ANTHROPOGENIC IMPACTS ON THE WATER AND ENERGY BALANCE OF AN URBAN
SEMI-ARID ENVIRONMENT
A thesis submitted to the Faculty and the Board of Trustees of the Colorado School of Mines in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Civil and Environmental Engineering). Golden, Colorado Date _______________________ Signed: _______________________ Bryant Reyes Signed: _______________________ Dr. Terri S. Hogue Thesis Advisor Signed: _______________________ Dr. Reed M. Maxwell Thesis Co-Advisor Golden, Colorado Date _______________________ Signed: _______________________ Dr. Terri S. Hogue Professor and Head Department of Civil and Environmental Engineering
ABSTRACT
As the world rapidly urbanizes, a grasp of water resources within an urban context becomes crucial to both the policy and scientific communities. Yet many of the operational and management models used to inform these policies lack vital parameterizations needed to simulate the hydrologic system holistically. Anthropogenic processes and changes to the
hydrologic cycle caused by urbanization (land use change, denser building patterns, increases in imported water and water use, increased surface imperviousness, increased subsurface
infrastructure, etc.) are known to have significant and interacting impacts on the hydrologic system as a whole; only recently has the hydrologic community been able to quantify these effects and understand their behavior. The work presented here assesses the processes simulated by an integrated, coupled land surface and hydrologic model at various spatial scales in the urban domain. Throughout this work data pertaining to Ballona Creek watershed in Los Angeles, California is used in both model building and analysis. The watershed contains highly urbanized and diverse portions of the cities of Santa Monica and Los Angeles, along with more natural land surfaces in the northern portions of the watershed, with a wide range of urban land cover and land use scenarios in a semi-arid environment. We begin this work by utilizing two land cover datasets for the City of Los Angeles: (1) the National Land Cover Database (NLCD) dataset at a 30-m resolution; and (2) an ultra-high-resolution dataset at a 0.6-m resolution. Various
permutations and resolutions of the model are simulated for a spin-up and two-year study period. The impact of the highly organized, yet heterogeneous, land cover typical of the urban domain is shown to impact the runoff/runon process characteristics of these domains, creating variations in
overland flow and evapotranspiration (ET). Spatial scaling land surface and hydrologic parameters creates systematic diurnal biases in the surface energy budget in contrast to the seasonal biases causes by lateral flow processes. In addition to creating land surface parameters for the widely used NLCD urban land covers, this work illustrates nonlinear issues of scale and resolution and improves understanding of how these processes affect the surface energy and hydrologic budgets. Next, remotely sensed observations of land surface temperature and land cover are paired with domestic water use data to assess the direct impact of outdoor water use. We find a decrease of up to 3.2±0.02 Kelvin between low and high irrigation areas of similar land cover; simulations are able to capture this difference but underestimate absolute values throughout. Model simulations show that irrigation timing has a small impact on ET and runoff and that relatively low irrigation volumes push the semi-arid urban environment of Ballona Creek into a sub-humid regime. Finally, we utilize a range of land surface and hydrologic models applied at a high spatial resolution (1-km) to quantify some of the deficiencies seen in the
simulation of semi-arid urban environments and to provide a framework for future work in the field.
TABLE OF CONTENTS
ABSTRACT ... iii
LIST OF FIGURES ... ix
LIST OF TABLES ... xiii
ACKNOWLEDGMENTS ... xiv
DEDICATION ...xv
CHAPTER 1 MOTIVATION AND LITERATURE REVIEW ...1
Motivation ...1
Motivating Scientific Questions ...2
1.2.1 Structure of Dissertation ...4
Study Domain ...4
Introduction to Urban Hydrology ...5
1.4.1 Scope of Literature Review ...7
The Urban Hydrologic Flow Regime ...8
Groundwater in the City ...12
1.6.1 Groundwater Recharge Sources ...13
1.6.2 Groundwater-Surface Water Interactions ...15
1.6.3 Water Table Change ...16
1.6.4 Groundwater Quality ...17
1.6.5 Importance as a Resource ...17
Recharge Estimation Methods ...18
1.7.1 Water Balance Method ...19
1.7.4 Direct Measurement ...22
1.7.5 Indirect Measurement ...22
1.7.6 Theoretical Methods ...23
1.7.7 Watershed Models ...24
1.7.8 Groundwater Models ...25
1.7.9 Land Surface Models ...26
CHAPTER 2 IMPACT OF LATERAL FLOW AND SPATIAL SCALING ON THE SIMULATION OF SEMI-ARID URBAN LAND SURFACES IN AN INTEGRATED HYDROLOGIC AND LAND SURFACE MODEL ...27
Introduction and Background ...28
Methodology ...32
2.2.1 Model Description ...32
2.2.2 Land Cover ...35
2.2.3 Vegetation and Soil Parameters ...37
2.2.4 Meteorological Forcings ...39
2.2.5 Simulations ...40
Results and Discussion ...43
2.3.1 Increasing Spatial Discretization ...43
2.3.2 Land Surface Composition and Configuration ...45
2.3.3 Scaled Parameter Performance ...49
Conclusion and Implications ...55
CHAPTER 3 URBAN IRRIGATION SUPPRESSES LAND SURFACE TEMPERATURE AND CHANGES THE HYDROLOGIC REGIME IN SEMI-ARID REGIONS ...58
Introduction ...59
Methods...62
3.2.2 Landsat Land Surface Temperature ...63
3.2.3 Outdoor Water Use Data and LST Classification ...65
3.2.4 Land Cover, Irrigation, and Model Used ...65
3.2.5 Model Forcings ...67
Results ...68
3.3.1 Irrigation Impact on LST ...68
3.3.2 Simulated LST Change ...72
3.3.3 Sensitivities of Hydrologic Fluxes and Regime Change ...75
Discussion and Conclusions ...78
CHAPTER 4 URBAN FEATURES IN THE MODELING OF A SEMI-ARID URBAN ENVIRONMENT: BALLONA CREEK WATERSHED, LOS ANGELES, CALIFORNIA...81
Introduction ...82
Study Area and Methodology ...85
4.2.1 Models...86
4.2.2 Study Area ...87
4.2.3 Forcing Data ...88
4.2.4 Land Cover and Elevation ...90
4.2.5 Soil ...91
4.2.6 Comparison Datasets ...94
Results and Discussion ...96
4.3.1 Baseline Noah LSM and PF.CLM ...97
4.3.2 Impact to Runoff ...98
4.3.3 Impact to ET ...100
CHAPTER 5 CONCLUSIONS AND IMPLICATIONS ...110
Spatial Scale and Heterogeneity ...110
Urban Irrigation ...111
Urban Watershed Modeling ...113
Future Work: Urban Hyper-Resolution Models ...114
REFERENCES CITED ...115
LIST OF FIGURES
Figure 1.1 Conceptualization of the urban hydrologic and land surface features and fluxes
addressed in this work ... 2 Figure 1.2 Estimated potential increases in deep infiltration caused by urbanization in
various cities. Methods for estimating values for various cities vary by study. [Modified from figures in: Foster et al., 1999; Garcia-Fresca and Sharp, 2005;
Sharp, 2010] ... 7 Figure 1.3 Urban hydrologic processes with the urban domain set as the primary
boundary and interactions between surface water, groundwater, water users,
and infrastructure. ... 9 Figure 2.1 Land Cover datasets used in this study for Ballona Creek Watershed, Los
Angeles, CA: (a) NLCD 2001 30-m land cover dataset; (b) McPherson et al. 0.6-m land cover dataset; (c) NLCD and McPherson land cover data overlain over a satellite image of a neighborhood within the watershed; (d) Satellite and McPherson land cover data for a typical 30 m by 30 m “Medium Intensity
Developed” pixel ... 38 Figure 2.2 Representations of Study Simulations: (a) All simulations are 30 m by 30 m
