Professional Documents
Culture Documents
Comparison of Two Watershed Models For Addressing Stakeholder Flood Mitigation Strategies: Case Study of Hurricane Alex in Monterrey, México
Comparison of Two Watershed Models For Addressing Stakeholder Flood Mitigation Strategies: Case Study of Hurricane Alex in Monterrey, México
net/publication/318663053
CITATIONS READS
8 356
3 authors, including:
Some of the authors of this publication are also working on these related projects:
Climate change impact assessment on the hydrological cycle in Europe View project
All content following this page was uploaded by Enrique Vivoni on 15 October 2017.
Abstract: Extreme flooding in the metropolitan area of Monterrey, México has led to an interest from local stakeholders in potential mit-
Downloaded from ascelibrary.org by 189.204.175.21 on 07/08/17. Copyright ASCE. For personal use only; all rights reserved.
igation strategies in the Santa Catarina watershed. The authors analyze a set of hydraulic infrastructure options using two hydrologic models
of varying complexity in the context of a major flood caused by the landfall of Hurricane Alex in 2010. A consistent approach was used
to provide terrain, soil, vegetation, and meteorological data to each model—hydrological modeling system (HEC-HMS), and triangulated
irregular network (TIN)–based real-time integrated basin simulator (tRIBS)—and to test the models with streamflow and water level ob-
servations. Simulation analyses focus on the differential ability of the two models in capturing precipitation and watershed properties and its
effects on the hydrologic response in the presence of hydraulic infrastructure options. A scenario with a single, large dam reduced the flood
peak more favorably than three smaller structures. The model comparison is effective in addressing stakeholder-driven mitigation strategies
and revealing the added value of spatially-distributed approaches. DOI: 10.1061/(ASCE)HE.1943-5584.0001560. This work is made avail-
able under the terms of the Creative Commons Attribution 4.0 International license, http://creativecommons.org/licenses/by/4.0/.
Author keywords: Extreme event; Flood mitigation; Hydraulic infrastructure; Model complexity; Distributed hydrologic modeling;
Remote sensing; Linear reservoir; Stakeholder engagement.
the flood peak during Hurricane Alex (reprinted from Maiz 2010, with 2010; Hernández et al. 2014). This topic is of broad importance to
permission) the hydrologic modeling community given the wide number of op-
tions for conducting watershed simulations (Singh and Woolhiser
2002; Kampf and Burges 2007). In addition, demonstrating the
utility of fully-distributed models as tools for evaluating flood mit-
occurred between local stakeholders and researchers. In this pro- igation strategies to hurricane landfalls and other extreme events
cess, the authors implemented a model comparison approach opens up their broader adoption in other regions (e.g., Gutiérrez-
for addressing the dual need for a detailed investigation of hydro- López and Ramírez 2005; Slutzman and Smith 2006; Fang et al.
logic processes and the planning of mitigation strategies in a 2011; Karamouz et al. 2015).
watershed management context. The modeling framework con-
sisted of the application of a semidistributed model (Hydrological
Modeling System, HEC-HMS, HEC 2015) and a fully-distributed Methods
model (triangulated irregular network–based real-time integrated
basin simulator, tRIBS, Ivanov et al. 2004a) in a commensurate
fashion and taking advantage of the merits of each method (see Study Watershed and Its Hydraulic Infrastructure
Singh and Woolhiser 2002 for description of various modeling ap- The study region is the Santa Catarina watershed (SCW)
proaches). While hydrologic models have been compared previ- (1,831 km2 ) in northeastern México, which encompasses Monter-
ously in México (e.g., Cisneros-Iturbe et al. 2007; Velázquez et al. rey, capital of Nuevo León [Fig. 2(a)]. A subtropical, semiarid cli-
2015), this is the first attempt to analyze the comparative utility of mate characterizes the watershed (Návar and Synnott 2000), which
semidistributed and fully-distributed models in the design and is a subbasin of the San Juan River (33,538 km2 ) that provides
evaluation of flood mitigation strategies in a Mexican watershed. water to Monterrey through reservoir operations at El Cuchillo
As noted by Fatichi et al. (2016), computational advances, data Dam (Scott et al. 2007). The metropolitan area occupies the lower
availability, and improved process coupling have increased the use areas of the SCW and has a population of nearly 4 million, with
of fully-distributed models in a wider set of engineering problems, an annual growth rate of 1.3% from 1990 to 2010 (INEGI 2011).
including for watershed management and infrastructure design. As an important city in México (INEGI 2009), Monterrey has a
Fig. 2. (Color) (a) City of Monterrey in Nuevo León, México; (b) Santa Catarina watershed based on a 30 m digital elevation model and the locations
of rain gauges, current hydraulic infrastructure (Rompepicos Dam located at 25.556°N and 100.397°W), and the stream gauge at the basin outlet
(Cadereyta Station); (c) land cover classes and the definition of basin units used in the flood event diagnosis; geographical data is in a projected
coordinated system of UTM Zone 14N and a datum of WGS 1984
Grasslands 2.07 near Monterrey [Fig. 2(b)], with 72-h totals ranging from 250 to
Human settlements 6.57 800 mm, and in some locations exceeding the mean annual precipi-
Mixed woodlands 20.83 tation in the semiarid region (SMN 2010). Ramírez (2010) esti-
Secondary shrublands 23.09 mated daily return periods between 20 and >100 years at daily
Shrublands 32.92 CONAGUA sites, though these tended to underestimate rainfall
Unvegetated 0.16 as compared to several automated stations. Furthermore, Hurricane
Urban areas 10.48
Alex made landfall prior to the primary rainy season occurring in
Note: 2.12% of the watershed area has soils classified as not available. August and September in northeastern México (Shreve 1944). The
a
Data from ISRIC (2013). low number of rain gauges (11 from daily CONAGUA sites, and 8
b
Data from INEGI (1993). from automated stations) in the SCW limited the ability to charac-
terize the spatiotemporal distribution of meteorological conditions
diversified economy that supports a sprawling amount of urban during Hurricane Alex. For this purpose, the authors obtained
land cover on relatively mild slopes. In contrast, the upper reaches meteorological fields from the North American land data assimila-
of the SCW are predominantly rural areas because of the highly- tion system (NLDAS, Mitchell et al. 2004), a reanalysis product at
sloped mountain ranges [Fig. 2(b)] that form part of the 1,000 km 12 km, hourly resolution consisting of rainfall, air temperature,
long Sierra Madre Oriental in northeastern México (e.g., Maqueda wind speed, solar radiation, pressure, and relative humidity. Fol-
et al. 2008; Ferriño-Fierro et al. 2010). Table 1 lists the areal cover- lowing Robles-Morua et al. (2012, 2015), the authors applied a
age of soil classes from the Food and Agricultural Organization daily, mean-field bias correction to the NLDAS rainfall field over
(FAO) and land cover classes in the SCW, with the majority of the the period June 1 to July 31, 2010 for forcing both models in a
urban area and human settlements occurring in the lower reaches of similar fashion. Fig. 3(a) compares the daily total precipitation
the basin [Fig. 2(c)]. While only a small fraction of the SCW is used averaged over the SCW from rain gauges and NLDAS, indicating
for agriculture, downstream areas in the San Juan River support a general underestimation in the NLDAS fields. A mean-field, daily
bias correction (Cázares-Rodríguez 2016) avoided the sharp dis-
important activities in this sector that are also subject to flood haz-
continuities in the rain gauge patterns [Fig. 3(b)] and resulted
ards (Scott et al. 2007). The primary land covers in the upper
in a large improvement in precipitation magnitude in NLDAS
reaches of the SCW are submountainous shrublands and secondary
[Fig. 3(c)]. Nevertheless, inconsistencies in the subdaily timing of
shrublands at low to midelevations, and mixed woodlands on the
precipitation and other meteorological variables remain within the
higher mountain slopes. The sequence of folded ridges in the SCW
NLDAS fields used as forcing to the two hydrologic models.
