Assessment of Floodplain Vulnerability during Extreme Mississippi

Feb 10, 2014 - Mississippi River. Our flow modeling and spatial mapping analysis attributes this vulnerability to a combination of erodible soils,...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/est

Assessment of Floodplain Vulnerability during Extreme Mississippi River Flood 2011 Allison E. Goodwell,† Zhenduo Zhu,† Debsunder Dutta,† Jonathan A. Greenberg,‡ Praveen Kumar,†,* Marcelo H. Garcia,† Bruce L. Rhoads,‡ Robert R. Holmes,§ Gary Parker,† David P. Berretta,∥ and Robert B. Jacobson⊥ †

Department of Civil and Environmental Engineering, University of Illinois at Urbana−Champaign, 205 North Mathews Avenue, Urbana, Illinois 61801-2352, ‡ Department of Geography, University of Illinois at Urbana−Champaign, 605 East Springfield Avenue Champaign, Illinois 61820, United States § U.S. Geological Survey, Office of Surface Water, ∥U.S. Army Corps of Engineers, Memphis District, and ⊥U.S. Geological Survey CERC, Columbia, Missouri 65201-9634, United States S Supporting Information *

ABSTRACT: Regional change in the variability and magnitude of flooding could be a major consequence of future global climate change. Extreme floods have the capacity to rapidly transform landscapes and expose landscape vulnerabilities through highly variable spatial patterns of inundation, erosion, and deposition. We use the historic activation of the Birds Point-New Madrid Floodway during the Mississippi and Ohio River Flooding of 2011 as a scientifically unique stress experiment to analyze indicators of floodplain vulnerability. We use pre- and postflood airborne Light Detection and Ranging data sets to locate erosional and depositional hotspots over the 540 km2 agricultural Floodway. While riparian vegetation between the river and the main levee breach likely prevented widespread deposition, localized scour and deposition occurred near the levee breaches. Eroded gullies nearly 1 km in length were observed at a low ridge of a relict meander scar of the Mississippi River. Our flow modeling and spatial mapping analysis attributes this vulnerability to a combination of erodible soils, flow acceleration associated with legacy fluvial landforms, and a lack of woody vegetation to anchor soil and enhance flow resistance. Results from this study could guide future mitigation and adaptation measures in cases of extreme flooding.

■

INTRODUCTION Extreme floods can transform landscapes, impact lives, and cause loss of property and livelihood. Because predicted changes in precipitation and climate patterns could affect regional flood frequency,1 it is imperative that we develop appropriate flood mitigation and adaptation measures.2 Recent studies have called for the urgent development of strategies for disaster resilience3,4 and the need to assess landscape vulnerability.5 Large floods, such as the extreme Mississippi and Ohio River Floods of Spring 2011,6 create highly variable spatial patterns of erosion and deposition.7,8 These localized areas, or “hotspots”, of change expose underlying landscape vulnerabilities and provide opportunities to retrospectively assess their causes in order to guide future actions. We treat this modern historical flood as a scientifically unique field-scale stress test,9 and analyze the indicating factors of hotspots of floodplain vulnerability to erosion and deposition. Erosion and deposition associated with overbank flow, lateral migration, and avulsive cutoffs shape the floodplains and control soil properties along large meandering rivers such as the © 2014 American Chemical Society

Lower Mississippi. In this region, these processes over time allow for the development of a highly productive agricultural landscape. To protect these valuable lands from flooding, humans have built levees that hydrologically disconnect them from the river. To manage extreme floods, that threaten nearby communities or the structural integrity of the levees, the modern arsenal of flood-control techniques includes intentional levee breaches or spillway activations in optimal locations.10,11 One such location is the Birds Point New Madrid (BPNM) Floodway, an agricultural region located west of the Mississippi River just south of its confluence with the Ohio River, near the city of Cairo, Illinois (Figure 1a,b). Current policy which was established following the disastrous 1927 Flood permits levees surrounding the BPNM Floodway to be intentionally breached during extreme floods to prevent upstream and downstream Received: Revised: Accepted: Published: 2619

October 25, 2013 February 4, 2014 February 10, 2014 February 10, 2014 dx.doi.org/10.1021/es404760t | Environ. Sci. Technol. 2014, 48, 2619−2625