with 15 layers and initialized groundwater at layer 13; (b) seven resolutions (1×1 up to 50×50) of a homogenous land cover are tested; (c) the
heterogeneous simulations are apportioned into pervious (bare soil, grass, and tree) and impervious (concrete) sections according to values in Table 2.1 for the four urban land covers and are modeled in three different slope directions (impervious to pervious, pervious to impervious, and parallel to both) to
capture the variability within a 30 m by 30 m urban region ... 41 Figure 2.3 Total yearly (top) overland flow [mm yr-1] and (bottom) evapotranspiration
[mm yr-1] fluxes for a homogeneous land surface at 7 spatial resolutions ... 45
Figure 2.4 Total yearly hydrologic budget (evapotranspiration, overland flow, and soil storage) for simulations of heterogeneous (50×50 pixel, 0.6-m resolution) land cover at 3 configurations (I→P, P→I, I↓P) and effective/scaled (1×1
pixel, 30-m resolution) parameterization ... 47 Figure 2.5 Comparisons of latent heat flux for every timestep between the three
configurations and the average of the heterogeneous configurations (50×50) vs. the scaled/effective parameter simulations (Eff.) for high, medium, and
low intensity developed and developed open space. Colors show the season in
which the time step occurred. ... 50 Figure 2.6 Same as Figure 2.5 for the sensible heat flux of high intensity developed and
developed open space ... 51 Figure 2.7 Same as Figure 2.5 for the ground heat flux of high intensity developed and
developed open space ... 51 Figure 2.8 Mean diurnal surface energy fluxes of effective/scaled (solid lines) and the
average of the heterogeneous configuration (dashed lines) simulations for the
four urban land cover types. ... 53 Figure 3.1 Study domain and LST classification methodology. (a) Estimated outdoor
water use data for a typical month (April 2005), (b) high and low water use areas overlying the NLCD 2001 land cover (Developed, High Intensity; Developed; Medium Intensity; Developed; Low Intensity; Developed; Open Space; “Natural” refers to all other land cover), (c) medium intensity
developed LST pixels from the April 21, 2005 Landsat overpass within high
and low water use areas. ... 63 Figure 3.2 Box plots of monthly LST pixels aggregated across years for (a) Developed,
Open Space, (b) Developed, Low Intensity, (c) Developed, Medium Intensity, and (d) Developed, High Intensity. The box represents the middle 50% of data, line and circle within the box are the median and mean respectively, the whiskers encompass 96% of the data, and the points are outliers. Mean difference between urban land cover in low (dark color) and high (light color) water use areas for pre-water use restriction years (1/2000 through 12/2007) is shown above the box plot for each month. All mean differences are significant
(p < 0.01) except for (*) ... 70 Figure 3.3 Box plots for low and high water use areas of low intensity LC. Similar to
Figure 3.2 for a (top) dry year and (bottom) wet year. During the spring and summer months of the dry year, irrigations impact on the LST is greater than during the wet year; all mean differences are significant (p < 0.01) except for
“*”. See Table 3.2 for mean differences for all land cover types. ... 71 Figure 3.4 Comparison of simulated and observed Landsat LST for all overpass times.
Horizontal bars represent the standard deviation of values observed for that scene; vertical bars represent the full range of model results from all simulations. 1:1 and mean bias lines shown as dark and light grey dashed
Figure 3.5 Mean LST differences between low and high irrigation in Landsat LST vs. modeled simulations. These are calculated using the low and high irrigation areas and simulation scenarios aggregated through the study period by month similar to Figure 3.2 and Figure 3.3. The r2 values along the 1:1 line are
shown with and without spring months (3, 4, 5) included in the calculation. ... 76 Figure 3.6 Study period mean simulated evapotranspiration and runoff. All simulated
irrigation timings and volumes for the study period are shown. The shading represents the range of timing, which is small throughout with the largest
range seen for the open space LC at +300% simulation. ... 77 Figure 3.7 Budyko and water-energy plots for simulations. Full study period mean
modified (a) aridity index vs. evaporative index and (b) evaporative efficiency vs. evaporative index for each irrigation volume (mean of all irrigation
timings). Irrigation volume is represented by size of symbol and the “mean” high water use area volumes (middle 50% of high irrigation rates for domain and time period) are shown in (a). Black lines in (a) are the theoretical water and energy limits and the dotted blue lines are transition points in aridity
(Arora, 2002; Ponce et al., 2000). ... 77 Figure 4.1 (a) USGS 30-m National Land Cover Database land cover type, the (b)
derived 1-km resolution land cover used in the PF.CLM simulations, and the
(c) UMD land cover dataset used for the Noah LSM. ... 87 Figure 4.2 Soil data used for (a–d) PF.CLM and (e) Noah LSM simulations. (a) and (c)
show the 30-m data from which the 1-km soil in (b) and (d) are derived. For PF.CLM, the top layer uses unique soil properties for the urban land cover types and the soil in shown in (b) for natural land cover pixels. The 100-m bottom layer uses the data shown in (d) obtained from the Gleeson et al. (2011) dataset. (e) The CONUS-SOIL (Miller and White, 1998) dataset is
used for all four layers of the Noah model. ... 93 Figure 4.3 Comparison of Ballona Creek watershed average domain LST from the
MODIS MOD11 and MYD11 products and Landsat 5 and 7 30-m LST for the
88 overlapping scenes ... 95 Figure 4.4 Monthly total runoff (mm/month) for Jan. 2000 through Sep. 2010 as observed
from the LACDPW Ballona Creek Gauge and estimated by four model scenarios. The Nash-Sutcliffe model efficiency coefficient is shown in
brackets for each simulation. ... 99 Figure 4.5 Average monthly total evapotranspiration for the full simulation period (Jan.
intensity developed, (d) medium intensity developed, and (e) high intensity
developed land cover types. ... 101 Figure 4.6 Average seasonal diurnal cycles for the full simulation period (Jan. 1, 2000–
Dec. 31, 2010) of four model simulations for the (a) full study area, (b) developed open space, (c) low intensity developed, (d) medium intensity
developed, and (e) high intensity developed land cover types. ... 104 Figure 4.7 Comparison of remotely sensed (MODIS and Landsat) LST to modeled LST
for (a) full study domain, (b) developed open space, (c) low intensity
developed, (d) medium intensity developed, and (e) high intensity developed urban land cover types. Two distinct clusters of points appear in the figures
LIST OF TABLES
Table 2.1 Average composition of NLCD Urban Land Cover Types in terms of the
McPherson Land Cover Dataset ... 37 Table 2.2 Land surface parameters used in the PF.CLM simulations. Parameters for the
NLCD urban types were created using a weighted mean average method using
GLDAS1 parameters and the values found in Table 2.1. ... 39 Table 2.3 Soil parameters used in the PF.CLM simulations. Soil parameters are
calculated for the NLCD top layer using the percent impervious and values for
sandy loam (Schaap and Leij, 1998) and concrete (Schneider et al., 2012) ... 40 Table 3.1 Total yearly irrigation and irrigation plus precipitation rates for all simulations.
