate and transport modules.     

Variations in trace element geochemistry in the Seine River Basin based on floodplain deposits and bed sediments

Between 1990 and 1995 a series of bed sediment, suspended sediment and fresh floodplain samples were collected within the Seine River Basin, in France, to evaluate variations in trace element geochemistry. Average background trace element levels for the basin were determined from the collection and subsequent analyses of bed sediment samples from small rural watersheds and from a prehistoric (5000 BP) site in Paris. Concentrations are relatively low, and similar to those observed for fine‐grained bed sediments from unaffected areas in the United States and Canada. However, the concentrations are somewhat higher than the reference levels presently adopted by French water authorities for areas north of the Seine Basin, which have similar bedrock lithologies. Downstream trace element variations were monitored in 1994 and 1995 using fresh surficial floodplain samples that were collected either as dried deposits a few days after peak discharge, or immediately after peak discharge (under ≤30 cm of water). Chemical comparisons between fresh floodplain deposits, and actual suspended sediments collected during flood events, indicate that, with some caveats, the former can be used as surrogates for the latter. The floodplain sediment chemical data indicate that within the Seine Basin, from the relatively unaffected headwaters through heavily affected urban streams, trace element concentrations vary by as much as three orders of magnitude. These trace element changes appear to be the result of both increases in population as well as concomitant increases in industrial activity. (This article is a US government work and is in the public domain in the United States.)     

Automated calibration applied to watershed‐scale flow simulations

Complexity in simulating the hydrological response in large watersheds over long times has prompted a significant need for procedures for automatic calibration. Such a procedure is implemented in the basin‐scale hydrological model (BSHM), a physically based distributed parameter watershed model. BSHM simulates the most important basin‐scale hydrological processes, such as overland flow, groundwater flow and stream–aquifer interaction in watersheds. Here, the emphasis is on estimating the groundwater parameters with water levels in wells and groundwater baseflows selected as the calibration targets. The best set of parameters is selected from within plausible ranges of parameters by adjusting the values of hydraulic conductivity, storativity, groundwater recharge and stream bed permeability. The baseflow is determined from stream flow hydrographs by using an empirical scheme validated using a chemical approach to hydrograph separation. Field studies determined that the specific conductance for components of the composite hydrograph were sufficiently unique to make the chemical approach feasible. The method was applied to the Big Darby Creek Watershed, Ohio. The parameter set selected for the groundwater system provides a good fit with the estimated baseflow and observed water well data. Copyright © 1999 John Wiley & Sons, Ltd.     

The Kings River Experimental Watersheds: Infrastructure and data

The Kings River Experimental Watersheds (KREW) were established in 2002 to expand our knowledge of catchment physical, chemical, and biological processes in Sierra Nevada headwater forests, and to better understand the impacts of prescribed burning and forest thinning on these processes. Two elevation strata (high and low) were selected for the KREW sites, with four independent catchments and one nested catchment within each stratum. Both high and low elevation study areas were instrumented for continuous measurements of meteorology, streamflow, and turbidity. Atmospheric and stream chemistry, suspended sediment concentration, and bedload sediment delivery were measured on a regular schedule. Soil chemical and physical properties and vegetation were systematically sampled before and after the initial thinning and prescribed burning treatments, which were implemented between 2012 and 2016. Post-treatment data collection continues today as we explore opportunities for the second round of possible treatments. The critical research infrastructure and long-term baseline data collection has been instrumental in building partnerships with downstream managers, end users, non-governmental organizations, academic researchers, and national research programmes. Contributions to date include fundamental understanding of magnitude and variability of nutrient deposition; carbon, nutrient, and major ion dynamics in headwater streams; aquatic algae and macroinvertebrate populations; vegetation composition and structure; and streamflow responses to precipitation in the two elevation strata. Data from the experimental watersheds also support calibration and validation of diverse hydrologic models used for water resources planning.     