Environmental Science & Technology

Article

Figure 1. BPNM Floodway Topography and Geomorphic Impacts. (a) Location of BPNM Floodway. (b) BPNM Floodway Digital Elevation Model (DEM) from 2005 Lidar survey showing inflow and outflow crevass locations, prominent meander scar ridges, locations of USGS sensors used for model validation, and elevation transect in (c). The USGS sensors, following from north to south have Identification numbers as follows: 911329, 2022416, 2022415, 2014656, 911320. Water levels at sensors are shown in SI Figure S2. (c) Floodway elevation transect, location shown in b. Levees and meander ridges are identified as sharp gradients. (d) differential Lidar DEM (2011−2005 elevations) showing erosion and deposition response of the north breach (A), O’Bryan Ridge gully erosion (B1, B2), and IFOF #1 near Big Oak Tree State Park (C).

into the floodplain widened and connected the breaches. Inflow/Outflow breach (IF/OF) #2 was activated on May 3. Inflow from the upstream breach had by this time created a hydraulic gradient such that floodwaters at the southern end of the Floodway flowed out of IF/OF #2 and the 1500 Ft. Gap. IF/OF #1 near Big Oak Tree State Park was activated on May 5 and acted as an inflow breach. Parts of the agricultural floodplain were inundated for over a month.13,14 Geomorphic impacts due to the levee breaches include localized scour holes, rills, and gullies, in addition to sand and gravel deposition and ripple fields indicating flow direction.15 Prior to the Floodway activation, the USGS (United States Geological Survey) placed numerous water-level sensors within

flooding.12 During the historic flood stages of May 2011, the Floodway was activated for the first time since 1937. The U.S. Army Corps of Engineers used blasting agents on May 2, 3 and 5, 2011 to create artificial breaches at three locations: the north (upstream) breach (Figure 1b, Figure 2a,b), Inflow/Outflow (IF/OF) #1, and IF/OF #2 (Figure 1b). The first levee was breached at 10:00 pm on May 2, 2011 at the northern end of the Floodway (Figure 1b). Prior to the first breach, the downstream portion of the floodplain was inundated by backwater flooding through an intentional gap in the levee at the downstream end known as the 1500 Ft. Gap (Figure 1b). The north levee section was breached at six locations along a 2 km stretch, and the force of water flowing 2620

dx.doi.org/10.1021/es404760t | Environ. Sci. Technol. 2014, 48, 2619−2625

Environmental Science & Technology

Article

Figure 2. Images showing the location of the first crevasse and erosion at O’Bryan Ridge. (a) Riparian and agricultural corridor between first crevasse and Mississippi and Ohio River confluence (2/29/2012). (b) Riparian corridor and north crevasse during May 2011 Floodway activation (5/7/ 2011). (c) IKONOS24 natural color composite image shows sediment plume emanating from O’Bryan Ridge gulley scours during 2011 activation.

frequencies created by flight lines (SI Figure S4). We also omitted known spurious erosion and deposition signatures in locations of wetlands, forest vegetation, and standing water from DEM change analysis (SI Figure S4d,e). We corroborated the lack of erosion or deposition in several of these areas through a field inspection on July 3, 2012. After these corrections, we used a LiDAR region outside the Floodway as a reference area to detect potential bias between the DEMs. We estimated a bias of 11 cm toward deposition in the differential DEM, and subtracted the bias. Finally, we applied an elevation change significance threshold of ±14 cm over the Floodway to exclude uncertainty associated with LiDAR vertical accuracy (SI: LiDAR Filtering and Analysis, SI Figure S5). The estimated net erosion over the Floodway is 4.9 million m3, equivalent to a depth of nearly 1 cm eroded over the entire Floodway. We used USGS flow and sediment data from upstream on the Mississippi River to obtain an estimate of 995 000 m3 of total sediment input to the main breach (SI Figure S6), corresponding to an average sediment concentration of approximately 0.2 kg/m3 during the activation period. The 7.6 m spatial resolution AVIRIS data used for woody vegetation mapping in this study was obtained by NASA in July 2011 after the floodwaters had receded from the region. See the SI for details on the acquisition, analysis, and classification of the AVIRIS and LiDAR data sets. HydroSed2D17,18 was run on a 200 m irregular grid over the entire BPNM Floodway. We initiated the simulation at the time