Irrigation is applied for 1 hour every other day at the specified hour of the
simulation (00, 06, 12, or 18 local time). ... 67 Table 3.2 Mean differences between high and low irrigation rates for dry (WY2002) and
wet (WY2005) years. *Mean difference not statistically significant (p > 0.01) ... 73 Table 4.1 Irrigation rates for simulations by land cover type and PF.CLM scenario. The
irrigation is applied for 1-hour every other day at 6PM local time ... 88 Table 4.2 Descriptive statistics of meteorological forcings for both Noah LSM and
PF.CLM simulations. “Precipitation NLDAS” is derived from NLDAS-2 and used for the Noah LSM simulations and “Precipitation Local” is derived from
local stations and is used for PF.CLM simulations. ... 89 Table 4.3 Urban land cover land surface parameters used for PF.CLM simulations. For
the Noah LSM we use the standard land surface parameters provided by the
LIS modeling framework. ... 91 Table 4.4 Soil parameters used in PF.CLM simulations. Soil parameters are calculated for
the NLCD top layer using values for soil found in Schaap and Leij (1998) and Schneider et al. (2012). S01, S09, S12, S19, and S22 are fully-saturated layers
and are obtained through the dataset described in Gleeson et al. (2011). ... 92 Table 4.5 Change in monthly total ET, monthly total runoff, and diurnal average LST
over the 11-year study period due to changing emissivity (𝜀), soil thermal conductivity (λsoil), and hydraulic conductivity (K) for urban land cover in
ACKNOWLEDGMENTS
I wish to acknowledge the many people and institutions that have allowed me to pursue this degree:
The Eugene V. Cota-Robles Fellowship at the University of California, Los Angeles; the National Science Foundation (NSF) Graduate Research Fellowship (DGE-1057607); NSF Water Sustainability and Climate Grants (EAR-12040235, EAR-1204787); NSF Engineering Research Center for Reinventing the Nation’s Urban Water Infrastructure (EEC-1028968); National
Aeronautics and Space Administration (NASA) and the University of Virginia Intensive Summer School for Computing in Environmental Sciences; and high-performance computing support from NASA Center for Climate Simulation, the Colorado School of Mines Golden Energy Computing Organization, and Yellowstone (ark:/85065/d7wd3xhc) provided by NCAR's Computational and Information Systems Laboratory sponsored by the NSF.
My committee members who have provided insights and their valuable time to improving this work: Dr. Amanda S. Hering, Dr. John E. McCray, and Dr. Christa D. Peters-Lidard. Fellow graduate researchers who have shared ideas, grievances, and solutions throughout the years: Tristan Acob, Dr. Jennifer Jefferson, Dr. Alicia Kinoshita, Dr. Sonya Lopez, Dr. Kim Manago, Paul Micheletty, Dr. Caroline Mini, and Dr. Pouya Vahmani. My mother, father, brother, sister, and extended family who have kept me sane.
Finally, I would like to acknowledge my advisors, Dr. Terri S. Hogue and Dr. Reed M. Maxwell, for believing in me and my work.
DEDICATION
CHAPTER 1
MOTIVATION AND LITERATURE REVIEW
Motivation
Currently, the United Nations estimates that over 50% of the world’s population lives in urbanized areas, whereas this figure is higher in the United States (82%) and other developed nations (United Nations, 2011). However, by 2050 these proportions are projected to increase to 90% in the US and nearly 70% globally, with the largest increases occurring in developing countries and semi-arid regions. Our ability to provide resources to nearly any point in the world, regardless of natural availability, is partly attributable for this urbanization. In many cities, especially those located in semi-arid/arid regions, importing water from more plentiful and pristine locations is common. The combined impacts of increased urbanization and climate change will likely threaten the sustainability of such water management schemes. Increasingly however, because the climate attenuating nature of groundwater as a source of potable water (Green et al., 2011), semi-arid/arid cities have committed to efficiently increasing their use of local groundwater resources to combat possible shortages (e.g. LADWP, 2010). Given these pressures on water sustainability, a grasp of the water resources available within an urban context becomes crucial to both the policy and scientific communities. Accurate models of the hydrologic environment are the essential tool in assessing and projecting water resource availability.
The literature on the pathways that water travels within urban watersheds, the processes that affect it, and how they change have been summarized by Fletcher et al. (2013) and Schirmer et al. (2013). The various fluxes that are introduced and changed due to urbanization as well as
the general impact of that change are shown in Figure 1.1. For instance, the groundwater component of the hydrologic cycle goes through significant change due to urbanization and many fluxes are introduced that are absent in natural environments. It is important to note that the changes brought on by urbanization not only include physical alterations (increased surface imperviousness, channelized flow, increased sub-surface infrastructure etc.) but also changes to the water cycle due to direct human interactions (increased use of imported water, landscape irrigation, industrial water use, etc.). Much of the work completed in the literature has been in the quantification of these fluxes in a variety of cities, using a variety of methods, and considering various water use and climate scenarios. However, much of this work fails to provide methods that can be applied to a variety of urban areas, at different spatial and temporal scales, for water resource forecasting and policy making. There are several noted deficiencies in review of the previous work, including: 1) few studies have been conducted in semi-arid/arid cities, and 2) models are lacking key sources and sinks of water in highly developed urban areas.
Figure 1.1 Conceptualization of the urban hydrologic and land surface features and fluxes addressed in this work
Motivating Scientific Questions
processes (evapotranspiration, vegetation dynamics, energy budget cycling, etc.). Coupled models are most useful for understanding complex water resource problems because they consider feedback processes that affect the timing and rates of flux in the water and energy cycles and can account for processes that occur at the land surface(Singh and Woolhiser, 2002). We utilize such a model for this work (PF.CLM; Ashby and Falgout, 1996; Dai et al., 2003; Jones and Woodward, 2001; Kollet and Maxwell, 2006; Maxwell, 2013; Zeng et al., 2002) to allow us to obtain robust results for our simulations.
The research presented in this dissertation focuses on the improvement and understanding of the simulation of anthropogenic effects on evapotranspiration (ET), runoff, groundwater recharge, land surface temperature (LST), and other land surface fluxes in a highly developed, semi-arid region. Specifically, we seek to probe the following questions:
1. How can we capture the heterogeneity inherent in the urban land surface in
hydrologic and land surface models? Are the impacts of spatial resolution relevant at urban scales, and what are the impacts of scaling parameters to the accurate
simulation of relevant hydrologic and land surface fluxes?
2. Are the impacts of domestic and municipal outdoor water-use substantial enough, in an urban semi-arid environment, to be detectable at similar scales to our models? What are the characteristic impacts of this water-use flux on how the hydrologic and energy fluxes produced by different urban land cover types?
3. Does including outdoor water use and heterogeneous urban land cover in our models improve the accuracy of the estimated water and energy fluxes? Does the inclusion of this process provide new insights into the unique urban semi-arid hydrologic
1.2.1 Structure of Dissertation
First, the related issues of spatial scale, heterogeneity, land cover, and lateral flow in the modeling of semi-arid urban environments are discussed in Chapter 2. Next, Chapter 3 discusses a novel approach to the analysis of remotely sensed data to assess the impacts of the urban irrigation flux, which is prominent in semi-arid cities, on land surface and hydrologic processes. Finally, Chapter 4 discusses the work to ascertain the impacts of the urban irrigation flux on watershed level through the use of a full groundwater-surface water model. We undertake all of our analyses utilizing meteorological inputs and data from Ballona Creek watershed, which contains highly urbanized and diverse portions of the cities of Santa Monica and Los Angeles, California along with more natural land cover in the northern portions of the watershed in the Santa Monica Mountains. Los Angeles and the surrounding metropolitan area rely on extensive water importation as well as local groundwater resources (an average of 10-20% of yearly total potable water use; LADWP, 2010) for potable water and, as a result, serves as a good test bed for many scientific questions of interest. Chapter 5 provides overall conclusions and descriptions of the results from each of the previous chapters.