leads to a trellis-like stream network consisting of ephemeral chan-
nels that increases flood hazards upstream of the city during storm
events (Ramírez 2010). Simulation Approaches
Recurring floods from hurricane landfalls prompted the con-
struction of a flood control structure at Corral de Palmas, known Model Descriptions
more commonly as Rompepicos Dam [Fig. 2(b)]. The dam has a HEC-HMS is a semidistributed hydrologic model that is widely
gravity curtain composed of roller-compacted concrete, an elevation used in watershed management and planning because of its ease
of 70 m and a length of 240 m (Ramírez 2011). The structure has of implementation, flexible number of hydrologic representations,
two outlet structures: a secondary rectangular opening of 6 × 6 m and low computational demand (HEC 2015). Prior applications in-
at the base of the dam with a maximum capacity of 838 m3 =s and clude assessments of proposed reservoirs in decision-making con-
a main Creager spillway with a crest length of 60 m and a capacity texts (e.g., Emerson et al. 2003; Ganoulis et al. 2008; Tingsanchali
of 3,376 m3 =s at maximum water level (Ramírez 2011). Completed and Tanmanee 2012; Robles-Morua et al. 2015). Among the various
in 2004, Rompepicos Dam controls flooding generated in the options, the authors selected the soil moisture accounting scheme
uppermost reaches of the SCW and has served its design purposes for infiltration to carry out continuous simulations in response to
during Hurricane Emily in 2005 and Hurricane Alex in 2010. For Hurricane Alex that also required representations of canopy inter-
Hurricane Alex, only two hydrologic observations were available: ception and evapotranspiration. Table 2 describes the hydrologic
a visually-estimated maximum water level behind Rompepicos processes selected in HEC-HMS that were guided by prior applica-
Dam (2.5 m below the spillway or 650 m3 =s) and the continuous tions (Fleming and Neary 2004; Robles-Morua et al. 2015) with the
discharge record at the basin outlet (Cadereyta Station) by the aim to be as consistent as possible with the second modeling ap-
Comisión Nacional de Agua (CONAGUA). However, both of proach. The tRIBS model is a fully-distributed hydrologic model
these observations are considered to be uncertain given the flood that uses a triangulated irregular network (TIN) to represent a water-
damages occurring along the measured river reach at the Cadereyta shed at multiple resolutions (Vivoni et al. 2004), as opposed to the
Fig. 3. (Color) (a) Time series of the basin-averaged daily precipitation (mm=day) from the rain gauge network and the original NLDAS product
(12 km, 1 h resolution); spatial distribution of total precipitation (mm) for June and July 2010 obtained from (b) rain gauges using a Thiessen polygon
interpolation; (c) bias-corrected NLDAS product
subbasin areas used in HEC-HMS. Fully-distributed models capture kinematic wave routing through the stream network and the effects
spatial variations of landscape and meteorological conditions more of reservoirs on the flood wave using level-pool routing.
faithfully, but at the cost of a larger computational demand, and
thus with more limitations in decision-making contexts. Several Model Domain, Parameterization, and Calibration
prior applications of tRIBS have focused on flood event simula- A common set of topographic, soil, and vegetation products
tions (e.g., Vivoni et al. 2006; Nikolopoulus et al. 2011; Moreno were used for setting up and parameterizing the two hydrologic
et al. 2013; Hawkins et al. 2015) and have demonstrated skill in models in the SCW. The watershed domain was delineated from
representing the spatially-distributed basin response. Table 2 de- a 30 m digital elevation model (DEM) obtained from the advance
scribes the physical processes represented in tRIBS, including spaceborne thermal emission and reflection radiometer (ASTER)
the level-pool reservoir routing (Mays 2010) developed by the au- [Fig. 2(b)] for both modeling approaches. The authors applied ter-
thors to match the formulation in HEC-HMS. Overall, the selection rain analysis procedures from the HEC-GeoHMS (HEC 2013)
of physical processes in each model captured the continuous soil package to derive the basin boundary upstream of the Cadereyta
moisture dynamics influenced by precipitation events and the ef- Station and the stream networks for use in both models. A stream
fects of vegetation on interception and evapotranspiration as well as cell threshold of 10 km2 was used to maintain a low number of
subbasins in HEC-HMS (a total of 61), as depicted in Fig. 4(a), an polygons used in tRIBS as computational elements and an equiv-
enhancement with respect to prior modeling efforts in the SCW alent cell size of 56 m (Ivanov et al. 2004a; Vivoni et al. 2005).