the Floodway that collected hourly data for the first 16 days after the breach (several sensor locations are indicated in Figure 1b Floodway map).16 The USGS also traversed breach openings in boats mounted with acoustic Doppler current profilers (ADCPs) to collect daily or twice-daily inflow and outflow measurements at the north breach, IFOF #1, IFOF #2, the 1500 Ft. Gap, and the Combined Outflow (consists of 1500 Ft. Gap and IFOF #2). These unprecedented observations along with pre- and postflood LiDAR data collected by USACE (U.S. Army Corps of Engineers) provide the basis for the modeling and analysis presented here. We first perform a 2-D model simulation of flow over the entire Floodway using the HydroSed2D model17 and validate against the USGS discharge and water-level data.16 We then perform analysis of the LiDAR data (SI: LiDAR Acquisition) to identify regions of erosion and deposition (SI: LiDAR Filtering and Analysis), which is combined with model simulations and data pertaining to soil properties and vegetation (SI: Soils and Vegetation) to identify factors that best explain the patterns of landscape vulnerability to erosion and deposition.

■

EXPERIMENT AND METHOD The 2005 and 2011 LiDAR data sets were collected for the U.S. Army Corps of Engineers (SI: LiDAR Acquisition). The differential DEM (2005 elevations subtracted from 2011 elevations) exhibited flight line errors, which we corrected using a Fourier filtering technique that removed dominant 2621

dx.doi.org/10.1021/es404760t | Environ. Sci. Technol. 2014, 48, 2619−2625

Environmental Science & Technology

Article

Figure 3. Floodway Vulnerability Map. (a) Vulnerability Map of BPNM Floodway combining flow velocities, natural vegetation, and soil K/T erosion factor. The gray overtones on the color map of flow velocities indicate areas of higher vulnerability due to high K/T factor and high flow velocities. (b) O’Bryan Ridge vulnerability map overlaid by observed gully erosion. (c) Ten Mile Pond vulnerability map, where no observed erosion or deposition near the meander scar could be detected through LiDAR analysis or field observations.

■

of the first levee breach at the north levee and simulated the first 60 h after the breach (SI Figure S1). Initial and boundary conditions at inflows and outflows were based on USGS watersurface elevation measurements during the Floodway activation. Woody vegetation was incorporated by modifying the Manning’s n coefficient between 0.035 and 0.2 (SI Figure S2, S3). The model was validated using USGS flow and watersurface elevation measurements from the Floodway and used to calculate shear stresses (SI: 2D Flow Model, Figure S1). We calculated ratios of dimensionless shear stress to the critical shear stress for incipient sediment motion (SI: Equations). See the SI for details of the flow modeling and calculation of the dimensionless stresses and conditions for sediment motion. We obtained soil data from the Natural Resources Conservation Service (NRCS) Soil Data Mart19 for Mississippi and New Madrid Counties in Missouri. We used the ratio K/T to assess landscape sensitivity to erosion, where the soil K Factor is a measure of soil erodibility used in the Revised Universal Soil Loss Equation (RUSLE)20,21 and the soil T Factor (tons/acre/year) is a measure of sustainable or tolerable erosion for agricultural productivity.22 The flow model results and the soil and vegetation maps were used to spatially map vulnerability to erosion due to flooding. We compare these maps to observed impacts from the LiDAR.

RESULTS AND DISCUSSION Erosion and Deposition Patterns. The maximum inflow rate through the 2.8 km wide north breach on May 3, 2011, as measured by the U.S. Geological Survey, was 11 400 m3/s.16 Simulations with the HydroSed2D flow model17 over the Floodway captured this flow rate and the flood stage, especially when we account for resistance to flow by vegetation (SI Figure S2c,d). Water spilling through the upstream breach carved several scour holes up to 2 m deep and 60 m long, depositing eroded sediment up to 400 m downstream (differential LiDAR shown in Figure 1d(A)). Inflow into IF/OF #1 generated a 250 m long scour hole and a 50 000 m2 area of deposition that was 0.5 to 1.5 m thick north of the scour hole (Figure 1d(C)). The most prominent erosional features were 1 km long and 3 m deep gullies excavated at O’Bryan Ridge, approximately 13 km downstream of the upstream breach (Figure 1d(B1, B2)). At this location, floodwaters cascaded over the outer bank of a relict meander bend oriented perpendicular to the flow (Figure 2c). Within the Floodway, O’Bryan Ridge and similar ridges along the outer banks of relict meander scars formed over the past 6000 years23 remain the principal areas of steep gradients (2−3%) (Figure 1c). As erosion patterns demonstrate, these relict geomorphic features can strongly influence floodplain responses to contemporary flood events. After applying the ±14 cm threshold to the differential LiDAR, we determined that erosion and deposition occurred 2622