A simple index explains annual atrazine transport from surface runoff‐prone watersheds in the north‐central USA

Year‐to‐year dynamics in weather affect both the timing of application and the potential hydrologic transport of pesticides. Further, the most commonly used pesticides dissipate in the environment during the growing season. Interactions among these factors – hydrology, timing of application and dissipation kinetics – hinder the detection of temporal trends in transport. It is increasingly important to be able to discern such trends, to judge effectiveness of management practices or to determine whether observed changes were caused by management or weather. In previous work, a cumulative vulnerability index was developed to account for these three factors. It explained 63% of annual variation in atrazine load in the Goodwater Creek Experimental Watershed (GCEW). The objectives of the current work were (i) to generalize the cumulative vulnerability index to explicitly account for variation in watershed size, area treated with atrazine and average application rate; (ii) to test the overall performance on watersheds showing such variation; and (3) to test whether the generalized index properly accounted for the additional input parameters. The generalized index was tested using data from GCEW (73.7 km²) and seven additional watersheds in the northeast Missouri claypan region that varied in size from 212 to 1180 km² and from 4% to 23% of watershed area planted to corn or sorghum. Across 32 site‐years, the generalized index explained 84% of variation in annual atrazine load. Further, tests of residuals showed no dependence on either watershed area or fraction of area planted to corn and sorghum, indicating that these parameters were properly integrated into the index. The performance of the index supports the conclusion that data obtained from GCEW is representative of the Mark Twain Lake Basin and likely the entire Central Claypan Major Land Resource Area. Published 2012. This article is a U.S. Government work and is in the public domain in the USA.     

Identification of changes in floods and flood regimes in Canada using a peaks over threshold approach

Recent flood events in Canada have led to speculation that changes in flood behaviour are occurring; these changes have often been attributed to climate change. This paper examines flood data for a collection of 132 gauging stations in Canada. All of these watersheds are part of the Canadian Reference Hydrometric Basin Network (RHBN), a group of gauging stations specifically assembled to assist in the identification of the impacts of climate change. The RHBN stations are considered to have good quality data and were screened to avoid the influences of regulation, diversions, or land use change. Daily flow data for each watershed are used to derive a peaks over threshold (POT) dataset. Several measures of flood behaviour are examined based on the POT data, which afford a more in‐depth analysis of flood behaviour than can be obtained using annual maxima data. Analysis is conducted for four time periods ranging from 50 to 80 years in duration; the latter period results in a much smaller number of watersheds that have data for the period. The changes in flood responses of the watersheds are summarized by grouping the watersheds by size (small, medium, and large) and also by hydrologic regime (nival, mixed, and pluvial). The results provide important insights into the nature of the changes that are occurring in flood regimes of Canadian rivers, which include more flood exceedances, reduced maximum flood exceedance magnitudes for snowmelt events, and earlier flood events. Copyright © 2016 John Wiley & Sons, Ltd.     

Does hillslope trenching enhance groundwater recharge and baseflow in the Peruvian Andes?

As Andean glaciers rapidly retreat due to climate change, the balance of groundwater and glacial meltwater contributions to stream discharge in tropical, proglacial watersheds will change, potentially increasing vulnerability of water resources. The Shullcas River Watershed, near Huancayo, Peru, is fed only partly by the rapidly receding Huaytapallana glaciers (<20% of dry season flow). To potentially increase recharge and therefore increase groundwater derived baseflow, the government and not‐for‐profit organizations have installed trenches along large swaths of hillslope in the Shullcas Watershed. Our study focuses on a nonglacierized subcatchment of the Shullcas River Watershed and has 2 objectives: (a) create a model of the Shullcas groundwater system and assess the controls on stream discharge and (b) investigate the impact of the infiltration trenches on recharge and baseflow. We first collected hydrologic data from the field including a year‐long hydrograph (2015–2016), meteorological data (2011–2016), and infiltration measurements. We use a recharge model to evaluate the impact of trenched hillslopes on infiltration and runoff processes. Finally, we use a 3‐dimensional groundwater model, calibrated to the measured dry season baseflow, to determine the impact of trenching on the catchment. Simulations show that trenched hillslopes receive approximately 3.5% more recharge, relative to precipitation, compared with unaltered hillslopes. The groundwater model indicates that because the groundwater flow system is fast and shallow, incorporating trenched hillslopes (~2% of study subcatchment area) only slightly increases baseflow in the dry season. Furthermore, the location of trenching is an important consideration: Trenching higher in the catchment (further from the river) and in flatter terrain provides more baseflow during the dry season. The results of this study may have important implications for Andean landscape management and water resources.     