where 19 subbasins had been specified (Ramírez 2010). The hydro- Fig. 4 shows differences in the domain representation in the two
graphic procedure of Vivoni et al. (2004) was applied to derive the models through the spatial distribution of the terrain slope. The
multiple-resolution TIN domain, resulting in 580,434 Voronoi complex terrain of the SCW yields variations in slope that are
Fig. 4. (Color) (a) HEC-HMS subbasin definition with the mean aggregated slope for each subbasin and schematic network representation; (b) tRIBS
Voronoi polygon network with the slope field and the stream network captured in the model; hydraulic infrastructure used in the simulations
(Rompepicos Dam, Entry to the City Dam, Detention Dams) and the stream gauge at the basin outlet (Cadereyta Station) are displayed in each
model representation
et al. (2004b). As a result, gridded soil parameters from ISRIC Vegetation Parameters—HEC-HMS
(2013) were averaged within land cover polygons, as were vegeta- C (mm) Maximum canopy storage 0.01–2.5 RS
tion parameters derived from the moderate resolution imaging D (mm) Maximum depression storage 0.176–2.36 L
spectroradiometer (MODIS) sensor. Cázares-Rodríguez (2016) fur- Soil Parameters—tRIBS
ther describes the use of pedotransfer functions (Van Genuchten K s (mm=h) Saturated hydraulic conductivity 0.01–34.74 L
1980) and vegetation relations (Méndez-Barroso et al. 2014) within θs Saturated soil moisture content 0.4–0.44 L
the model parameterizations. While this procedure reduced the na- θr Residual soil moisture content 0.06–0.07 L
tive resolution of the original soil and vegetation products (avail- λ0 Pore distribution index 0.165–0.277 L
able at ≥1 km), it provided a consistent means to input landscape Ψb (m) Air entry bubbling pressure 1.5 to 0.12 L
f (mm−1 ) Conductivity decay parameter 0.0008–0.011 C
properties to both models. Additional aggregation at the level of
ar Anisotropy ratio 25–90 C
subbasins was performed in HEC-HMS, as shown for terrain slope n Total porosity 0.44–0.49 L
in Fig. 4(a). Table 3 lists the vegetation, soil, and routing param- ks (J=msK) Volumetric heat conductivity 1.33 C
eters, their range of values, and sources used in the HEC-HMS and Cs (J=m3 K) Soil heat capacity 2,400,000 C
tRIBS simulations for the SCW. The range of values represents the
Soil parameters—HEC-HMS
variations across the land cover polygons within the watershed. The
K (mm=h) Maximum infiltration rate 0.01–13.57 C
spatial resolution and classification fidelity represented in the mod- Ss (mm) Soil storage 50–150 C
els exceed those in previous regional studies (e.g., Maqueda et al. T s (mm) Tension storage 25–50 C
2008; Ramírez 2010; Návar 2012). GW s (mm) Groundwater storage 50–150 C
Model calibration of soil and vegetation parameters (Table 3) Sp (mm=h) Soil percolation rate 1.27–5.4 L
followed previous HEC-HMS and tRIBS model applications where GW p (mm=h) Groundwater percolation rate 1–5.4 C
values were obtained from literature (L) or remote sensing (RS) GW c (h) Groundwater coefficient 100 C
products for similar site conditions (e.g., Van Genuchten 1980; Routing parameters—tRIBS
Schaap et al. 1999; Mitchell et al. 2004; Ivanov et al. 2004a, b; n Manning’s channel roughness 0.35 L
Mays 2010; Robles-Morua et al. 2012, 2015; Singh and Jain 2015) aB Channel width-area coefficient 2 C
assumed to be spatially uniform within each land cover polygon and bB Channel width-area exponent 0.5 C
considered to be a reliable means for reducing the overall parameter cv Hillslope velocity coefficient 2.45 C
space to be sampled in the model calibration exercise. As r Hillslope velocity exponent 0.4 C
shown in Table 3, the number of soil and vegetation parameter val- Routing parameters—HEC-HMS
ues is large for each model given the spatial variations represented n Manning’s channel roughness 0.35 L
across the land cover polygons, whereas the routing parameter val- nf Flow plane roughness 0.1–0.32 L
ues are spatially uniform within the stream networks depicted in cl (%) Channel losses 0.02–0.1 C
each model. Model initialization was aided by the dry conditions Note: For sources, C = model calibration; L = literature review; and RS =
prior to the arrival of Hurricane Alex. As discussed by Vivoni et al. remote sensing. Additional details on sources provided in Cázares-
(2010), semiarid regions in northern México are characterized by Rodríguez (2016).
ephemeral rivers within alluvial basins with deep groundwater
tables. High evaporative demands and low rainfall during the early
summer effectively reset the conditions of hydrologic systems to dry stakeholder engagement process. The authors also inspected the
states prior to the rainy period occurring later in the summer season. streamflow volume of the simulations for the Cadereyta Station
This is in contrast to more humid settings with smaller degrees of during the manual calibration exercise. Simulations were tested in
seasonality (e.g., Nikolopoulos et al. 2011; Nied et al. 2013; Massari a validation activity against the maximum water levels recorded at
et al. 2014) where time-variable initial conditions need to be ac- the Rompepicos Dam included as a hydraulic infrastructure during
counted for accurately. As such, a dry state was assumed in each the exercise (see Fig. 4 for its location). As such, the model cali-
model following prior flood forecasting efforts (e.g., Chu and bration and validation approaches are based on streamflow data
Steinman 2009; Hawkins et al. 2015). at the basin outlet and the maximum reservoir level at an internal
A manual calibration exercise was conducted with respect to location, respectively, exceeding prior efforts in model testing for
the streamflow record at Cadereyta Station where observations the extreme event under analysis (Ramírez 2010), but still focus-
were available at 1–6 h intervals by using the peak error as an ing attention on the primary concern for stakeholders—the peak
objective function given the uncertainties present in other aspects discharge—with less emphasis resulting in the other characteristics
of the streamflow observation and the emphasis of this metric in the of the calibrated hydrographs. Model calibration involved varying a
pedotransfer functions, or remote sensing were not varied during (HEC-HMS) and 0.84 (tRIBS) with the streamflow record and rel-
calibration, whereas the sensitive parameters identified earlier could atively small peak errors (Table 4) amounting to 0.83% (HEC-
not be derived easily through these methods. This is consistent with HMS) and 5.84% (tRIBS) of the peak discharge. In addition, the
the goal of reducing the over-parameterization of semidistributed Nash-Sutcliffe model efficiency (NSE) of 0.66 (tRIBS) and 0.38
and fully-distributed hydrologic models which are often criticized (HEC-HMS) indicate a better model performance for the fully-
for having an excessive number of variable parameters during a cal- distributed approach. Overall, the authors consider that the calibra-
ibration process (see discussion in Fatichi et al. 2016). tion exercise applied to each model yielded a good agreement to the
CONAGUA records at the basin outlet when considering: (1) losses
Hydraulic Infrastructure Scenarios of streamflow data and their uncertain values during a destructive
Using the calibrated models, the authors evaluated the sensitivity of flood event, (2) differences in precipitation timing among rain
the flood response in the SCW during Hurricane Alex to the pres- gauges and NLDAS, and (3) similarities achieved in both model
ence of hydraulic infrastructure. Based on stakeholder suggestions responses given their underlying differences. Having built confi-
during two workshops, four scenarios were considered in a consis- dence in the calibration at the basin outlet, testing at Rompepicos
tent fashion for each model: (1) removal of current hydraulic infra- Dam shows that the models matched well the estimated peak out-
structure (Rompepicos Dam); (2) evaluation of current conditions flow of 650 m3 =s (dashed horizontal line) and have a similar peak
(i.e., calibration case); (3) inclusion of an additional large dam at inflow to the reservoir of 1,500 m3 =s, thus achieving a peak reduc-
the Entry to the City location; and (4) inclusion of three small de- tion of 43% by the dam in both models. Unfortunately, there were
tention dams at locations with no current hydraulic infrastructure no streamflow observation into the Rompepicos Dam for model
(i.e., uncontrolled flow). Fig. 4 presents the locations of the pro- evaluation purposes. Note that semidistributed and fully-distributed
posed hydraulic infrastructure, with the same level-pool routing hydrologic model evaluations are not commonly conducted at in-
methodology applied in both models for the scenarios. To include ternal locations in a basin (see Ivanov et al. 2004b). In addition, the
the large dam at the Entry to the City site considered by CERNL close correspondence in the reservoir inflow and outflow behavior
(2010), the authors created a conceptual engineering design for the in the two models indicates the adequacy of the level-pool routing
new dam based on adjusting the elevation-discharge-storage rela- method developed for this study. Nevertheless, a few differences
tions of Rompepicos Dam (Ramírez 2011) using the topographic are noted between the models, namely an earlier rising limb and
conditions of the new location and retaining the overall dimensions quicker recession limb in HEC-HMS, as explored subsequently.