Study Domain
The history of the City of Los Angeles from a small pueblo on the banks of the Los Angeles River to the second largest metropolis in the United States is one that is tightly intertwined with its access (or lack thereof) to water. The creation of the California and Los Angeles Aqueducts to transport water from the Sierra Nevada Mountains and the Owens Valley, respectively, as well as water from the Colorado River Aqueduct allowed Los Angeles to
correctly takes into account this importation of water has been an elusive goal. One aspect in particular that is often overlooked is the possible groundwater recharge that occurs as a result of the use of this imported water. As this imported water travels through the city’s distribution system, when lawns all over Los Angeles are watered, there is potential for imported water to recharge the local groundwater system. Quantifying, and possibly modeling, these phenomena are important for understanding the water resources the city possesses and how future changes in water use patterns may impact this resource. The importance of groundwater as a resource for metropolitan Los Angeles calls for a greater understanding of its true state. The underlying groundwater basins in Los Angeles are the Central Basin, Santa Monica Basin, West Coast Basin, Hollywood Basin, and San Fernando Basin. More information is provided within each of the chapters regarding the study domain.
Introduction to Urban Hydrology
The modern field of urban hydrology began in earnest in the late 19th and early 20th
century (Delleur, 2003) as cities began to rapidly expand and the need for urban water
infrastructure projects became essential. However, McPherson (1979) places the rise of the term “urban hydrology” to about the late 1950’s, nearly half a century after the U.S. became a
majority urban nation. The field was focused at the time on the management of storm water and flood mitigation. For this reason, the impact of urbanization on rainfall-runoff and surface water processes and quality (Brabec et al., 2002; Field and Cibik, 1980; Galster et al., 2006; Jacobson, 2011; Paul and Meyer, 2001; Shuster et al., 2005) are generally better understood than the effects of urbanization on other hydrologic processes. Urbanization has significant effects on all
the years to include subjects such as surface water-groundwater interactions (Sophocleous, 2002) and land-surface atmosphere interactions (Collier, 2006; Kanda, 2007), among others.
Delleur (2003) concludes, after chronicling the evolution of urban hydrology from the drainage of rainwater and wastewater in ancient civilizations to the emergence of hydrologic modeling in the 1960s, that urban water problems “can no longer be evaluated in isolation but will have to be looked at in an integrated way.” Reviews of urban groundwater, and the urban potential recharge flux specifically, have been conducted as early as Lerner (1986). Subsequent examinations of the flux (Lerner, 2002, 1990; Schirmer et al., 2013; Sharp, 2010; Sharp et al., 2003; Vázquez-Suñé et al., 2005) have advanced the field and agreed on various points of contention. One of these points is the finding that the recharge flux is generally increased from natural conditions in nearly all currently studied urbanized areas when accounting for
uncertainty, regardless of the prevailing hydro-climatic regime of the area (Figure 1.2). Note that these findings are estimates for citywide values; recharge decreases in many local areas,
increases in others, and new recharge pathways are formed due to urbanization (Bhaskar et al., 2016a, 2016b).
Understanding the groundwater system and associated recharge flux is vital to predicting various important natural processes. Climate change has the potential to cause large changes in groundwater recharge rates around the world through changing precipitation and temperature (Dragoni and Sukhija, 2008; Green et al., 2011; Herrera-Pantoja and Hiscock, 2008; Liu, 2011; Thampi and Raneesh, 2012). Concurrently, feedbacks between the groundwater system and land surface processes have been shown to significantly alter energy fluxes (Ferguson and Maxwell, 2010), further complicating the effects of climate change on regional hydrology. In addition to these hydrologic considerations, groundwater is one of the most valuable water resources
available to man (Cutter, 2007), and recharge estimates are essential to groundwater resource assessment and management. Conjunctive use, especially in semi-arid cities, will be vital in combating expected future water shortages (Bray et al., 2007; Foster, 2001).
Figure 1.2 Estimated potential increases in deep infiltration caused by urbanization in various cities. Methods for estimating values for various cities vary by study. [Modified from figures in: Foster et al., 1999; Garcia-Fresca and Sharp, 2005; Sharp, 2010]
1.4.1 Scope of Literature Review
The following literature review focuses on describing the urban hydrologic cycle and the various methods available to study the groundwater recharge process within the urban domain. First, Section 1.5 presents a brief overview of the urban hydrologic flow regime and the relevant fluxes. Section 1.6 is a more detailed examination of the groundwater component of the urban hydrologic flow regime detailing the deleterious effects of urbanization, the importance of
groundwater in the city, and the potential sources and pathways of groundwater recharge.
Section 1.7 provides a brief overview of the non-hydrologic modeling techniques used in various case studies for recharge estimation. Hydrologic models and modeling methods that have been used to study urban groundwater are discussed in Sections 1.7.7 through 1.7.9.
The Urban Hydrologic Flow Regime
In 1942 O. E. Meinzer, often called the “father of modern groundwater hydrology,” stated, "The central concept in the science of hydrology is the so-called hydrologic cycle—a convenient term to denote the circulation of the water from the sea, through the atmosphere, to the land; and thence, with numerous delays, back to the sea by overland and subterranean
routes." (Meinzer, 1942) Much in the same way then, the central concept in the science of urban hydrology is the so-called urban hydrologic cycle—defined similarly to the previous term but including the effects of urbanization. Linton (2008) provides a detailed history of the
development of the hydrologic cycle and the various changes the standard illustration and concept have gone through. Linton takes the unique point-of-view of the hydrologic cycle as invention by 19th and early-20th century hydrologists and engineers. The current form of the hydrologic cycle can be traced back to the work conducted by them and has since been
established as suitable on a global scale. However, in many places around the world where the standard model has been greatly altered by both natural and anthropogenic processes, the
standard concept of the hydrologic cycle is inadequate to fully understand the hydrologic regime at the regional scale.
There are many challenges in conceptualizing the urban hydrologic cycle. Since every city is within a unique hydro-climatic system (humid, semi-humid, semi-arid, arid) and possesses differences in infrastructure systems, geographic locations, geologic setting, and
socio-demographics, no one hydrologic cycle illustration will be correct for every region. Another challenge of conceptualizing the urban hydrologic cycle is identifying and choosing the appropriate water boundary. Instead of creating the traditional hydrologic cycle for this paper, which assumes a closed system, we chose to create a conceptual urban hydrologic flow regime diagram (Figure 1.3) that includes all of the relevant fluxes when the urban domain is taken as the primary boundary of interest.
Figure 1.3 Urban hydrologic processes with the urban domain set as the primary boundary and interactions between surface water, groundwater, water users, and infrastructure.
The fluxes that cross the theoretical urban domain are shown around the perimeter of the domain in Figure 1.3:
• Imported Water Flux – Cities generally require additional sources of potable water to supplement the available local water resources; this flux includes pumping sources, water brought in through aqueduct systems, and other sources of potable water flowing into the urban domain.
