Modeling Cumulative LID Features for Floodplain Impacts in an Urban Watershed in Houston, TX Christina Hughes, Phil Bedient cmhughes@rice.edu May 2, 2014 1
Modeling Cumulative LID Features for Floodplain Impacts in an Urban Watershed in Houston, TX Christina Hughes, Phil Bedient cmhughes@rice.edu May 2, 2014 2
Research Questions How can LID be best represented in a hydrologic model? To what degree can spatial resolution enhance LID modeling capabilities? Is there an appreciable difference in the hydrologic effect of modeling LID as a weighted change in parameters? What cumulative impact on water quantity can small-scale LID features have when implemented across a large area? What combinations of LID features are most effective? 3
Background - LID in Houston The Woodlands - 1972 LID Design Competition - 2010 Harris County LID Design Criteria - 2011 Bagby St. receives national recognition 2013 COH Cottage Grove pilot project - 2014 Limitations: Convention Quantity vs quality Limited space in urban environment Unique topography and climate 4
Bagby Street Bioretention/Rain garden Greenroads Certified Development Project Series of rain gardens treatment train Land Use: Commercial/Residential 5
Rain Barrels Capacity: varies (100-350 gal) Less storage than cisterns Easily installed at downspouts Individual initiative No current estimate on use Land Use: Residential 6
Distributed Modeling Allows for more spatial variation than lumped Higher resolution at large scale Gridded data availability Rainfall Urban Reservoir Channel Riparian Successfully used to model Houston area watersheds: Doubleday et al. 2013 Teague et al. 2011 Duncan et al. 2011 Fang et al. 2010 7
Watershed Modeling Lumped Model vs. Distributed Model Uniform parameters 8
LID Modeling in The Woodlands, TX Simulated existing LID features by changing roughness, imperviousness, adding detention Compared to undeveloped and theoretical conventional urbanization Significant impact of watershed-scale LID >100% peak and volume reduction for 100-year storm Riparian corridors Large scale detention Roughness associated with land cover 9 Source: Doubleday et al. 2013
White Oak Bayou 111 square mile watershed 151 miles open streams Harris County Urban 90% Developed (2011) 18% High 39% Medium 23% Low 10 Source: HCFCD; NLCD 2011
White Oak Bayou HEC RAS 10 year Floodplain 11
Vflo Model Elevation LiDAR 2001 Land cover NLCD 2011 Soil NRCS 2003 Land Use - HGAC 2011 Channels & gauges Land Use Land Cover Resolution: 50,000 cells Cell: 2.8 acres (350 ft x 350 ft) Soil DEM 12
Vflo Model Channel Overland 13
Model Performance and Calibration (1) Hurricane Ike 2008 10-15 in; 48 hrs (2) July 2012 7-9 in; 168 hrs (3) September 2013 5 in; 24 hrs (1) (2) Parameters adjusted: overland roughness channel roughness hydraulic conductivity soil depth initial saturation (3) 14
Modeling LID Rain gardens residential Increased roughness and infiltration Rain barrels residential Pure abstraction Green roofs public buildings Increased roughness and infiltration Decreased impervious surface Bioswales high-traffic roadways Increased roughness and infiltration Decreased impervious surface Design Storms 10 yr, 2 yr 41% Residential land use 15
Methodology - Residential LID Assumptions: 1 home = 2000 sq ft building with 2000 sq ft yard 30.6 homes per 2.8 acre cell 100% of homes have one rain garden 100% of homes use one rain barrel Home No change in imperviousness assume features built in existing pervious area 16 Yard/Road
Methodology Representing LID Features Weighted by area Rain garden: 225 sq ft area (5 ft x 45 ft) 3 ft depth, sandy loam n = 0.4 (McCuen 2005) No detention Modeled by increased roughness and Green & Ampt parameters Rain barrel: Rain barrel 350 gallons per home 2000 sq ft of roof area captured Modeled as 1.55 in abstraction for entire cell Roof Rain Garden 17
Results 7.6 in/24 hr (10 yr) 4% reduction in peak 2% reduction in volume 30000 25000 White Oak Outlet - 10 Year Event Current: 29200 cfs Rain Garden: 28900 cfs Combined: 28100 cfs 20000 Discharge (cfs) 15000 10000 5000 18 0 6/1/2007 0:00 6/2/2007 12:00 6/4/2007 0:00 Time Current Conditions Rain Gardens Rain Gardens and Rain Barrels
Results 7.6 in/24 hr (10 yr) 14000 12000 Upper White Oak - 10 Year Event Current:13700 cfs Rain Garden: 13400 cfs Combined: 13000 cfs 5% reduction in peak 3% reduction in volume 10000 Discharge (cfs) 8000 6000 4000 2000 0 6/1/2007 0:00 6/2/2007 0:00 6/3/2007 0:00 6/4/2007 0:00 Time Current Conditions Rain Gardens Rain Gardens and Rain Barrels 19
Results 4.4 in/24 hr (2 yr) White Oak Outlet - 2 Year Event 10% reduction in peak 5% reduction in volume 14000 12000 Current: 14800 cfs Rain Garden: 14400 cfs Combined: 13200 cfs 10000 Discharge (cfs) 8000 6000 4000 2000 0 4/29/2014 0:00 4/30/2014 0:00 5/1/2014 0:00 Time 20 Current Conditions Rain Gardens Rain Gardens and Rain Barrels
7000 6000 Results 4.4 in/24 hr (2 yr) Upper White Oak - 2 Year Event Current: 6700 cfs Rain Garden: 6400 cfs Combined: 5900 cfs 12% reduction in peak 8% reduction in volume 5000 Discharge (cfs) 4000 3000 2000 1000 0 4/29/2014 0:00 4/29/2014 12:00 4/30/2014 0:00 4/30/2014 12:00 5/1/2014 0:00 Time Current conditions Rain Gardens Rain Gardens and Rain Barrels 21
Observation Largest reduction upstream Higher density of LID features White Oak Bayou Total: 41% Residential Upper White Oak Bayou: 49% Residential Highest LID density = Little White Oak Bayou (59% res.) Lowest LID density = Cole Creek (14% residential) Cole Creek Little White Oak Bayou 22
Cole Creek - 2 Year Event Little White Oak - 2 Year Event 1400 4500 1200 Current: 1400 cfs Rain Garden: 1400 cfs Combined: 1390 cfs 4000 3500 Current:4400 cfs Rain Garden: 4300 cfs Combined: 3800 cfs 1000 3000 Discharge (cfs) 800 600 Discharge (cfs) 2500 2000 400 1500 1000 200 500 0 4/29/2014 0:00 4/30/2014 0:00 5/1/2014 0:00 Time 0 4/29/2014 0:00 4/30/2014 0:00 5/1/2014 0:00 Time Current Conditions Rain Gardens Rain Gardens and Rain Barrels 1% reduction in peak 2% reduction in volume 14% reduction in peak 9% reduction in volume 23
Conclusions Fully distributed model useful for representation of individual LID features across large area Higher density LID leads to a greater cumulative effect on downstream flows Up to 9% volume reduction and 14% peak reduction for 2 yr storm using only rain gardens and rain barrels LID feature size and scale has appreciable effect on stormwater flows Large-scale features (Woodlands) more effective for large storms 100 year storm wash out small-scale LID features Conservative estimation of an optimistic scenario Future work/ in progress: Combination of LID features and comparison of impact (green roofs, bioswales) Impervious to pervious area routing and interception via higher resolution nested subareas Floodplain impact analysis 350 ft 350 ft 24