of Rompepicos Dam. This site is attractive because of its location
upstream of Monterrey and its ability to contain both uncontrolled
flows and the discharge from Rompepicos Dam. As a lower cost Comparisons of Internal Hydrologic Variability
alternative, the three small detention dams were located in key sites
where important contributions were simulated to the overall flood The internal variability of the flood response in the two hydrologic
response during Hurricane Alex. The authors conceptualized the models was compared by delineating three units (labeled Units 1,
design of smaller, concrete detention structures with a spillway 2, and 3) and selecting a representative channel site (labeled Site I,
to apply the level-pool routing at the three sites using local infor- II, and III) in each area. Fig. 6 presents the spatial location of the
mation on topography, channel dimensions, and upstream contrib- units and sites, while Table 5 summarizes the topographic, land
uting areas. These sites are intended to collectively reduce and cover, and infrastructure properties for each unit (also see the unit
delay the flood contributions from uncontrolled subbasins upstream boundaries in Figs. 2 and 3). This delineation afforded the ability to
of Monterrey. Comparisons of the scenarios were carried out for separate regions in the SCW according to their contributions to the
different units in the SCW exhibiting varying terrain properties Hurricane Alex flood event: (1) Unit 1 is upstream of Rompepicos
(labeled Sites I, II, and III) and along various main channel loca- Dam and consists of high forest cover and complex terrain, (2) Unit
tions (labeled Sites A, B, C, and D) to determine the flood sensi- 2 is a region with uncontrolled flow between Rompepicos Dam
tivity in both models. and the Entry to the City and is characterized by intermediate forest
cover and terrain ruggedness, and (3) Unit 3 contains flatter areas
with high amounts of impervious areas located downstream of the
Results and Discussion Entry to the City. The landscape differences between the three units
help to explain the variations in the flood response between the
hydrologic models at the three sites as quantified in Table 6. Note
Hydrologic Model Performances
that Site I (Unit 1) exhibits the same pattern identified at the in-
Numerical experiments using the two hydrologic models were con- flow of Rompepicos Dam, where HEC-HMS has a quicker re-
ducted over the period of June 28 to July 7, 2010. Rainfall in the sponse characterized by a lower time lag (10.9 h) as compared to
Fig. 5. (a) Cumulative discharge at the basin outlet (Cadereyta Station, in 106 m3 ) from observations and model simulations along with the basin-
averaged precipitation from rain gauges and the bias-corrected NLDAS fields; (b) simulated inflow and outflow discharges (m3 =s) at the Rompepicos
Dam from both models with the reported peak outflow labeled as “Observed”
Table 4. Model Performance Metrics for Simulations as Compared to the Larger differences between HEC-HMS and tRIBS in areas of
Observed Discharge at the Cadereyta Station complex terrain (Unit 1) as compared to flatter regions (Unit 3)
Metric tRIBS HEC-HMS were consistently found at other internal sites. To evaluate if other
factors, such as impervious urban cover, might cause differences
Peak Error (m3 =s) 251.72 35.94
Peak Error (%) 5.84 0.83 between the models, the authors evaluated the flood response along
Mean Error (m3 =s) 103.15 260.73 a set of main channel sites. Fig. 7 shows the locations and simulated
Mean Error (%) 10.34 26.15 flood events at four sites (Sites A, B, C, and D), while Table 6 pro-
NSE 0.66 0.38 vides a set of flood metrics for each model at these sites and the
CC 0.84 0.70 basin outlet. The HEC-HMS and tRIBS models show consistent
B 0.99 0.80 differences along the main channel, with a noticeable decrease
RMSE (m3 =s) 747.51 1,011.06 in the differences in peak timing among sites and an increase in
Note: Metrics are defined following Vivoni et al. (2006). Peak error (m3 =s) the difference in peak discharge as the contributing area grows
is the error between observed and simulated peak discharges, whereas mean (i.e., from Sites A to D). An important change in the flood response
error (m3 =s) is the error between observed and simulated mean discharge occurs between Sites C and D located upstream and downstream of
over the entire period. Peak and mean error (%) are normalized with respect the main metropolitan area of Monterrey. The substantial growth in
to the observed values. B = bias between the simulated and observed total
streamflow volume is attributed to runoff produced in the imper-
runoff volume for the entire simulation; CC = correlation coefficient; NSE =
Nash-Sutcliffe model efficiency; and RMSE = root mean squared error. vious regions in each model. Despite this contribution, the overall
differences in flood peak amount and timing between the two mod-
els only varies moderately (Table 6). This suggests that the amount
tRIBS (17.4 h). In addition, Site I shows a large difference in the of urban cover does not explain the flood response differences be-
peak discharge between the two models (729 m3 =s in HEC-HMS tween the semidistributed and fully-distributed models such that
and 533 m3 =s in tRIBS). At Sites II and III, differences between there is more sensitivity to the aggregation of terrain properties
HEC-HMS and tRIBS are progressively reduced, with nearly indis- (slope) as compared to land cover conditions (impervious cover).