dx.doi.org/10.1021/es404760t | Environ. Sci. Technol. 2014, 48, 2619−2625

Environmental Science & Technology

Article

over 12% and 8% of the Floodway, respectively, with corresponding average depths of 0.27 m and 0.3 m. The LiDAR data and field excursions also demonstrate the effect of vegetation on erosional and depositional impacts at O’Bryan Ridge (SI Figure S7). For example, one small patch of land covered by dense vegetation located just upstream (north) of the highest gradient of the ridge was not eroded, while surrounding areas were scoured up to depths of 2 m. Vulnerability to Erosion. We designate a landscape as vulnerable if potential erosion or deposition over some time horizon is likely to decrease its functional value without human intervention. We used the following framework to characterize vulnerability to erosion. Let η (x,y,t) represent the elevation at any point (x,y) and time t. The rate of change of elevation may be written as (∂η(x,y,t)/∂t) = f (η,I,̅ Θ̅ ) where I ̅ and Θ̅ are vectors of input drivers (e.g., flow velocities, shear stresses) and land-surface characteristics (e.g., soil erodibility, slope, vegetation cover, etc.), respectively. We can quantify the total volume change Δ over an area A and time period T as Δ(A,T) = ∫ A ∫ T (∂η/∂t) dAdt. Vulnerability characterizes the expectation of the occurrence of a large Δ, through erosion or deposition, such that it impacts the agricultural or other utility of the landscape. The dominant functional value in the Floodway is row-crop agricultural production. For example, in the Fall of 2012, farmers had to regrade land dissected by large gullies at O’Bryan Ridge to restore the land to agricultural use.27 Landscape functional values could also be defined for various ecosystem services that could accrue under floodplain restoration scenarios.25,26 Factors that affect Δ due to erosion include (a) exposure to high velocities and shear stresses of floodwaters that can cause severe erosion; and (b) landscape sensitivity, or the capacity of soils to resist28 erosion by floodwaters29 that depends on soil erodibility and vegetation cover. Although developed for rill-dominated systems and not deep gullies that formed in the eroding floodplain, we use K and T as broad measures of sensitivity. A high K/T Factor indicates high sensitivity due to high erodibility and/or low erosion tolerance. On the basis of available data, we choose the following as indicators of vulnerability: K/T, the presence or absence of woody vegetation, flow velocities, and the ratio of dimensionless shear stresses to critical shear stresses (τb*/τc*). To assess the heterogeneous nature of vulnerability to erosion within the Floodway, we combined vulnerability indicators to create maps for the entire Floodway. We focus specifically on two relict meander scars, Ten Mile Pond and O’Bryan Ridge, which exhibited different responses to the 2011 event (Figure 3). Woody vegetation, most of which is located within meander scars in the southern low-lying area, covers about 16% of the Floodway. The extents of woody vegetation at Ten Mile Pond and O’Bryan Ridge are 24% and 8% of their land areas, respectively. The soil K and T Factors range from 0.24 to 0.55 and 3 to 5 tons/acre/year, respectively, within the Floodway. The Ten Mile Pond mean K/T factor is 32% larger than that at O’Bryan Ridge (Figure 4a), while both locations exhibit above average (>0.074) K/T Factors just upstream of the meander scar ridges (indicated by gray shading in Figure 3). Simulated depth-averaged flow velocities are 51% higher near O’Bryan Ridge than at the Ten Mile Pond (Figure 4b) because of higher gradients, a 50% narrower floodplain width, and lower roughness due to the lack of vegetation (SI Figure S3). The relatively low flow velocities (0.074) and below average (1) and low ( 1) and low vulnerabilities.

