Challenges in Urban Hydrologic Modeling: A Baltimore Case Study
To effectively and sustainably manage water resources in urban areas we need to better understand the effects of urbanization on the hydrologic cycle and conversely, the ways that surface and groundwater quality and quantity can affect humans. We present work on coupled modeling groundwater and surface water in Dead Run, a small urban watershed in Baltimore, Maryland. We use ParFlow, which models three dimensional, variably saturated subsurface - surface flow. This poster reviews some of the challenges that have been encountered in modeling elevations and slopes for overland flow in the Dead Run case. Elevations from a Digital Elevation Model (DEM) in an urban landscape may not be able to adequately define surface flow paths as streams may have been moved, channelized, piped underground, or otherwise modified. Knowledge of the locations of streams may not be adequate, as the surface stream expressions may be disconnected due to piping in between. Here we present the beginnings of ways to deal with the challenges of modeling urban as well as future plans to incorporate urbanization within existing models. This work will be expanded to a regional hydrologic model, which will be coupled with an urban growth model of the Baltimore region to explore the predictions and feedbacks between the two models.
Interactions Between Flow Patterns and Infrastructure in an Urban Stream
Channel form and flow pattern in urban streams are often strongly influenced by the configuration of urban infrastructure, with potential consequences for flowpaths, residence times, and biogeochemical cycling. Although there is a common perception that urban channels are characterized by simplified geometry and loss of habitat complexity, in some cases culverts and other in-channel structures may add hydraulic complexity and lengthen residence times for water and associated constituents. In this study we examine the influence of in- channel structures on longitudinal profiles and configuration of pools and riffles in two urban streams reflecting different periods of historical development and different practices with respect to stormwater management and encroachment on the riparian corridor. The analysis involves using a 2-D depth-averaged finite difference hydraulic model to simulate travel and residence times and flow patterns over a range of stages along stream reaches in the Dead Run and Red Run tributaries of Gwynns Falls, in the main study watershed of the Baltimore Ecosystem Study. This research seeks to assess 1) varying sensitivity of flow patterns and residence times to the influence of in-channel structures with changes in flow stage, and 2) differences in the relative influence of infrastructure in watersheds with differing age of development and management practices.
Evaluating Effects of Land-use Change on Stream Hydrology and Water Quality in the Reedy River Watershed
Conversion of land cover from forested to urban is a major cause of nonpoint source pollution of the surface waters. During active land conversion/development, suspended sediment eroded from exposed soils often is the primary source of surface water degradation. Despite the use of Best Management Practices (BMPs), the overall protective goals of erosion prevention and sediment control regulations are not always achieved. The objective of this study is to understand and predict the impact of construction activities on stream hydrology and water quality and to assess the overall, collective effectiveness of BMPs that were implemented. Hydrologic, sediment, and nutrient data were collected from several streams that are tributaries of the Reedy River, South Carolina. These streams drain catchments (~1 km2) with varying degrees of active disturbance due to housing construction. Some were highly disturbed, others moderately disturbed, while others served as relatively undisturbed references. Stationary and handheld instruments that use Doppler technology were employed to measure the flowrate in these streams. Upon comparing measurements from these two types of instruments, it has been observed that the stationary instrument consistently over-estimated flowrates because it sampled only a small, faster moving portion of the stream. Therefore, one empirical and several physically based procedures (conveyance-slope method, and rectangular channel approximation method) were developed to correct the time series of flow made with the stationary instrument. Sediment and nutrient concentrations in the outflows from the disturbed catchments are compared with those of the undisturbed catchments to quantify the extent of disturbance. Peak flows from disturbed catchments were up to two orders of magnitude greater, and sediment yields were up to three orders of magnitude greater than those from the reference catchments. Several metal concentrations (e.g. Fe, Mn, Al, Mg, and K) follow a similar trend and are strongly correlated with the sediment. Using the formulation of the Universal Soil Loss Equation, the product of cover-management factor (C) and support practice factor (P) is calculated for each land-use type. Since all other factors in the USLE (R, K, L, and S) have same values across the catchments at any given time, the product (PC) quantifies the extent of disturbance in terms of sediment yield. Preliminary relative values of this product (PC) for agricultural land = 2.4, forest land = 1.0, fully developed urban land = 0.9, and construction site = 150.