tinguishable flood metrics for Site III (Table 6). This is explained Fig. 7 also demonstrates a few important features of the flood re-
by the different representations of slope (Fig. 4) in the two models, sponse represented in both models, namely (1) the contribution of
with the milder slopes in Unit 3 leading to smaller differences than Unit 1 (Site A) to the basin outlet is muted by Rompepicos Dam;
the more variable slope conditions in Unit 1. As a result, the simu- (2) Unit 2 (Site B) has an important amount of uncontrolled flows
lated flood response is more sensitive to the selection of a semi- (49 and 78% of the reservoir outflow volume in HEC-HMS and
distributed or fully-distributed approach in areas where a greater tRIBS); and (3) the urban area (between Site C and D) has a smaller
aggregation of terrain properties is performed. contribution to the flood volume downstream of the city (Site D) as
Fig. 6. Comparison of the hydrologic responses from HEC-HMS and tRIBS at (a) the three internal sites, along with hourly precipitation upstream of
each location; (b) Site I in Unit 1; (c) Site II in Unit 2; (d) Site III in Unit 3
Table 5. Terrain Properties and Land Cover Characteristics for Three Units Table 6. Simulated Flood Metrics at Interior Sites, Along the Main
(Units 1, 2, and 3) Channel, and the Basin Outlet
Parameter Statistic Unit 1 Unit 2 Unit 3 Peak discharge Time lag Volume
2 Location Model (m3 =s) (h) (106 m3 )
Area (km ) — 732 397 702
Elevation (m) Mean 2,108 1,617 720 Interior sites
Standard deviation 492 521 345 Site I tRIBS 533 17.42 75.49
Slope (degrees) Mean 26.94 26.51 11.42 HEC-HMS 728.9 10.92 77.16
Standard deviation 12.59 13.82 12.73 Site II tRIBS 247 17.30 34.54
Impervious area (%) — 0 0 44.6 HEC-HMS 307.4 11.92 27.41
Forested area (%) — 32.2 15.5 11.2 Site III tRIBS 335.4 11.49 39.61
Hydraulic infrastructure — 1 0 0 HEC-HMS 330.6 12.92 32.16
Main channel
Site A tRIBS 1,449 18.86 202.87
compared to the mountainous basin (only 42% of the total flood HEC-HMS 1,473 11.92 154.89
volume at the Cadereyta Station is produced in the urban area in Site B tRIBS 1,415 19.11 360.51
both models). This latter finding is significant because stakeholder HEC-HMS 1,401 12.92 231.16
perceptions varied widely as to the relative importance of the Site C tRIBS 2,104 16.86 457.29
mountainous basin and metropolitan area in generating the flood. HEC-HMS 2,174 13.92 308.03
In addition, both models consistently indicated the importance of Site D tRIBS 3,803 18.11 685.16
uncontrolled flows from natural and urban tributaries downstream HEC-HMS 3,927 15.92 530.94
Outlet tRIBS 4,059 18.92 714.60
of Rompepicos Dam to the overall flood magnitude at the basin
HEC-HMS 4,274 16.92 575.06
outlet.
Note: Peak discharge (m3 =s) is the maximum streamflow value. Time lag
(h) is the time difference between the flood peak and the centroid of the
Spatial Patterns of Hydrologic Model Response basin-averaged rainfall. Volume (106 m3 ) is the total amount of streamflow
Next, the authors compared the spatial variability of the hydro- over the entire flood period.
logic response in each model to identify the areas in the SCW with
the highest contributions to the flood event. It is well known
that interactions between terrain, soil, and vegetation patterns with 2010) for the HEC-HMS and tRIBS models. Because the variables
meteorological forcing create preferential areas of soil moisture ac- of interest are different for each formulation (i.e., saturation fraction
cumulation and runoff production (e.g., Smith and Hebbert 1979; in HEC-HMS, and root-zone relative moisture in tRIBS) the values
Sivapalan and Wood 1986; Ivanov et al. 2004b; Mascaro et al. 2015). shown in the colorbars do not match. Despite this, qualitative
Fig. 8 presents a comparison of the time-averaged soil moisture comparisons can be made among the semidistributed and fully-
conditions during the entire simulation period (June 28 to July 7, distributed models and the spatial patterns within each model can
Fig. 7. (Color) Comparison of the hydrologic responses from HEC-HMS and tRIBS at the four locations along the main channel as shown in the inset
along with hourly basin-averaged precipitation; Site A is the outflow from Rompepicos Dam, Site B is at the Entry to the City, Site C is upstream of
the urban area, and Site D is downstream of the urban area
Fig. 8. (Color) Comparison of the spatial distribution of time-averaged soil moisture as (a) saturation fraction distribution from HEC-HMS; (b) root
zone amount in the top 1 m from tRIBS
be assessed relative to the landscape properties and meteorological bottoms consisting of shrublands and secondary shrublands.
data. The primary trends captured in both models are (1) relatively Notably, the representation of soil moisture is more highly resolved
drier soils in areas upstream of Rompepicos Dam (Unit 1) due to in tRIBS, where spatial differences between drier mountain ridges
lower rainfall accumulations [Fig. 3(c)]; (2) a large contrast between and wetter valley bottoms are depicted well.
impervious urban areas and surrounding rural lands exhibiting Fig. 9 presents a comparison of the total runoff (mm) produced
higher wetness [Fig. 2(c)]; and (3) relatively wetter soils in valley during the simulation by each model and the runoff associated with
Fig. 9. (Color) Spatial distribution of the total runoff (mm) generated during the simulation period from (a) HEC-HMS; (b) tRIBS; spatial distribution
of the runoff (mm) generated from (c) infiltration-excess and (d) saturation-excess mechanisms simulated in tRIBS
two of the primary mechanisms in tRIBS (infiltration-excess and the proposed scenarios. While similar efforts have been conducted
saturation-excess runoff) (Ivanov et al. 2004a; Vivoni et al. with HEC-HMS (e.g., Emerson et al. 2003; Robles-Morua et al.
2007). Overall, a similar runoff pattern was produced in the semi- 2015), this study reports the first use of reservoir routing and infra-
distributed and fully-distributed approaches, driven in large part by structure scenarios with tRIBS. Fig. 10 shows the locations and
the rainfall accumulation [Fig. 3(c)]. Note that the NLDAS 12 km upstream contributing areas (CA) of the hydraulic infrastructure
pixels are visible in the runoff distribution from tRIBS as the com- scenarios, with additional details on the site locations presented in
putational elements are much smaller than the meteorological forc- Cázares-Rodríguez (2016). The large dam at the Entry to the City
ing, as opposed to the rainfall aggregation occurring in HEC-HMS. has a high upstream CA (1,129 km2 or 62% of the SCW area) and
Nevertheless, the close correspondence between the approaches serves as a clear demarcation between the upper rural area (Unit 1
builds confidence in the runoff capabilities of the fully-distributed upstream of Rompepicos Dam and Unit 2) and the lower urban
model with respect to the modeling tool commonly accepted by (Unit 3) area of the watershed. The location of the large dam (also
local stakeholders (Ramírez 2010). In addition, tRIBS adds a con- labeled as Site B in Fig. 7 and Table 6) coincides with a switch
siderable amount of spatial detail that captures the high-resolution between wet valley areas producing saturation-excess runoff and
interaction of rainfall characteristics with soil and terrain properties downstream urban areas where infiltration-excess runoff is the
in the SCW, which ultimately leads to runoff production in a land- dominant mechanism. In contrast, the three small detention dams
scape and discharge in the stream network. Aggregation of the have much lower CA ranging from 106 to 170 km2 , placed at stra-
watershed soil and terrain information in HEC-HMS results in a tegic locations in terms of uncontrolled runoff production during
substantial loss of information. Furthermore, the fully-distributed Hurricane Alex as simulated by the two hydrologic models. Two
model is able to simulate spatial patterns of different runoff mech- of the three detention dams correspond to analysis locations (Site II
anisms indicating that (1) the metropolitan area of Monterrey pro- and III) in Fig. 6 and Table 6. Both stakeholder strategies (large
duces runoff via the infiltration-excess mechanism due to the large
dam at the Entry to the City dam and three small detention dams)
amounts of impervious surfaces, (2) drier areas in the upper basin
retained the existing Rompepicos Dam within the two models.