erosion at this location based on analysis of LiDAR data (SI Figure S4e). At O’Bryan Ridge, erosion depths are higher for grid cells that have above average (>0.074) K/T factor than for those with below average K/T (erosion depth frequency histogram is shown in Figure 4c). As erosion depth increases, the probability that a given cell has above average K/T increases (Figure 4d). Similarly, large erosion depths are preferentially associated with high (>1, indicating incipient motion) values of τb*/τc* (Figure 2623

dx.doi.org/10.1021/es404760t | Environ. Sci. Technol. 2014, 48, 2619−2625

Environmental Science & Technology

Article

4e,f). When the effects of K/T and τ*b /τ*c are combined, the greatest erosion depths are associated with grid points that have both above average K/T Factor and high τb*/τc* (Figure 4g,h) compared to cases where one or both of these factors are low. These results show that, in addition to legacy geomorphic features, soil and vegetation characteristics also dictate vulnerability to erosion. Model simulations that distinguish between woody vegetation and agricultural fields show decreased flow velocities within and downstream of forest patches near Ten Mile Pond and O’Bryan Ridge compared to simulations omitting vegetation (SI Figure S3). At O’Bryan Ridge, gully scouring occurred adjacent to several vegetated patches, while the forested patches and areas downstream of them were protected (Figure 3b map shows relation between vegetation, flow, and erosion at O’Bryan Ridge). Vegetated areas were found to be up to 1 m higher than the surrounding fields at the meander scar ridge, indicating a potential history of soil erosion due to agriculture.30 One of these small vegetated patches located just upstream of the high gradient at the ridge mitigated upstream and downstream gully erosion (photos and LiDAR at patch shown close up in SI Figure S7). Vulnerability to Deposition. Deposition in the Floodway was highly localized, occurring mainly near levee-breach scours and downstream of the gullies and vegetated patches at O’Bryan Ridge. Spatially, these deposition hotspots are closely linked to erosional hotspots. We contrast the lack of widespread deposition within the BPNM Floodway to two well-studied events: (1) major sand accumulation in the Bonnet Carré (BC) Spillway during the same 2011 event11 and (2) substantial deposition at some levee breaches during the 1993 Mississippi River Flood. The BC Spillway, near New Orleans, Louisiana, was activated by the USACE during the May 2011 Flood. Although only the top 5 m of river flow entered the spillway, nearly 5 million m3 of sand was deposited over its area of less than 25 km2. The high deposition rate in the spillway is attributed to a direct sediment flux toward its opening, which is located along the inner bank of a bend downstream of the bend apex where sediment concentrations are high and secondary flows are directed inward. In contrast to this optimal location for sand deposition, the upstream levee of the BPNM Floodway is separated from the river by a 2 km strip of floodplain that consists of agricultural fields and riparian vegetation (Figure 2a,b). Regardless of the effect of this corridor, sediment concentration data (SI Figure S6) indicates that less than 1 million m3 of sediment likely entered the Floodway. We can also contrast the 2011 BPNM Floodway deposition to impacts of the Mississippi River Flood of 1993, which caused over 1000 breaches31 and exposed landscape sensitivities upstream of the Floodway.32,33 The 1993 Miller City levee-break complex, located several km upstream of the BPNM Floodway, was directly adjacent to the Mississippi River, resulting in extensive sand deposits over 4 m thick and net deposition of 8.2 million m3 of sediment on the floodplain.31 In this case, proximity to the river, in addition to uncontrolled breach conditions, contributed to high deposition rates. In the 2011 BPNM Floodway activation, low river sediment load, controlled wide breach conditions, riparian woody vegetation, and a lack of river-Floodway connectivity limited deposition. Implications. This vulnerability analysis of the BPNM Floodway indicates that topographic legacies related to floodplain-shaping geomorphological processes, such as meander scar ridges, dictate landscape characteristics including gradient, floodplain width, and local elevation. These character-

istics, in turn, impact the likelihood of erosion and deposition during flood events. On a shorter time scale, land use and land cover characteristics also impact landscape response to flooding. This study uses an intentional levee breach and subsequent inundation as a unique experiment to retrospectively analyze drivers of vulnerability to erosion and deposition in an agricultural floodplain. In the BPNM Floodway, the development of gullies during the flood corresponded to erodible soils, lack of forest vegetation, and high simulated flow velocities at a legacy meander scar ridge. Another ridge farther downstream was not eroded because an increase in floodplain width along with dense forest vegetation diminished flow velocities and shear stresses. While these small forested regions reduced vulnerability to erosion inside the Floodway, a corridor of riparian vegetation between the river and Floodway limited vulnerability to deposition. This riparian corridor reduced connectivity between the river and Floodway and prevented river sediment load from entering. Quantitative measures of the indicating factors of vulnerability allowed us to create a vulnerability map for the entire landscape. Building resilience to natural disasters such as extreme flooding is of regional, national, and global concern. The results of this study suggest that restoration of large-river floodplains not only benefits natural ecosystems, but also potentially mitigates flood hazard and risk, increasing resilience to future extreme floods.34−36 Although predictive models of regional climate change are uncertain due to complex feedbacks, a global increase in flooding will depend on the extent of climate warming. Potential changes in flood variability due to climate change or land use practices emphasize the need to develop methods to assess resilience and vulnerability. Vulnerability assessments like the one presented in this study could inform future flood control and restoration strategies aimed at mitigating erosion and deposition of floodplain landscapes. Such assessments can be used to identify regions of high vulnerability in diverse landscapes prone to potentially catastrophic landscape change in future extreme floods.