Influence of Antecedent Moisture and Rainfall Rate on the Hydrologic Response of and Nitrate Leaching from Intact Soil Monoliths
The transfer of excess nitrate from agricultural soil during and following rainfall events is a major environmental problem in many regions worldwide. Although field and modelling studies have linked discharge and nutrient export from catchments to spatial variables such as soil type and agricultural management practices, less is known about temporal variability in nutrient export. Several field studies have shown that the hydrologic response of catchments to storm events differs with variable antecedent soil moisture conditions as well as storm type/rainfall rate. Variable hydrologic responses are in turn, expected to affect the export of nutrients such as nitrate from agricultural landscapes. This research uses an experimental approach to characterize the combined influence of antecedent soil moisture and rainfall rate on runoff quantity and quality from agricultural soil. Two sets of experiments were performed on large, intact soil monoliths in a laboratory setting, where soil monoliths were subject to three moisture treatments (dry, moist, wet/field capacity) and two different rainfall simulations (2.5mm/h and 20.5mm/h). Under high rainfall rates, nitrate export for the event ranged from 1125mg m-2 (wet soil) to 81.5mg m-2 (dry soil) and ranged from 1053mg m-2 (wet soil) to 429.1mg m-2 (dry soil) under low rainfall rates. Antecedent moisture conditions had a greater effect on the depth of runoff and the total mass of nitrate leached than rainfall rate, with wetter soils producing more runoff (runoff ratios = 0.37, 0.33, respectively for the 2.5mm/h and 20.5mm/h events) and causing more nitrate export (1125 and 1053 mg m-2 event-1) than dry soil (runoff ratios = 0.033 and 0.105; nitrate export = 89.2, 429.1 mg m-2 event-1). The soil moisture response and the progression of the wetting front occurred earliest in wet soil and latest in dry soil, as illustrated by runoff volumes and soil moisture profiles recorded throughout both rainfall simulations. Patterns differed between high rainfall rates and low rainfall rates in terms of the flow- weighted export of nitate. Following high rates of rainfall, wet soil produced higher flow-weighted nitrate export (128.5mg m-2 mm-1), whereas dry soil produced higher flow-weighted nitrate (155.6mg m-2 mm-1) export at low rainfall rates. Moisture profiles showed a clear contrast in the progression of the wetting front between high and low rainfall rates. Matrix flow, utilizing a wide array of pore sizes, dominated during low rainfall rates at all three moisture levels and high rainfall rates with dry soil, whereas preferential transport appeared to dominate during high rainfall rates. Wet soils under high rainfall rates, which produced the most preferential flow, produced both the highest total mass of nitrate export and the highest flow-weighted nitrate export. In general, antecedent soil moisture had a greater effect on the total mass of nitrate leached than rainfall rates. The findings of these experiments will increase the understanding of the relationship between antecedent soil moisture, rainfall rates and leaching behaviour in agricultural soils and will help guide best management practices in the management of soil nutrients for agriculture.
Storm to Seasonal Nitrate Flushing and Relationship to Hydroclimatic Conditions
Coupled hydrological and geochemical cycles have historically been used to understand the mechanisms controlling chemical export at the watershed scale. Numerous studies have focused on stream geochemistry dynamics and particularly on the effects of nitrogen deposition on catchments and ecosystems. However, few studies have evaluated the variability in nitrate behavior in watersheds with different climate and hydrologic characteristics. Furthermore, storm nitrate dynamics (flushing) and corresponding relationship to discharge patterns, precipitation intensity, atmospheric deposition, and watershed properties is an area with few published studies. In the current study, we utilize hydrologic, geochemical and atmospheric data to better understand the primary processes controlling nitrate dynamics at monthly and seasonal time scales in ten watersheds located in the U.S. Stream data (discharge and nitrate concentrations) was obtained from the United States Geological Survey (USGS). Atmospheric deposition data was gathered from regional National Atmospheric Deposition Program (NADP) sites. Collected data were assessed to evaluate nitrate dynamics for the period 2000-2006. Intra-storm nitrate variability was also evaluated in a watershed in Southern California (Arroyo Seco) where estimates of dry deposition are considered some of the highest in North America. Results from our analyses show that nitrate concentration appears to respond differently in the studied watersheds, exhibiting dilution effects in some watersheds and concentration effects in others. Discharge and nitrate concentrations during the 2009 winter reveal that the Arroyo Seco watershed is characterized by distinctive flushing of nitrates within storms. The role of specific hydrological events (size, intensity and duration) on the nitrate mobilization proved to be important as well as the accumulated nitrate pool available for mobilization. In general, results show a strong linkage between hydrological and biogeochemical controls of nitrate exports. Results from this study support previous studies regarding nitrate flushing during wet seasons and also add new insight on enhanced export in areas with increased deposition rates of nitrates.