[Fig. 8(b)] have moderate amounts of infiltration-excess runoff due
Fig. 11 summarizes the effects of the hydraulic infrastructure
to the formation of shallow layer of surface saturation in regions of
scenarios from both models at three watershed locations of increas-
moderate to high slope (also see Ivanov et al. 2004a), and (3) wetter
ing CA that are relevant to the Monterrey metropolitan area: (1) En-
areas in valley bottoms and mountain ridges [Fig. 8(b)] that reach
try to the City, (2) Site C, a confluence upstream of the main urban
saturation of the entire soil profile primarily produce saturation-
areas (Fig. 7), and (3) Cadereyta Station at the basin outlet. The
excess runoff. The ability to distinguish between these runoff
mechanisms can help select among different flood mitigation strat- scenario labeled Current Conditions refers to the calibrated flood
egies and identify key locations for the placement of hydraulic in- responses from the semidistributed and fully-distributed approaches
frastructure options. that contain the flood mitigation achieved by the Rompepicos Dam.
The favorable effects of the existing infrastructure are captured
in the scenario labeled No Dams, where Rompepicos Dam results
Impacts of Flood Mitigation Strategies in a reduction of the peak discharge of approximately 39% at
The authors assessed the impact of the current hydraulic infrastruc- the Entry to the City and 14% at Cadereyta Station, an indication
ture in the SCW and the potential for flood mitigation afforded by that the Hurricane Alex flood event would have been more severe
Fig. 10. (Color) Location of Rompepicos Dam and the hydraulic infrastructure scenarios with a large dam at the Entry to the City and three small
detention dams; for each site, the contributing area is shown within the context of the HEC-HMS subbasins and tRIBS stream network
(i.e., exceeding 5,000 m3 =s at the basin outlet) without the existing mitigation strategies using two hydrologic models of varying
hydraulic infrastructure. The two stakeholder-driven mitigation complexity but setup in a commensurate fashion using similar
strategies varied in their effectiveness upstream of the Monterrey underlying datasets. The modeling activities were the outcome of
metropolitan area (Entry to the City and Site C), whereas differences a stakeholder engagement process revealing that credibility in the
were reduced at the Cadereyta Station because of the contributions fully-distributed model could be established via comparisons to
from impervious urban areas downstream of the hydraulic infra- the semidistributed model that already had attained a level of
structure. The inclusion of a large dam at the Entry to the City had familiarity and trust by the participants. The model comparison also
the effect of decreasing and delaying the flood peak at all down- served the dual purposes of providing a detailed investigation of the
stream locations in both models because of its capacity to separate spatiotemporal flood response and the evaluation of mitigation
in time the contributions from upper areas (Units 1 and 2) and lower strategies in a watershed management context. The application
areas (Unit 3) to the overall flood wave. At the Cadereyta Station, a of the semidistributed and fully-distributed models was challenging
reduction of the peak discharge of 17% (tRIBS) and 23% (HEC- because of the sparse ground observations and the limited number
HMS) was obtained due to the proposed dam at the Entry to the of hydrologic modeling studies in México. In addition, the limited
City. Note that Hurricane Alex would have led to a water level ex- number of hydrologic observations (e.g., streamflow, soil moisture,
ceeding the main Creager spillway at the Entry to the City site (see groundwater depth) during Hurricane Alex prevented a more de-
sharp discharge rise in Large Dam scenario on July 2) if similar tailed set of model evaluation exercises, as performed when more
characteristics to the Rompepicos Dam were used in the design. As extensive data is available (c.f., Xiang et al. 2014; Robles-Morua
a comparison, inclusion of the small detention dams had a smaller et al. 2015; Mascaro et al. 2015). Nevertheless, the authors applied
mitigation effect at the three downstream locations, with a lower a commensurate set of spatiotemporal datasets, hydrologic process
peak reduction at the Cadereyta Station (12% in tRIBS and 10% representations, and calibration efforts in HEC-HMS and tRIBS to
in HEC-HMS) and a lower overall effect on the flood peak timing. enable a fair comparison of the two approaches and closely match
At Site C where the effects of all three structures can be deter- the limited set of hydrologic observations during Hurricane Alex
mined, the flood peak discharge was reduced by 17% (tRIBS) and
(see additional details in Cázares-Rodríguez 2016). For instance,
20% (HEC-HMS) relative to the current conditions. Thus, the stra-
both models utilized the same topographic and land cover datasets,
tegic placement of smaller hydraulic structures in regions with un-
employed a continuous representation of soil moisture dynamics,
controlled flows, as illustrated through the example of the three
and included a level-pool routing scheme to depict the effects of
small detention dams (see additional details in Cázares-Rodríguez
hydraulic infrastructure on the flood wave propagation. By ap-
2016), holds promise in collectively achieving an important degree
plying HEC-HMS with improved processes and datasets com-
of flood mitigation upstream of Monterrey as shown through this
mensurate with tRIBS, significant enhancements were made to the
model comparison.