URBAN DOMAIN ATMOSPHERIC WATER
IMPORTED
WATER FLOWOUT
URBAN DOMAIN INFRASTRUCTURE Infrastr. Atmospheric Water Imported Water Groundwater Inflow Surface Water Inflow
Outflow UR BA N D O MA IN AT MO SPH ER IC W AT ER IMPO RT ED W AT ER O UT FL O W UR BA N D O MA IN IN FR AST RU CT UR E UR BA N D O MA IN AT MO SPH ER IC W AT ER IMPO RT ED W AT ER O UT FL O W UR BA N D O MA IN IN FR AST RU CT UR E UR BA N D O MA IN AT MO SPH ER IC W AT ER IMPO RT ED W AT ER O UT FL O W UR BA N D O MA IN IN FR AST RU CT UR E UR BA N D O MA IN AT MO SPH ER IC W AT ER IMPO RT ED W AT ER O UT FL O W UR BA N D O MA IN IN FR AST RU CT UR E Urban Domain
• Atmospheric Water Fluxes – These fluxes are representative of precipitation and evapotranspiration. Both of these fluxes have been shown to increase due to urbanization (Collier, 2006; Kanda, 2007).
• Surface Water and Groundwater Inflow – Surface water and groundwater inflow that crosses into the urban domain.
• Infrastructure Outflow – Infrastructure outflow includes pumping from within the domain to outside the domain, wastewater outflow and any other anthropogenic outflows that may cross the urban boundary.
• Surface Water and Groundwater Outflow – Surface water and groundwater outflows are those natural water fluxes that either originate within the domain or pass through the domain.
Within the urban domain we represent the four broad places that water is stored and the fluxes between them:
• Infrastructure – Any of the potable and non-potable water related man-made
infrastructure present within the domain. Inflow fluxes include: imported water flow, groundwater pumping flow, municipal, industrial, commercial, and private
wastewater discharge, groundwater infiltration, and surface water flow through culverts and conduits. Outflow fluxes include: Potable water delivery to users,
artificial and urban induced (leakage etc.) groundwater recharge, and wastewater and potable water flow out of the domain.
• Surface Water – Surface water storage present within the domain include reservoirs, open-channels and natural rivers and lakes. Inflow fluxes include: precipitation runoff, imported water runoff, infrastructure inflow (outfall pipes, industrial water inflow,
etc.), and natural baseflow, interflow and surface water flow into the domain. Outflow fluxes include: municipal, industrial, commercial and private uses, irrigation and other diversion, and natural groundwater recharge and surface water flow out of the urban domain.
• Groundwater – Groundwater storage underlying the urban domain include confined and unconfined aquifers and soil moisture. Inflow fluxes include: natural, artificial, urban induced recharge, and natural groundwater flow into the domain. (See Section 1.6 for more detail on sources of groundwater within the urban domain.) Outflow fluxes include: anthropogenic pumping, flow into surface water bodies, capillary rise, flow into leaky infrastructure and natural groundwater flow out of the urban domain. • Users – All municipal, industrial, commercial, and private users who interact with
potable and non-potable water within the domain. Inflow fluxes include: Potable water use (imported and local), and non-potable water use by municipal, industrial and commercial users. Outflow fluxes include: Wastewater discharge and outdoor water use that may runoff to surface water or recharge groundwater.
Note that Figure 1.3 illustrates the relevant fluxes in relation to the urban domain. The fluxes are not proportional to their real-world magnitudes and may not all be present as described above in a given domain. Similar figures may be created using the watershed, aquifer system, or sewershed as governing domains. As with all scientific concepts, it is important to look at the hydrologic cycle critically and with knowledge of who created these illustrations and for what purpose. The “invention” of the urban hydrologic cycle, then, should be carried out carefully and must be mindful of the hugely varying hydrologic, historic, infrastructural, geographic, and societal differences found in cities around the world. Lerner (2002, 1990), Foster et al. (1999),
Yang et al. (1999), and Vázquez-Suñé et al. (2005) all provide urban hydrologic cycle
illustrations that are useful for various conceptual uses. For instance, Foster et al. (1999) provide an illustration of how the urban density overlying hydrogeology, and the hydrogeology itself, lead to differing levels of groundwater contamination in Merida, Mexico and Santa Cruz, Bolivia. Yang et al. (1999) and Lerner (2002), on the other hand, choose to show the urban recharge pathways of precipitation and water supply explicitly. Each of the figures show similar fluxes but emphasizes distinct processes to make their point. Schirmer et al. (2013) and Hibbs and Sharp (2012), meanwhile, provide illustrations of the expected changes urbanization brings to the fluxes mentioned above.
Groundwater in the City
The next century presents enormous water related challenges for cities around the world. Populations are projected to increase, cities will grow larger and become denser, and at the same time climate change will threaten many sources of potable water (Gober, 2010). In many cities, especially those located in semi-arid/arid regions, importing water from more plentiful and pristine locations is common. This practice has long been understood to be an unsustainable and potentially unreliable form of providing potable water, especially as populations increase and the climate changes. The use of local sources of water is vital to sustain potable water use in urban areas. Groundwater is already of major importance in providing water to cities, however, the underutilization of this local source does occur in more developed countries where importing water is prevalent (Foster, 2001). Increasing the use and recharge of groundwater and other local sources from storm events, recycling, and other sources will be vital to mitigating the deleterious effects of mismanaged groundwater and reduce the projected scarcity of water (Larsen et al., 2016; Niemczynowicz, 1999). However, if more stormwater is to be recharged, important
questions on how this will affect hydrologic processes within the urban domain must be answered. Integration of urban groundwater fluxes into urban water management plans is necessary for long-term, sustainable, planning (Wolf et al., 2006). The relationship between water and energy, often called the “water-energy nexus” (Kenway et al., 2011b), is increasingly gaining importance in cities as well. The sustainability of cities depends on the efficient use of these two inextricably linked resources. Large gaps in research exist, especially in semi-arid regions where more energy and economic resources are placed into obtaining potable water (Gober, 2010).
1.6.1 Groundwater Recharge Sources
As discussed in Section 1.5, water in the urban domain has more recharge pathways than water in natural settings. Natural groundwater recharge processes may be broken down into three types (de Vries and Simmers, 2002):
• Direct Recharge – the vertical recharge flux occurring uninterrupted through the vadose zone, at times referred to as diffuse recharge (Nolan et al., 2007), often simply calculated as Infiltration = Precipitation – Evapotranspiration – Runoff for a given area. However, this can be much more difficult to calculate given local geology. As aridity increases in a region, this type of recharge becomes less significant.
• Indirect Recharge – the recharge flux occurring at the interface between surface water bodies and the groundwater system; at times referred to as focused recharge (Nolan et al., 2007). As aridity increases in a region, this type of recharge becomes the primary source of groundwater in natural environments (Alley et al., 2002).
• Localized Recharge – the recharge flux occurring between non-well-defined flow channels and the groundwater system. Local topographic and geologic features (karst
formations, etc.) cause this type of recharge and may be significant in certain regions. The terms “localized recharge” and “indirect recharge” are often used
interchangeably (e.g. Alley et al., 2002; Scanlon et al., 2002) but we propose to use the more detailed definitions used by de Vries and Simmers (2002) to differentiate between important recharge processes in urban areas.