Determination of predevelopment denudation rates of an agricultural watershed (Cayaguás River, Puerto Rico) using in-situ-produced Be in river-borne quartz

Accurate estimates of watershed denudation absent anthropogenic effects are required to develop strategies for mitigating accelerated physical erosion resulting from human activities, to model global geochemical cycles, and to examine interactions among climate, weathering, and uplift. We present a simple approach to estimate predevelopment denudation rates using in-situ-produced cosmogenic ¹⁰Be in fluvial sediments. Denudation processes in an agricultural watershed (Cayaguás River Basin, Puerto Rico) and a matched undisturbed watershed (Icacos River Basin) were compared using ¹⁰Be concentrations in quartz for various size fractions of bed material. The coarse fractions in both watersheds bear the imprint of long subsurface residence times. Fine material from old shallow soils contributes little, however, to the present-day sediment output of the Cayaguás. This confirms the recent and presumably anthropogenic origin of the modern high denudation rate in the Cayaguás Basin and suggests that pre-agricultural erosional conditions were comparable to those of the present-day Icacos.     

Effects of landuse change on surface runoff and sediment yield at different watershed scales on the Loess Plateau

Erosion and sediment yield from large and small watersheds exhibit different laws. Variations in surface runoff and sediment yield because of landuse change in four watersheds of different scales from 1 km2 to 73 km2 were analyzed. Due to reforestation and farmland terracing, surface runoff and sediment yield reduced by 20-100% and 10-100% respectively. Reductions in surface runoff were differed significantly under different precipitation regimes. For the large watershed (73 km2) landuse change had similar effects on surface runoff regardless of changing of precipitation. For the small watershed (1 km2) landuse change had fewer effects on surface runoff under high precipitation. The relative changes of sediment yield in the four watersheds under reforestation and farmland terracing decreased as precipitation increased from 350 mm to 650 mm, then increased as precipitation increased from 650 mm to 870 mm. Where initial forest coverage rate was below 45%, sediment yield decreased dramatically as forest coverage rate increased. Watershed management with aiming at reducing both surface runoff and sediment yield should be conducted both on sloping surfaces and in channels in large watersheds.     

The Colorado East River Community Observatory Data Collection

The U.S. Department of Energy's (DOE) Colorado East River Community Observatory (ER) in the Upper Colorado River Basin was established in 2015 as a representative mountainous, snow-dominated watershed to study hydrobiogeochemical responses to hydrological perturbations in headwater systems. The ER is characterized by steep elevation, geologic, hydrologic and vegetation gradients along floodplain, montane, subalpine, and alpine life zones, which makes it an ideal location for researchers to understand how different mountain subsystems contribute to overall watershed behaviour. The ER has both long-term and spatially-extensive observations and experimental campaigns carried out by the Watershed Function Scientific Focus Area (SFA), led by Lawrence Berkeley National Laboratory, and researchers from over 30 organizations who conduct cross-disciplinary process-based investigations and modelling of watershed behaviour. The heterogeneous data generated at the ER include hydrological, genomic, biogeochemical, climate, vegetation, geological, and remote sensing data, which combined with model inputs and outputs comprise a collection of datasets and value-added products within a mountainous watershed that span multiple spatiotemporal scales, compartments, and life zones. Within 5 years of collection, these datasets have revealed insights into numerous aspects of watershed function such as factors influencing snow accumulation and melt timing, water balance partitioning, and impacts of floodplain biogeochemistry and hillslope ecohydrology on riverine geochemical exports. Data generated by the SFA are managed and curated through its Data Management Framework. The SFA has an open data policy, and over 70 ER datasets are publicly available through relevant data repositories. A public interactive map of data collection sites run by the SFA is available to inform the broader community about SFA field activities. Here, we describe the ER and the SFA measurement network, present the public data collection generated by the SFA and partner institutions, and highlight the value of collecting multidisciplinary multiscale measurements in representative catchment observatories.     