application of the model for Hurricane Alex as compared to prior
efforts (Ramírez 2010, 2011). In addition, the overall good agree-
Summary and Conclusions ment between the two models, in particular with respect to the spa-
tial distribution of total runoff, the reservoir inflow and outflow
The authors analyzed a major flood event caused by Hurricane dynamics, and the cumulative discharge at the basin outlet, yielded
Alex in 2010 in Monterrey, México and quantified different flood a level of confidence in the ability of both models to be relevant for
Fig. 11. Comparison of flood responses from (a, c, and e) HEC-HMS and (b, d, and f) tRIBS for different hydraulic infrastructure scenarios at
(a and b) Entry to the City; (c and d) Site C, a confluence upstream of the urban area; (e and f) Basin Outlet at Cadereyta Station; the scenarios are
labeled No Dam (removal of Rompepicos Dam), Current Conditions (with Rompepicos Dam), and Detention Dams and Large Dam, both in addition
to Rompepicos Dam; the hourly basin-averaged precipitation from the bias-corrected NLDAS product is shown in all cases
watershed planning and management. Furthermore, the model Cadereyta Station for Hurricane Alex) is critical for mitigating
comparison showed that fully-distributed approaches add signifi- hurricane-induced flooding in the region; and
cant value to the representation of the basin hydrology and its prac- • Two stakeholder-driven flood mitigation strategies varied in ef-
tical use for infrastructure planning as summarized in the following: fectiveness upstream of Monterrey. A single, large dam at the
• Aggregation of landscape properties and meteorological for- Entry to the City reduced and delayed the flood peak more fa-
cing in semidistributed hydrologic models yields more signifi- vorably than three smaller detention dams placed in regions with
cant discrepancies in regions of complex terrain as compared to uncontrolled flow. The overall reduction of the flood peak at the
fully-distributed approaches. Specifically, HEC-HMS simula- basin outlet across all scenarios ranged from 10 to 23%, indi-
tions exhibited an earlier rising limb and a quicker recession limb cating the extreme events generated during hurricane landfall
in areas where the subbasin slope overestimated the actual slope are difficult to fully contain. Nevertheless, the strategic use of
distribution represented in tRIBS. Furthermore, the hydrologic small detention dams in concert with the existing infrastructure
response in semidistributed models is less sensitive to the effects holds promise as a collective measure to minimize flood recon-
of aggregation of impervious urban cover; struction efforts in Monterrey.
• The semidistributed and fully-distributed models allowed for a These results suggest that fully-distributed hydrologic models
spatiotemporal evaluation of the Hurricane Alex flood event, are amenable tools for watershed planning and management
resulting in new insights on the runoff generation sites and the purposes, including the evaluation of proposed hydraulic infra-
flood wave propagation. For instance, urban areas in Monterrey structure, that have distinct advantages over more commonly ap-
had a smaller flood contribution (42%) than the mountainous plied semidistributed modeling packages such as HEC-HMS.
basin (58%) to the flood volume at the basin outlet, resolving These advantages included: (1) an increased model resolution cap-
an important source of conflict among stakeholders participat- turing spatial details of soil moisture and runoff generation in
ing in the workshops. Similarly, the flood control derived from regions of complex terrain, (2) a higher fidelity of hydrologic proc-
the Rompepicos Dam (39% at the Entry to the City and 14% at esses allowing for the spatial characterization of runoff generation
Nestos Help Basin.” Water SA, 34(4), 461–467. Mays, L. W. (2010). Water resources engineering, Wiley, Hoboken, NJ.
Gutiérrez-López, A., and Ramírez, A. I. (2005). “Hydrological prediction Méndez-Barroso, L. A., et al. (2014). “A modeling approach reveals differ-
in two towns using the index-flood method.” Ingeniería Hidraúlica en ences in evapotranspiration and its partitioning in two semiarid ecosys-
México, 20(2), 37–47. tems in northwest México.” Water Resour. Res., 50(4), 3229–3252.
Hawkins, G. A., Vivoni, E. R., Robles-Morua, A., Mascaro, G., Rivera, E., Mitchell, K. E., et al. (2004). “The multi-institution North American land
and Dominguez, F. (2015). “A climate change projection for summer data assimilation system (NLDAS): Utilizing multiple GCIP products
hydrologic conditions in a semiarid watershed of central Arizona.” and partners in a continental distributed hydrological modeling system.”
J. Arid Environ., 118, 9–20. J. Geophys. Res., 109(D7), D07S90.
HEC (Hydrologic Engineering Center). (2013). HEC-GeoHMS geospatial Moreno, H. A., Vivoni, E. R., and Gochis, D. J. (2013). “Limits to flood
hydrologic modeling extension user’s manual, version 10.1, U.S. Army forecasting in the Colorado front range for two summer convection
Corps of Engineers, Davis, CA. periods using radar nowcasting and a distributed hydrologic model.”
HEC (Hydrologic Engineering Center). (2015). Hydrologic Modeling Sys- J. Hydrometeorol., 14(4), 1075–1097.
tem HEC-HMS Applications Guide, U.S. Army Corps of Engineers, Návar, J. (2012). “Modeling annual discharge of six México’s northern
Davis, CA. rivers.” Revista Ambiente Agua, 7(1), 36–50.
Hernández, A., and Bravo, G. C. (2010). “Reseña del Huracán Alex del Návar, J., and Synnott, T. J. (2000). “Soil infiltration and land use in
Océano Atlántico [Review of Hurricane Alex on the Atlantic Ocean].” Linares, N.L., México.” Terra Latinoamericana, 18(3), 255–262.
〈http://smn.cna.gob.mx/ciclones/tempo2010/atlantico/Alex2010a.pdf〉 Nied, M., Hundecha, Y., and Merz, B. (2013). “Flood-initiating catch-
(Jul. 1, 2016). ment conditions: A spatio-temporal analysis of large-scale soil mois-
ture patterns in the Elbe River basin.” Hydrol. Earth Syst. Sci., 17(4),
Hernández, M. A., Amador, A., Sánchez, S. T., and Solera, A. (2014).
1401–1414.
“Integrated water management in the Tambula-Picachos sub-basin,
Nikolopoulos, E. I., Anagnostou, E. N., Borga, M., Vivoni, E. R., and
Guanajuato, México.” Tecnología y Ciencias del Agua, 5(6), 159–165.
Papadopoulos, A. (2011). “Sensitivity of a mountain basin flash flood
INEGI (Instituto Nacional de Estadistica y Geografia). (1993). “Conjunto
to initial wetness condition and rainfall variability.” J. Hydrol., 402(3),
de datos vectorial uso de suelo y vegetación escala 1:250, 000 Serie II
165–178.
(continuo nacional). [Vectorial data set on land use and vegetation
Pasch, R. J. (2010). “Tropical cyclone report Hurricane Alex (AL012010)
1:250, 000 scale series II (national map).].” Aguascalientes, México.
25 June–2 July 2010.” 〈http://www.nhc.noaa.gov/data/tcr/AL012010
INEGI (Instituto Nacional de Estadistica y Geografia). (2009). “SCNM:
_Alex.pdf〉 (Jul. 1, 2016).
Sistema de cuentas nacionales de México: Producto interno bruto
Porse, E. (2014). “Risk-based zoning for urbanizing floodplains.” Water
por entidad federativa 2005–2009. [SCNM: National account system
Sci. Technol., 70(11), 1755–1763.
in México: Gross domestic product by federal entity 2005-2009.].”