■

ASSOCIATED CONTENT

S Supporting Information *

Details on LiDAR and AVIRIS acquisition and processing, flow modeling, sediment estimates, and soil and vegetation data. This material is available free of charge via the Internet at http://pubs.acs.org/.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: (217) 333-4688; fax: (217) 333-0687; e-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We gratefully acknowledge NSF RAPID Grant No. EAR 1140198 and AVIRIS Data Collection supported by NASA. The USGS provided flow and depth data used for model validation. The USACE provided the processed Lidar 2005 and 2011 DEMs through a Memorandum of Understanding with the University of Illinois. The USACE and USGS both provided insights into the levee breach operation and impacts. A.G. was supported by the Carver and SURGE Fellowships at 2624

dx.doi.org/10.1021/es404760t | Environ. Sci. Technol. 2014, 48, 2619−2625

Environmental Science & Technology

Article

(20) Renard, K.; Foster, G.; Weesies, G.; Porter, J. RUSLERevised Universal Soil Loss Equation. J. Soil Water Conserv. 1991, 46, 30−33. (21) Neitsch, S.; Arnold, J.; Kiniry, J.; Williams, J.; King, K. Soil and Water Assessment Tool: Theoretical Documentation, version 2005; Texas, U.S.A., 2005, (22) McCormack, D.; Young, K.; Kimberlin, L. Current criteria for determining soil loss tolerance. Determin. Soil Loss Tolerance 1982, 95−111. (23) Fisk, H. Geological Investigation of the Alluvial Valley of the Lower Mississippi River; War Department, U.S. Army Corps of Engineers, 1944. (24) Global Land Cover Facility, IKONOS Imagery; http://glcf.umd. edu/data/ikonos/, 2013. (25) Faulkner, S.; Barrow, W., Jr.; Keeland, B.; Walls, S.; Telesco, D. Effects of conservation practices on wetland ecosystem services in the Mississippi Alluvial Valley. Ecol. App. 2011, 21, S31−S48. (26) Jenkins, W. A.; Murray, B. C.; Kramer, R. A.; Faulkner, S. P. Valuing ecosystem services from wetlands restoration in the Mississippi Alluvial Valley. Ecol. Econ. 2010, 69, 1051−1061. (27) Olson, K.; Morton, L. Restoration of 2011 flood-damaged Birds Point−New Madrid Floodway. J. Soil Water Conserv. 2013, 68, 13A− 18A. (28) Luers, A. The surface of vulnerability: An analytical framework for examining environmental change. Global Environ. Change−Human Policy Dimens. 2005, 15, 214−223. (29) Brunsden, D. A critical assessment of the sensitivity concept in geomorphology. Catena 2001, 42, 99−123. (30) Knox, J. Agricultural influence on landscape sensitivity in the Upper Mississippi River Valley. Catena 2001, 42, 193−224. (31) Jacobson, R.; Oberg, K. Geomorphic Changes on the Mississippi River Flood Plain at Miller City, Illinois, as a Result of the Flood of 1993; USGS, 1997; Vol. 1120. (32) Gomez, B.; Mertes, L.; Phillips, J.; Magilligan, F.; James, L. Sediment characteristics of an extreme flood 1993 Upper Mississippi River Valley. Geology 1995, 23, 963−966. (33) Gomez, B.; Phillips, J.; Magilligan, F.; James, L. Floodplain sedimentation and sensitivity: Summer 1993 flood, upper Mississippi River valley. Earth Surf. Proc. Landforms 1997, 22, 923−936. (34) Opperman, J.; Galloway, G.; Fargione, J.; Mount, J.; Richter, B.; Secchi, S. Sustainable floodplains through large-scale reconnection to rivers. Science 2009, 326, 1487−1488. (35) Fokkens, B. The Dutch strategy for safety and river flood prevention. Extreme Hydrological Events: New Concepts for Security; Springer: New York, 2007; pp 337−352. (36) Sparks, R. E. Need for ecosystem management of large rivers and their floodplains. BioScience 1995, 45, 168−182.