Using integrated modelling to understand seasonal vegetation dynamics and its relationship to runoff generation

Vegetation dynamics and hydrological processes are major components of terrestrial ecosystems, and they interact strongly with each other. Studies of hydrological responses to vegetation dynamics are usually conducted on a long-term scale, whereas the hydrological responses within a single year have rarely been studied. In the present study, Poyang Lake runoff (PYL-R) model, a new hydrological model coupled with leaf area index (LAI) remote sensing products, was established and applied to simulate the runoff process in the Poyang Lake Watershed. The simulation results obtained in three sub-watersheds of the Poyang Lake Watershed (Ganjiang Watershed, Xinjiang Watershed, and Fuhe Watershed) agreed well with the observations (Nash efficiency coefficient values and R values exceeded 0.6 and 0.9, respectively). The PYL-R experiment (PYL-R-E) model was designed as a contrast model without considering the impact of LAI. The simulated monthly runoff results obtained using the PYL-R and PYL-R-E models were compared, and the within-year changes in the differences between the two results were analysed to evaluate and quantify the impact of vegetation dynamic on runoff. From January to July, when LAI values increased by around 2.6 m² m⁻², monthly runoff depth differences between PYL-R and PYL-R-E results increased by 35.25, 27.98, and 29.14 mm in the Ganjiang, Xinjiang, and Fuhe watersheds, respectively. Dense vegetation caused high interception and evapotranspiration during summer, which largely reduced runoff. By contrast, during winter, the effect of vegetation was weaker on runoff process whereas the impacts of other factors (e.g., precipitation) were higher. The sensitivity of monthly runoff to vegetation dynamics varied greatly throughout the whole year. In particular, during August and September, the LAI-caused runoff changes were very high, accounting for 28–42% of monthly runoff in the sub-watersheds. Our findings clarify the effects of changes in vegetation on hydrological processes over short time scales, thereby providing insights into the effects of scale on eco-hydrological processes.     

Luquillo Experimental Forest: Catchment science in the montane tropics

Catchments in the Luquillo Experimental Forest (LEF) of Puerto Rico are warm, wet and tropical with steep elevational relief creating gradients in temperature and rainfall. Long-term objectives of research at the site are to understand how changing climate and disturbance regimes alter hydrological and biogeochemical processes in the montane tropics and to provide information critical for managing and conserving tropical forest ecosystems globally. Measurements of hydrology and meteorology span decades, and currently include temperature, humidity, precipitation, cloud base level, throughfall, groundwater table elevation and stream discharge. The chemistry of rain, throughfall, and streams is measured weekly and lysimeters and wells are sampled monthly to quarterly. Multiple data sets document the effects of major hurricanes including Hugo (1989), Georges (1998) and Maria (2017) on vegetation, biota and catchment biogeochemistry and provide some of the longest available records of biogeochemical fluxes in tropical forests. Here we present an overview of the findings and the data sets that have been generated from the LEF, highlighting their importance for understanding montane tropical watersheds in the context of disturbance and global environmental change.     

Effect of subsurface drainage on runoff and sediment yield from an agricultural watershed in western Oregon,U.S.A.