Ramírez, A. (2010). “Estudio hidrológico de la cuenca del Río Santa
Aguascalientes, México.
Catarina, N.L. [Hydrologic study of the Santa Catarina River in the state
INEGI (Instituto Nacional de Estadistica y Geografia). (2011). “Population of Nuevo León].” Technical Rep. Commissioned to Integrate the
by county in Nuevo León state.” Aguascalientes, México. Reconstruction of the Santa Catarina River after Hurricane Alex,
ISRIC–World Soil Information (2013). “SoilGrids: An automated system Comisión Nacional del Agua (CONAGUA), Monterrey, México.
for global soil mapping.” 〈http://soilgrids1km.isric.org〉 (Jan. 15, 2014). Ramírez, A. (2011). “Evaluacion del funcionamiento de la Presa Rompe-
Ivanov, V. Y., Vivoni, E. R., Bras, R. L., and Entekhabi, D. (2004a). “Catch- picos [Evaluation of the Rompe Picos Dam performance].” Technical
ment hydrologic response with a fully distributed triangulated irregular Rep. Commissioned to Integrate the Reconstruction of the Santa
network model.” Water Resour. Res., 40(11), W11102. Catarina River after Hurricane Alex, Comisión Nacional del Agua
Ivanov, V. Y., Vivoni, E. R., Bras, R. L., and Entekhabi, D. (2004b). “Pre- (CONAGUA), Monterrey, México.
serving high resolution surface and rainfall data in operational-scale Robles-Morua, A., Che, D., Mayer, A. S., and Vivoni, E. R. (2015).
basin hydrology: A fully-distributed physically-based approach.” “Hydrological assessment of proposed reservoirs in the Sonora river
J. Hydrol., 298(1), 80–111. basin, México, under historical and future climate scenarios.” Hydrol.
Kampf, S. K., and Burges, S. J. (2007). “A framework for classifying and Sci. J., 60(1), 50–66.
comparing distributed hillslope and catchment hydrologic models.” Robles-Morua, A., Halvorsen, K., Mayer, A. S., and Vivoni, E. R. (2014).
Water Resour. Res., 43(5), W05423. “Exploring the application of participatory modeling approaches in the
Karamouz, M., Zahmatkesh, Z., Goharian, E., and Nazif, S. (2015). “Com- Sonora river basin, Mexico.” Environ. Model. Software, 52, 273–282.
bined impact of inland and coastal floods: Mapping knowledge base for Robles-Morua, A., Vivoni, E. R., and Mayer, A. S. (2012). “Distributed
development of planning strategies.” J. Water Resour. Plann. Manage., hydrologic modeling in northwest México reveals the links between
10.1061/(ASCE)WR.1943-5452.0000497, 04014098. runoff mechanisms and evapotranspiration.” J. Hydrometeorol., 13(3),
Kim, J., Ivanov, V. Y., and Katopodes, N. D. (2013). “Modeling erosion and 785–807.
sedimentation coupled with hydrological and overland flow processes Sánchez-Rodríguez, R., and Cavazos, T. (2015). “Amenazas naturales,
at the watershed scale.” Water Resour. Res., 49(9), 5134–5154. sociedad y desastres. [Natural threats, society and disasters.].” Conviv-
Kunkel, K. E., Pielke, R. A., and Changnon, S. A. (1999). “Temporal fluc- iendo con la Naturaleza, T. Cavazos, ed., ILCSA, Tijuana, México,
tuations in weather and climate extremes that cause economic and 4–49.
metropolitan area, México: Lessons from Hurricane Alex.” Water distributed hydrological model in a parallel computing environment.”
and cities in Latin America: Challenges for sustainable development, J. Hydrol., 409(1), 483–496.
I. Aguilar-Barajas, J. Mahlknecht, J. Kaledin, M. Kjellén, and A. Mejía- Vivoni, E. R., Entekhabi, D., Bras, R. L., and Ivanov, V. Y. (2007). “Con-
Betancourt, eds., Routledge, New York, 145–169. trols on runoff generation and scale-dependence in a distributed hydro-
Sivapalan, M., and Wood, E. F. (1986). “Spatial heterogeneity and scale in logic model.” Hydrol. Earth Syst. Sci., 11(5), 1683–1701.
the infiltration response of catchments.” Scale problems in hydrology, Vivoni, E. R., Ivanov, V. Y., Bras, R. L., and Entekhabi, D. (2004).
V. K. Gupta, I. Rodríguez-Iturbe, and E. F. Wood, eds., Springer, “Generation of triangulated irregular networks based on hydrological
Netherlands, 81–106. similarity.” J. Hydrol. Eng., 10.1061/(ASCE)1084-0699(2004)9:4(288),
Slutzman, J. E., and Smith, J. A. (2006). “Effects of flood control structures 288–302.
on flood response for Hurricane Floyd in the Brandywine Creek water- Vivoni, E. R., Ivanov, V. Y., Bras, R. L., and Entekhabi, D. (2005). “On the
shed, Pennsylvania.” J. Hydrol. Eng., 10.1061/(ASCE)1084-0699 effects of triangulated terrain resolution on distributed hydrologic
(2006)11:5(432), 432–441. model response.” Hydrol. Processes, 19(11), 2101–2122.
Smith, R. E., and Hebbert, R. H. B. (1979). “A Monte Carlo analysis of the Vivoni, E. R., Rodriguez, J. C., and Watts, C. J. (2010). “On the spatio-
hydrologic effects of spatial variability of infiltration.” Water Resour. temporal variability of soil moisture and evapotranspiration in a moun-
Res., 15(2), 419–429. tainous basin within the North American monsoon region.” Water
SMN (Servicio Meteorológico Nacional). (2010). “Normales clima- Resour. Res., 46(2), W02509.
tológicas por estado [Climatological compilations by state].” 〈http:// White, D. D., Wutich, A., Larson, K. L., Gober, P., Lant, T., and Senneville,
smn.cna.gob.mx/index.php?option=com_content&view=article&id=184& C. (2010). “Credibility, salience, and legitimacy of boundary objects:
tmpl=component〉 (Jan. 15, 2014). Water managers’ assessment of a simulation model in an immersive
Starkl, M., Brunner, N., López, E., and Martínez-Ruíz, J. L. (2013). “A decision theater.” Sci. Public Policy, 37(3), 219–232.
planning-oriented sustainability assessment framework for pen-urban Xiang, T. T., Vivoni, E. R., and Gochis, D. J. (2014). “Seasonal evolution
water management in developing countries.” Water Res., 47(20), of ecohydrological controls on land surface temperature over complex
7175–7183. terrain.” Water Resour. Res., 50(5), 3852–3874.