For a more detailed review of flow and storage in natural groundwater systems please see Alley et al. (2002). All three of these fluxes are present within the urban domain, albeit at a much lower rate than in natural systems due to altered rates of evapotranspiration and increased surface runoff. The decrease in these natural recharge fluxes are largely counteracted, and at times surpassed, by increases in urban specific recharge processes. Septic tanks, potable water supply mains, sewer and storm water drains, and groundwater spreading grounds all constitute the basic water infrastructure of cities. This infrastructure is often poorly maintained and leakage is a common occurrence (Lerner, 1986). Sewer, drainage and other unpressurized pipe networks can represent both sources (exfiltration from infrastructure) and sinks (infiltration into infrastructure) for groundwater systems depending on the height of water table with respect to the infrastructure (Rutsch et al., 2008). Pressurized pipes, such as potable water mains, will cause an increase in recharge if compromised. These types of pipes have been found to lose anywhere from 3% to over 50% (Puust et al., 2010) of their total water carrying capacity to leakage, depending on the level of maintenance of the infrastructure and volume of water transported, accounting for large loses in potable water and associated expenses. Urban specific recharge then includes the following anthropogenic sources:
• Urban Induced Recharge – As discussed above, this type of recharge originates from the extensive man-made infrastructure present in the urban domain. It may emanate
from gravity mains, pressurized pipes, tunnels, septic systems, etc. The recharge flux may be spatially positive or negative with respect to the groundwater system
depending on level of maintenance and groundwater depth.
• Urban Irrigation Excess – An often-overlooked source of recharge is outdoor water use within the urban domain. In some locations, imported potable water is used to irrigate lawns, golf courses, parks and other urban greenery throughout the year. Water that is not taken up by the root system either runs off to the storm drainage system or percolates through the soil layer to become groundwater.
• Artificial Recharge – Often used in water recycling/reuse (Bickerton et al., 2011; Chen et al., 2012; Exall, 2004) and saltwater intrusion barrier (Bray et al., 2007; Werner, 2010) schemes. This type of recharge is targeted and engineered to alleviate the stresses places on the groundwater system by increasing the amount of water present in the system.
1.6.2 Groundwater-Surface Water Interactions
Although often thought of as two isolated components of the hydrologic cycle,
groundwater and surface water interact in a variety of ways that were once ignored but are now considered to be extremely important, especially for ecological reasons (Alley et al., 2002). Some of these interactions include event flow, baseflow, interflow and the interactions that occur in the ecologically important hyporheic zone. These interactions “are governed by the positions of the water bodies with respect to groundwater flow systems, geologic characteristics of their beds, and their climatic settings” (Sophocleous, 2002). Sophocleous (2002) goes on to provide a comprehensive review of these interactions in natural settings along with the impacts humans and urbanization have on groundwater-surface water interactions. For instance, pumping may
induce surface water sources to recharge aquifers as water is drawn out resulting in surface water depletion. The channelization, diversion, and burying of rivers and streams within the urban domain often cause groundwater-surface water interactions to be overlooked in urban hydrologic studies. However, these interactions occur in various places throughout the urban domain,
predominantly at headwaters of stream channels where less urbanization and non-channelized portions of the stream network are observed (Hibbs et al., 2012).
1.6.3 Water Table Change
Along with losses of potable water sources, poorly managed groundwater systems can cause both long-term and catastrophic disaster, especially in the heavily built-up environments within the urban domain. The two most commonly researched and discussed consequences of groundwater depletion are salt-water intrusion in coastal areas (Bray et al., 2007; Werner, 2010) and land subsidence (Galloway and Burbey, 2011; Holzer and Galloway, 2005). Conceptually, land subsidence occurs when an aquifer system is depleted to the point that no water is available to fill in the voids created in the soil and a collapse of those air voids occurs. In the case of salt-water intrusion, the pumping of salt-water out of the groundsalt-water system causes saline groundsalt-water to move towards the site of the pumping. Salt-water intrusion and land subsidence cause many wide-ranging issues, including: the depletion of possible sources of potable water, building foundation and footing damage, subsurface infrastructure damage, and ecosystem losses. Water table rise can also present problems in some humid regions with increased risk of flooding, liquefaction, and landslides amongst other issues. Mitigating these issues requires data and good models, however, data on water table depths is limited in most urban areas and infiltration and outflow fluxes are hard to estimate. Remote sensing methods of obtaining groundwater table depths, such as the GRACE satellites, are useful when data is scarce and knowledge of the
groundwater system is necessary (Henry et al., 2011) and methods that incorporate other hydrologic components that can be validated help in estimating relevant fluxes.
1.6.4 Groundwater Quality
Groundwater in cities is subject to various industrial and commercial pressures that degrade its quality. The slow moving nature of aquifer systems, especially confined aquifers that can take centuries or millennia for water to pass through them (Foster and Chilton, 2003), cause quality issues to linger and make improving water quality a difficult task. Point sources of
groundwater pollution are often caused by industrial or commercial activity while diffuse sources of pollution are caused by the degraded nature of water that recharges in urban areas (Foster and Chilton, 2003). As cities become denser and more developed, point sources of pollution decrease while diffuse pollution becomes more prevalent. Lerner and Barrett (1996) identify issues with groundwater in the United Kingdom, identifying historic and present-day sources of pollution. Vizintin et al. (2009) use an urban water management model, a sewer infiltration/exfiltration model, and groundwater models to understand the effects of diffuse pollution on urban groundwater sources. They find that residential land-uses have a much smaller, yet still important, impact on groundwater quality when compared to agriculture or industry. Various water sources also contain unique geochemical tracers that can be used to estimate groundwater recharge values (Barrett et al., 1999) and help in fully understanding the urban water budget (Carlson et al., 2011).
1.6.5 Importance as a Resource
The next century presents enormous water related challenges for cities around the world. Populations are projected to increase, cities will grow larger and become denser, and at the same time climate change will threaten many sources of potable water (Gober, 2010). In many cities,
especially those located in semi-arid/arid regions, importing water from more plentiful and pristine locations is common. This practice has long been understood to be an unsustainable and potentially unreliable form of providing potable water, especially as population’s increase and the climate changes. The use of local sources of water is vital to sustainably using water in urban areas. Groundwater is already of major importance in providing water to cities, however, the underutilization of this local source does occur in more developed countries where importing water is prevalent (Foster, 2001). Increasing the use and recharge of groundwater and other local sources from storm events, recycling, and other sources will be vital to mitigating the deleterious effects of mismanaged groundwater and reduce the projected scarcity of water (Niemczynowicz, 1999). However, if more stormwater is to be recharged, important questions on how this will affect hydrologic processes within the urban domain must be answered (Larsen et al., 2016). Integration of urban groundwater fluxes into urban water management plans is necessary for long-term, sustainable, planning (Wolf et al., 2006). The relationship between water and energy, often called the “water-energy nexus” (Kenway et al., 2011b), is increasingly gaining importance in cities. The sustainability of cities depends on the efficient use of these two inextricably linked resources. Gaps in research exist however (Kenway et al., 2011b), especially in semi-arid regions where more energy and economic resources are placed into obtaining potable water (Gober, 2010).
Recharge Estimation Methods
There are a multitude of groundwater recharge estimation techniques used in both urban and natural systems, in a variety of hydroclimatic settings, and at varying levels of detail. We classify these methods into three groups; water balance (1.7.1), physical (1.7.2–1.7.5), and theoretical methods (1.7.6–1.7.9). This is done to facilitate a brief discussion of methods that
have been applied in numerous previous studies and to provide terminology for the theoretical, or numerical, methods used (1.8) in the following chapters.