Nitrogen Dynamics Along a Headwater Stream Draining a Fen, Swamp, and Marsh in a Fractured Dolomite Watershed
Stream-wetland interaction has been shown to have a significant effect on nutrient cycling and downstream water quality. Additionally, connection to regional groundwater systems can dilute or enrich stream water with a number of dissolved constituents. This study demonstrates the resultant downstream change in dissolved nitrogen species as a hardwater stream emerges from a calcareous aquifer and traverses a calcareous fen, a cedar swamp, and a cattail marsh over two growing seasons, a very dry 2006 and a very wet 2007. Upon emergence at a number of groundwater seeps, the water contained appreciable nitrate levels averaging 2.72±0.42 mg NO3-N L-1, minimal organic nitrogen, and ammonium below detectable levels. Through the gently sloping calcareous fen, with a stream residence time of ~ 5 hours, NO3-N concentration decreases of 0.35 mg L-1 were observed. Concomitantly, stream recharge into the dolomite bedrock depressed stream discharge values significantly, further removing nitrate from the stream system. This resulted in the fen-bedrock system acting as an estimated net sink of 432 kg of NO3-N in the early summer of 2007, for example. In contrast, the hydrological-biogeochemical systems became decoupled through the swamp during the same period, where concentrations increased from 2.58±0.34 mg L-1 entering the swamp to 2.65±0.58 mg L-1 exiting, but streamflow decreased in general by 5 L s- 1. This resulted in the swamp, with its large depression storage, acting as a small net sink of nitrate (75 kg through the early summer), which would not be detected simply from concentration changes. The concentration-discharge relation realigned through the marsh, where significant groundwater entered the wetland, increasing both concentration and discharge, yielding a small export of 93 kg over the same time period. A series of tracer injections in each wetland type will be presented to compare the streamflow- concentration patterns with the measured nutrient spiralling. This research highlights the complexity of surface-groundwater interactions and multiple landscape elements on the resultant downstream fate and transport of nutrients.
Rain-Induced Bursts Of Nitrous Oxide May Account For Differences In Dissolved Nitrogen Export From Forested Catchments
Despite nearly 30 years of research, we are unable to account for differences in dissolved nitrogen (N) export among catchments in the sugar maple forest of the Turkey Lakes Watershed. Neighboring catchments with similar N inputs show major discrepancies in dissolved N (nitrate + ammonium + dissolved organic nitrogen) export. In this study, we hypothesized that gaseous N export from wetland soils accounts for this discrepancy. To test this hypothesis, soil nitrous oxide (N2O) efflux was measured during the snow free season (May 1 to October 30) in 2006, 2007, and 2008. Minimal N2O efflux (<1 g N/ha/day) was observed on days without rain. However, on days with rain, soil N2O efflux was significant from wetland area soils, with a linear increase of 0.016 g N/ha/day per millimeter of rain (r2 = 0.60, p<0.001); N2O efflux from upland soils was not significant. Process based monitoring of the wetland soil profile suggests that rain delivers water to the surface layers of the wetlands creating an oxygen poor environment where accumulated nitrate is first transformed to N2O and then to dinitrogen (N2). We could not measure N2. However, if we assumed a N2:N2O ratio of 10:1 from the literature, the discrepancy in dissolved N export among the catchments could be explained. Our findings suggest that rain can produce substantial bursts of N2O and N2 from forest soils and that failure to account for gaseous N export may lead to an underestimation of N loss from forested catchments.
Shopping for Hydrologically Relevant Connectivity Metrics in Humid Temperate Forested Systems
Connectivity is often said to be crucial for hydrologic prediction as it relates to the functional connectedness between catchment elements. If connectivity is to serve as an effective diagnostic classification tool of hydrological behaviour, then it clearly matters (1) how it is measured, and (2) whether the spatial metrics are strongly correlated not only to catchment-scale antecedent moisture conditions but also to streamflow discharges in a given environment. Previous studies have advocated that connectivity in shallow soil moisture patterns induces threshold-like changes in runoff in semi-arid rangeland catchments but not in temperate humid forested catchments. We argue that in the latter environments, capturing critical spatial organization in soil moisture patterns depends on the way the chosen connectivity metric is built. We therefore tested several 2-D and 3-D connectivity measures (i.e. entropy, effective upslope area, integral correlation and connectivity scale lengths, source-to-stream connectivity, flow path connectivity) in a temperate humid forested catchment (Laurentians, Canada). Computations were based on continuous soil moisture patterns collected on 16 occasions at soil depths of 5, 15, 30 and 45 cm and then transformed into indicator patterns using either time- variable or time-invariant thresholds. Assessments of connectivity were variable depending on the computed metric, as just a few measures were significantly correlated with both antecedent moisture conditions and catchment discharges. Antecedent moisture conditions were strongly associated to high connectivity between source areas having a 40% or higher soil moisture volumetric content. Also, significant discharges at the outlet were generated when physically connected source areas had a soil water volumetric content greater than 30%. Topography-based connectivity metrics reflected changes in catchment macrostate and stormflow response better than omnidirectional methods, as well as source-to-stream connectivity metrics were more hydrologically relevant than metrics that did not consider the stream channel. These conclusions stress the importance of shopping for the right connectivity metric for hydrologic prediction, especially in humid forested environments that exhibit much larger variability in soil hydrologic properties than semi-arid rangeland catchments.