Rainfall, watershed runoff and suspended-sediment concentrations for three small watersheds (0.46, 1.4 and 6.0 ha in size) were measured continuously for four winter rainfall seasons. The watersheds were fall-planted to winter wheat and were located on the hilly western margins of the Willamette Valley, Oregon. Following two rainfall seasons of data collection, a subsurface drainage system (consisting of a patterned arrangement of 10-cm plastic tubing at a depth of 1.0 m and a spacing of 12 m) was installed on the 1.4-ha watershed (watershed 2).Perched water tables were lowered and seepage was reduced on watershed 2 following the installation of the drainage system. The reductions were quantified with a water-table index (cumulative integrated excess). Watershed runoff and sediment yield from watershed 2 were decreased by ～65 and ～55%, respectively. These reductions were estimated from double mass curves and by statistical regression on a set of hydrograph variables. Maximum flow and average flow rates were decreased and the time from the beginning of a storm to the peak flow (lag time) increased.It is concluded that subsurface drainage can be an effective management practice for erosion control in western Oregon.     

Comparison of two stream gauging systems for measuring runoff and sediment yield for a semi‐arid watershed

Our ability to understand erosion processes in semi‐arid ecosystems depends on establishing relationships between rainfall and runoff. This requires collection of extensive and accurate hydrologic and sediment data sets. A supercritical flume with a total load traversing slot sediment sampler used on several sites at the Walnut Gulch Experimental Watershed (WGEW) near Tombstone, AZ has proven to be a reliable way to measure flow and sediment discharge from small watersheds. However, it requires installation of a costly structure that is only suitable for relatively small flows. A more commonly used method based on ease of installation and expense is the pump sampler. One example of this is a set of instrumentation developed by the Australian Commonwealth Scientific and Industrial Research Organization (CSIRO), in which the pump sediment sampler is part of an in‐channel, fully automated system for measuring water velocity, depth, turbidity and collecting runoff samples. A 3.7 ha arid watershed at WGEW was instrumented with both systems and hydrologic and sediment data were collected and compared during a 2 year period. Total sediment yield for the entire period measured by the CSIRO pump sampler (11.6 t ha^‐1) was similar to that by traversing slot sampler (11.5 t ha^‐1). The pump sampler accurately estimated the amount of fine (< 0.5 mm) sediment fractions exported, but consistently underestimated the coarse (>0.5 mm) sediment fractions. Median sediment diameter of samples collected by traversing slot and pump sampler were 0.32 and 0.22 mm, respectively. This study outlines the benefits and limitations of the pump sampler based system for monitoring sediment concentration and yield in high‐energy headwater catchments, and makes recommendations for improvement of its performance. Copyright © 2012 John Wiley & Sons, Ltd.     

Hydrological transit times in nested urban and agricultural watersheds in the Greater Toronto Area,Canada

Watershed mean transit times (MTTs) are used to characterize the hydrology of watersheds. Watershed MTTs could have important implications for water quality, as relatively long MTTs imply lengthier water retention, thereby allowing more time for pollutant transformation and more moderate release of pollutants into streams. Although estimates of MTTs are common for undisturbed watersheds, only a few studies to date have applied MTT models to urbanized watersheds. In the present study, we use δ¹⁸O to compare estimates of MTTs for paired suburban‐industrial and agricultural watersheds in Toronto, Canada. Although differences in precipitation δ¹⁸O between the two watersheds were negligible, there were significant differences in stream δ¹⁸O, suggesting differences in water transport pathways. Less damping between input precipitation δ¹⁸O and output stream δ¹⁸O in the suburban‐industrial watershed indicated a larger streamflow contribution from quick‐flow transport pathways. We applied three transit time models to quantify MTTs. Considering overall model fit, root mean square error, and uncertainty in model parameters, the exponential model performed the best of the three models. Optimized MTTs using this distribution across the suburban‐industrial subwatersheds ranged from 2.1 to 2.9 months, whereas those in the agricultural subwatersheds ranged from 2.7 to 7.5 months. The relatively small difference between urban and agricultural MTTs coincides with observations elsewhere. Model efficiencies could potentially be improved, and MTTs estimated more reliably, with a higher sampling frequency that captures a greater volume of overall discharge. Overall, this work provides a distinct first glimpse into the separation of MTTs between paired watersheds with such a large contrast in their land use.