These techniques are often applied to various portions of the hydrologic cycle, in conjunction or independently, to obtain potential or actual recharge estimates at varying space/time scales. Scanlon et al. (2006) review ~140 recharge studies in arid and semi-arid regions using a multitude of these methods. The choice of which technique to use is an important one, with implications for the robustness of results and soundness of the science. Scanlon et al. (2002) provide an approach and general guidelines for choosing an appropriate methodology. Their approach relies on a well-defined purpose for obtaining recharge estimates. For instance, if evaluation of available groundwater resources is important then techniques that are regional in scale and provide an adequate timescale should be used. Other factors are of great importance including cost, time constraints, and available data. See Scanlon et al. (2002) and Healy (2010) for more information.
1.7.1 Water Balance Method
The water budget method is a simple technique derived from the study of mass fluxes through a natural or constructed system. At its simplest, this technique can be described as accounting for all the mass or volume fluxes of interest, in this case water, passing through a defined domain as follows:
∆𝑆 = 𝑄* − 𝑄, (1.1)
Where ∆𝑆 is the change in storage of water within the domain, including changes in soil moisture, and natural and constructed groundwater and surface water storage (Kenway et al., 2011a). 𝑄* and 𝑄, are all the relevant input and output fluxes respectively. Kenway et al. (2011a) provide various methods used in urban water budget studies, highlighting the importance of
choosing an appropriate domain and the inclusion of urban specific fluxes. Recharge in these methods is often revealed through the change in storage term or a dedicated “residual term” in the calculations. This makes it difficult to understand the spatial variations present within the studied domain and how differing land use/land cover or infrastructural change affects recharge. Large uncertainties are created in the recharge estimate due to the indirect measurement of this value and its dependence on a large, detailed and accurate dataset.
Nevertheless, this method can be significantly useful in identifying major sources and sinks to a hydrologic system and potential causes of change in those systems (Passarello et al., 2012). Kim et al. (2001) and Bhaskar and Welty (2012) conclude that infiltration into and leakage from water infrastructure account for large portions of the water budget in their study domains, however the net effects of these processes results in negligible changes in Seoul and a net export of water in Baltimore further highlighting the importance of treating every system, especially urban systems, independently. Ngo and Pataki (2008) and Shen and Chen (2010), applying the water balance method to semi-arid regions, determine that more comprehensive studies of the hydrological processes and more rigorous methods of calculating water losses are needed for improved evaluation and management of these systems.
1.7.2 Physical Methods
These techniques are defined by the measuring of a physical property of the earth system, either directly or indirectly, that can then be applied to obtain a measure of recharge. Traditional methods have included a range of techniques briefly discussed in Sections 1.7.3 (chemical tracers) and 1.7.4 (lysimeters, etc.). Section 1.7.5 discusses fairly recent advances in remote sensing that have accelerated the use of multiple products in global and regional studies of water resources, ranging from direct observations of land surface temperature (LST) such as with
MODIS or Landsat instruments, to indirect measurements of groundwater storage from the GRACE satellite.
1.7.3 Chemical Tracers
Water originating from different locations possesses unique chemical and physical signatures, or tracers, which can be identified. By knowing the unique signatures of source waters and measuring the quantities or physical indicators of these tracers in receiving waters, we can quantify the composition, and sources of receiving waters. Physical characteristics, such as heat (Foulquier et al., 2009), have been used as tracers, along with naturally occurring
chemical properties and/or artificially applied tracers (Barrett et al., 1999; Hibbs et al., 2012; Huang and Pang, 2011; Vázquez-Suñé et al., 2010). In natural and urban settings, tracers have been used to understand the composition of both surface water bodies and groundwater aquifers. Most commonly in urban areas, the inherent chemical or physical properties of various sources of groundwater infiltration have been used as tracers. Theoretically, various urban sources of groundwater should exhibit very different chemical and physical properties (Barrett et al., 1999). Potable water may contain chlorination by-products, sewage may display certain chemicals commonly found in detergents and various organic compounds, and precipitation infiltrating through various different pathways should exhibit their own unique properties (Barrett et al., 1999). Of course, finding, measuring and understanding these tracers in a fast, cost-efficient, and repeatable manner is challenging.
This technique is well suited for the estimation of the total compositions of a given groundwater source. However, estimates of spatially distributed, or instantaneous recharge rates are difficult due to the low temporal resolution inherent in this technique and the cost and time necessary for high spatial resolution. Nonetheless, tracer techniques have been vital in
constructing historical recharge rates for various aquifers (Cartwright and Morgenstern, 2012; Currell et al., 2010; Manning et al., 2012).
1.7.4 Direct Measurement
Surface water instruments or techniques may include seepage meters and baseflow calculations done through the use of discharge records. Soil or groundwater instruments or techniques may include lysimeters, ground-penetrating radar (GPR; Huisman et al., 2003; Strobach et al., 2010), well records indicating water depth changes and direct measure of surface water flow (Delin et al., 2007; Healy and Cook, 2002). These techniques are limited by their spatial and temporal applicability, and time and cost to implement. Spatially, they represent point-scale measurements within a given time period. These limitations may be alleviated by the larger scale use of any of these instruments, which may further increase the time and cost (labor and capital costs) associated with the study. However, these methods may be an ideal solution for certain studies conducted in a small (i.e. watershed scale) domain with lower temporal needs.
1.7.5 Indirect Measurement
The advent of remote sensing technologies for earth observing applications has created new possibilities for groundwater system analysis (Becker, 2006). Remote sensing technologies rely on the detection of electromagnetic waves that are reflected off, emitted by, or transmitted through Earth’s surface and/or atmosphere by detectors placed on satellites, aircraft, or towers. All of these instruments are limited by the electromagnetic wave technology they utilize; for instance, only the first few centimeters of soil moisture have been able to be estimated through these techniques. There also exists a tradeoff between spatial and temporal resolution with these techniques, especially with satellite remote sensing.
MODIS (MODerate-resolution Imaging Spectroradiometer), Landsat, AMSR-E
(Advances Microwave Scanning Radiometer for EOS), and many other earth-observing satellite system based products may be used in conjunction (Kim and Hogue, 2012) or independently with various algorithms to determine soil moisture (Brocca et al., 2011; Owe et al., 2008), recharge (Khalaf and Donoghue, 2012; Szilagyi and Jozsa, 2012), and other hydrologically important parameters (Brunner et al., 2007; Leblanc et al., 2007). The Soil Moisture Active Passive (SMAP) satellite (Entekhabi et al., 2010) is in development for the specific purpose of measuring soil moisture at higher resolutions and with more accuracy than previous satellites (i.e. AMSR-E). The Gravity Recovery and Climate Experiment (GRACE) satellites have also been useful in estimating recharge at regional scales. The GRACE satellites are a pair of
satellites that can measure minute changes in Earth’s gravitational field, and therefore changes in mass at given locations, which through the use of various algorithms and models can estimate changes in water storage. GRACE derived products are provided at relatively low resolutions for urban studies, and at approximately monthly time steps, consequently, studies have mostly been limited to global, continental, or major river basin scales (Henry et al., 2011). Bhaskar and Welty (2012) have used GRACE satellite derived data to compare and validate results from a water balance conducted in an urban domain. Further work has been conducted and global and regional scales (Döll et al., 2014; Joodaki et al., 2014; Voss et al., 2013), demonstrating the potential use of this data as validation and or calibration products.