the University of Illinois. D.D. was supported by Ravindar K and Kavita Kinra Fellowship. Z.Z. and M.H.G. acknowledge the support from Chester and Helen Siess Professorship. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

■

REFERENCES

(1) Kumar, P. HYDROLOGY Seasonal rain changes. Nat. Clim. Change 2013, 3, 783−784. (2) Hirabayashi, Y.; Roobavannan, M.; Koirala, S.; Yamazaki, D.; Watanabe, S.; Kim, H.; Kanae, S. Global flood risk under climate change. Nat. Clim. Change 2013, 3, 816−821. (3) National Research Council Committee on Increasing National Resilience to Hazards and Disasters, Disaster Resilience: A National Imperative. 2012, 203. (4) National Research Council Committee on Challenges and Opportunities in Earth Surface Processes, Launching a National Conversation on Disaster Resilience in America: Workshop Summary; 2013, 57. (5) National Research Council Committee on Challenges and Opportunities in Earth Surface Processes, Landscapes on the Edge: New Horizons for Research on Earth’s Surface; 2010, 180. (6) Mississippi River & Tributaries System 2011 Post Flood Report; U.S. Army Corps of Engineers: Mississippi Valley Division. (7) Croke, J.; Todd, P.; Thompson, C.; Watson, F.; Denham, R.; Khanal, G. The use of multi temporal LiDAR to assess basin-scale erosion and deposition following the catastrophic January 2011 Lockyer flood, {SE} Queensland, Australia. Geomorphology 2013, 184, 111−126. (8) Falcini, F.; Khan, N. S.; Macelloni, L.; Horton, B. P.; Lutken, C. B.; McKee, K. L.; Santoleri, R.; Colella, S.; Li, C.; Volpe, G.; D’Emidio, M.; Salusti, A.; Jerolmack, D. J. Linking the historic 2011 Mississippi River flood to coastal wetland sedimentation. Nat. Geosci. 2012, 5, 803−807. (9) Brown, C.; Wilby, R. L. An alternate approach to assessing climate risks. EOS, Trans. Am. Geophys. Union 2012, 93, 401−402. (10) Jaffe, D.; Sanders, B. Engineered levee breaches for flood mitigation. J. Hydraulic Eng.-ASCE 2001, 127, 471−479. (11) Nittrouer, J.; Best, J.; Brantley, C.; Cash, R.; Czapiga, M.; Kumar, P.; Parker, G. Mitigating land loss in coastal Louisiana by controlled diversion of Mississippi River sand. Nat. Geosci. 2012, 5, 534−537. (12) Mississippi River Commission, The Mississippi River and Tributaries Project: Birds Point-New Madrid Floodway Information Paper; U.S. Army Corps of Engineers, 2007. (13) Olson, K.; Morton, L. The impacts of 2011 induced levee breaches on agricultural lands of Mississippi River Valley. J. Soil Water Conserv. 2012, 67, 5A−10A. (14) Olson, K.; Morton, L. The effects of 2011 Ohio and Mississippi river valley flooding on Cairo, Illinois, area. J. Soil Water Conserv. 2012, 67, 42A−46A. (15) Londoño, A. C.; Hart, M. L. Landscape response to the intentional use of the Birds Point New Madrid Floodway on May 3, 2011. J. Hydrol. 2013, (16) Koenig, T.; Holmes, R. J. Documenting the stages and streamflows associated with the 2011 activation of the New Madrid Floodway, Missouri. U.S. Geological Survey Professional Paper 1798-E 2013, 31. (17) Liu, X.; Landry, B.; Garcia, M. Two-dimensional scour simulations based on coupled model of shallow water equations and sediment transport on unstructured meshes. Coastal Eng. 2008, 55, 800−810. (18) Liu, X. HydroSed2D: A Two-Dimensional Numerical Model for Hydrodynamics and Sediment Transport, Version 1.0; University of Illinois at Urbana-Champaign, 2009. (19) NRCS. Soil Data Mart; Natural Resources Conservation Service, 2012. 2625

dx.doi.org/10.1021/es404760t | Environ. Sci. Technol. 2014, 48, 2619−2625