1.7.6 Theoretical Methods
In the following section, a theoretical method refers to those methods that mathematically solve the fundamental (physically or empirically based) equations of fluid flow and/or radiative transfer to model the hydrologic system. These mathematical models may be classified in various
ways by taking into account their distinct process descriptions, space and time scales, solution techniques, and usage. Generally, numerical models have been applied to distinct portions of the hydrologic cycle with process based mathematical representations of relevant processes. For our purposes, we broadly identify three major branches of numerical models of the hydrologic cycle: 1) watershed, or rainfall-runoff (RR) models that model stream generation and overland flow (Singh and Woolhiser, 2002); 2) groundwater flow models that can be applied to both the vadose and phreatic zones; and 3) land surface models (LSMs). These three classifications are broad in nature and define the focus of the models rather than the full range of processes modeled by each. Most of the numerical models that fall within these three classifications are able to provide some estimate of infiltration to the saturated zone, however each has its own set of drawbacks in regard to this estimate. Sections 1.7.7, 1.7.8, and 1.79 below seek to broadly describe each type of model, terms and potential limitations in regard to recharge estimation associated with each.
1.7.7 Watershed Models
Due to their attempt at modeling the full hydrologic cycle, these watershed models are the broadest of the three categories defined in this section. The idea of combining the many component models that had been developed in the late 1800’s and early 1900’s began with the seminal work of Crawford and Linsley (1966) in developing the Stanford Watershed Model (now HSPF), the basis for numerous contemporary models. This work continues to this day with the development of new watershed models, the improvement of physical representations of processes, improved parameter estimation and better understanding of process coupling (see Singh and Woolhiser, 2002). These models are often further classified as lumped or distributed, in reference to how they handle spatial heterogeneity, and can be predicated on empirically (mathematical equations derived from experimentation) or physically (mathematical equations
derived from the fundamental analysis of fluid flow) based equations. These classifications can be applied to all types of mathematical models but lumped empirical modeling is much more common in watershed models (TR-55, etc.). In terms of recharge estimation, watershed models are limited by many factors including the lack of explicit groundwater flow parameterization, poor ET estimation, and poor overland flow generation schemes. Many may only have simple shallow subsurface processes represented, lacking deeper aquifer system representation.
However, recent generations of watershed models have largely provided remedies to these issues by fully integrating surface and subsurface processes allowing for study of the interactions of these processes (Furman, 2008; Kampf and Burges, 2007).
1.7.8 Groundwater Models
Modern groundwater models (MODFLOW, Parflow, etc.) have the ability to run in a variably saturated state, most commonly by solving Richards’ Equation (Richards, 1931), which can adequately represent subsurface flow when given a reasonable upper boundary condition and adequate subsurface parameterization. Groundwater models spearheaded the use of distributed, grid-based approaches to hydrologic modeling through the use of numerical methods that made the solution of the complex differential equations feasible with modern computing. This type of modeling approach has, ostensibly, lead to the increased use of distributed modeling in
hydrology. However, the two largest assumptions commonly made when building a groundwater model are the value of the upper boundary condition and the requisite simplification of the complex hydrogeology of the modeled domain. In the past decade, these models have been increasingly explicitly coupled to or used in conjunction with watershed models to create a more robust and spatially distributed upper boundary condition.
1.7.9 Land Surface Models
Land surface models (LSMs), included here though originally developed by and for the atmospheric science community to provide a lower boundary condition for general circulation models (GCMs), have been increasingly important in advancing research in the hydrologic sciences (Pitman, 2003; Wang and Dickinson, 2012). These models resolve the energy budget at the land surface of which the latent heat flux is of extreme importance to hydrologists. LSMs in the hydrologic community are principally used as an upper-boundary condition for watershed or groundwater models by utilizing observed or model-derived meteorological data to force the models. They have also been instrumental in understanding the effects of land use/land cover change on hydrology and associated feedbacks with the energy budget (Maxwell and Kollet, 2008a). In hydrologic studies, these models are run at much higher resolutions (resolutions of 5 to 0.5-km or lower) than in atmospheric research (resolutions of much greater than 5-km are common). Along with the assumptions made in land use/land cover parameterization,
evapotranspiration calculation, and the homogeneity within a given model cell; LSMs often do not contain robust hydrologic formulations. Typically surface and subsurface lateral flow is unaccounted for between model cells and deeper subsurface processes are not explicitly
represented (Rihani et al., 2010). The coupling of these models with the more robust hydrologic representations provided by watershed and groundwater models can provide improvements on some of these limitations.
CHAPTER 2
IMPACT OF LATERAL FLOW AND SPATIAL SCALING ON THE SIMULATION OF SEMI-ARID URBAN LAND SURFACES IN AN INTEGRATED HYDROLOGIC AND
LAND SURFACE MODEL
Modified from a paper published in Hydrological Processes1
Bryant Reyes2, 3, 4, Reed M. Maxwell1, 5, 6, Terri S. Hogue1, 1, 1
Abstract
Understanding and representing hydrologic fluxes in the urban environment is
challenging due to fine scale land cover heterogeneity and lack of coherent scaling relationships. Here, the impacts of urban land cover heterogeneity, scale, and configuration on the hydrologic and surface energy budget (SEB) is assessed using an integrated, coupled land
surface/hydrologic model at high spatial resolutions. Archetypes of urban land cover are simulated at varying resolutions using both the National Land Cover Database (NLCD; 30 m) and an ultra high-resolution land cover dataset (0.6 m). The analysis shows that the impact of highly organized, yet heterogeneous land cover typical of the urban domain can cause large variations in hydrologic and energy fluxes within areas of similar land cover. The lateral flow processes that occur within each simulation create variations in overland flow of up to ±200%
1 Reprinted with permission from John Wiley and Sons; see Appendix A 2 Corresponding author: brreyes@mines.edu
3 Department of Civil and Environmental Engineering, Colorado School of Mines, Golden, CO 4 NSF Engineering Research Center ReNUWIt, Colorado School of Mines, Golden, CO 5 Department of Geology and Geological Engineering, Colorado School of Mines, Golden, CO 6 Hydrologic Sciences and Engineering Program, Colorado School of Mines, Golden, CO
and ±4% in evapotranspiration. The impact on the SEB is smaller and largely restricted to the wet season for our semi-arid forcing scenarios. Finally, we find that this seasonal bias, predominantly caused by lateral flow, is displaced by a systematic diurnal bias at coarser resolutions caused by deficiencies in the method used for scaling of land surface and hydrologic parameters. As a consequence, we have produced land surface parameters for the widely used NLCD urban land cover types. This work illustrates the impact of processes that remain
unrepresented in traditional high-resolutions land surface models and how they may affect results and uncertainty in modeling of local water resources and climate.
Introduction and Background
The rapid development and population growth in many of the world’s water stressed cities has amplified the importance of predicting and managing local urban water resources (McDonald et al., 2011; Shen and Chen, 2010). However, many of the tools used for this purpose within the research and operational modeling communities do not adequately represent the range of natural and anthropogenic processes observed at the urban scale. One important area of improvement is the surface/upper boundary conditions of physical hydrologic models and the surface parameterizations of land surface models (LSMs). The need for tools that account for physical hydrometeorologic and anthropogenic processes and interactions and their impact on local water resources at relevant spatial scales and various climatic/management scenarios is vital to policy makers, the research community, and for the sustainable use of potable water in cities around the world (Fletcher et al., 2013; Schirmer et al., 2013).
Urbanized watersheds have long been characterized by the changes in land use and land cover from natural conditions. Increases in imperviousness from street and building construction, introduction of industry and flood mitigation infrastructure, and the importation and extraction of
