Prediction and Reanalysis of the Flood and High Precipitation Event Canmore, Alberta February 11, 2014 Bruce Davison Al Pietroniro Nick Kouwen Anthony Liu Muluneh Mekonnen 3/11/2014 Page 1 A Cautionary Tale • • • • WATFLOOD and MESH Precipitation analysis WATFLOOD runs MESH run With a Silver Lining 3/11/2014 Page 2 WATFLOOD and MESH CLASS WATFLOOD MESH Second Generation Land Surface Scheme MOISTURE EXCHANGE ra IR up Ponded Water α veg Interception Tc Canopy Layer 1 Transpiration Leaves LAI rs Fraction vegetation cover & LAI Fn ( temperature & season) Snow Interflow LZS Base Flow Percolation Temperature and humidity IR up UZS Stomata Layer 2 ( light sensitivity changes rs ) Wind Leaf Drip Overland Flow Infiltration Evaporation Aerodynamic resistance ( ra ) WATFLOOD CLASS RADIATIVE EXCHANGE SENSIBLE EXCHANGE IR down Stems SAI Infiltration α soil Surface Runoff Layer 3 Fn ( soil wetness) Tg Drainage Surface Soil Temperature Upper layer wetness Upper soil layer Full soil column Lower Soil Temperature Root Distribution Fn ( vegetation type) Tgb Full column wetness Subsurface Runoff Sub-grid Variability Grid Square Sellers et al. 1997 Computational Unit = GRU Computational Unit = Grid Square a CS Mixture of 4 sub-areas C GS c Individual Pixels b Mixture of 4 sub-areas G Kouwen et al. 1993 Blend of 5 vegetation groups Blend of 5 vegetation groups CS GS C G The approach - WATFLOOD • Use existing watershed Manitoba Hydro WATFLOOD model for the headwaters of the North & South Saskatchewan rivers • Convert CaPA precipitation and temperature data form its native format to Green Kenue (GK) r2c formats • Re-calibrate the model parameters for the CaPA met data for 2002 – 2009 • Model the 2013 Calgary flood Distributed precip vs. CaPA • First, the conventional gauge data was compared with the CaPA data • Conventional data is distributed with the WATFLOOD pre-processors where the both the precipitation and temperature lapse rates and the radius of influence of each gauge are included in the DDS parameter fitting exercise • The next slide compares CaPA with conventionally distributed gauge preciptation Basin total precipitation 2002 - 2005 Gauge precipitation in mm 6000 4000 2000 Fit Results Fit 1: Through origin Equation Y = 1.251690945 * X Number of data points used = 51 Average X = 2519.55 Average Y = 3149.16 Residual sum of squares = 2.39859E+006 Coef of determination, R-squared = 0.995551 Residual mean square, sigma-hat-sq'd = 47971.8 0 0 4000 2000 CaPA precipitation in mm 6000 • As with the 2002 – 2005 comparison the 2013 storm precipitation appears under estimated when compared to published rainfall amounts Published precipitation map http://en.wikipedia.org/wiki/2013_Alberta_floods Max. precip: over 300 mm CaPA precipitation map June 16 – June 30, 2013 Max. precip.: 221 mm Model calibration approach • The usual approach to calibrating WATFLOOD is to first obtain the proper overall volume using only those parameters that affect evapotranspiration, sublimation and lake evaporation • The nest step is to adjust the timing of the hydrographs • The next ten slides show the result of 10 Dynamically Dimensioned Search (DDS) runs with WATFLOOD sim_ sp l0 _ d observed sim_spl1_d sim_spl1_g sim_spl1_m sim_spl1_n 500 flow in cms 400 33 - 05BH004 BOW RIVER AT CALGARY 7868 km^2 300 200 100 2003 300 2004 2004 27 - 05BB001 BOW RIVER AT BANFF 2210 km^2 flow in cms 200 100 0 2003 2004 2004 sim_ sp l0 _g observed sim_spl2_d sim_spl2_g sim_spl2_m sim_spl2_n 500 flow in cms 400 33 - 05BH004 BOW RIVER AT CALGARY 7868 km^2 300 200 100 2003 300 2004 2004 27 - 05BB001 BOW RIVER AT BANFF 2210 km^2 flow in cms 200 100 0 2003 2004 2004 observed sim_spl2_d07 sim_spl2_g07 sim_spl2_m07 sim_spl2_n07 500 flow in cms 400 33 - 05BH004 BOW RIVER AT CALGARY 7868 km^2 300 200 100 2003 300 2004 2004 27 - 05BB001 BOW RIVER AT BANFF 2210 km^2 flow in cms 200 100 0 2003 2004 2004 Computed Flow Observed Flow 500 Dv = -11.4 % 33 - 05BH004 BOW RIVER AT CALGARY 7868 km^2 400 cms 300 200 100 2004 250 2005 2006 27 - 05BB001 BOW RIVER AT BANFF 2210 km^2 200 cms 150 Dv = -26.7 % 100 50 0 2004 2005 2006 Computed Flow Observed Flow Dv = -11.4 % 33 - 05BH004 BOW RIVER AT CALGARY 7868 km^2 800 cms 600 400 200 02 03 04 05 06 07 08 10 09 27 - 05BB001 BOW RIVER AT BANFF 2210 km^2 300 cms 200 Dv = -26.7 % 100 0 2002 2003 2004 2005 2006 2007 2008 2009 2010 Computed Flow Observed Flow Dv = -24.3 % 1000 33 - 05BH004 BOW RIVER AT CALGARY 7868 km^2 800 cms 600 400 200 10 11 250 12 13 14 27 - 05BB001 BOW RIVER AT BANFF 2210 km^2 200 cms 150 Dv = -42.2 % 100 50 0 2010 2011 2012 2013 2014 Computed Flow Observed Flow 2000 Dv = -1.9 % 33 - 05BH004 BOW RIVER AT CALGARY 7868 km^2 1600 cms 1200 800 400 2012 500 2013 2014 27 - 05BB001 BOW RIVER AT BANFF 2210 km^2 400 cms 300 Dv = -26.3 % 200 100 0 2012 2013 2014 Dv = -10.3 % Qlz SIM OBS 2000 33 - 05BH004 BOW RIVER AT CALGARY 7868 km^2 1600 cms 1200 800 400 2012 800 2013 2014 27 - 05BB001 BOW RIVER AT BANFF 2210 km^2 cms 600 400 Dv = -32.3 % 200 0 2012 2013 2014 Dv = -11.1 % Qlz SIM OBS 2000 33 - 05BH004 BOW RIVER AT CALGARY 7868 km^2 1600 cms 1200 800 400 2012 600 2013 2014 27 - 05BB001 BOW RIVER AT BANFF 2210 km^2 cms 400 Dv = -31.1 % 200 0 2012 2013 2014 The approach - MESH • Use existing watershed MESH model for the South Saskatchewan River • Convert CaPA precipitation and temperature data form its native format to Green Kenue (GK) r2c formats • Pre-calibrated the model parameters for the CaPA met data for Oct 2002 – Oct 2004 • Model the 2013 Calgary flood Standalone MESH run
© Copyright 2024 ExpyDoc
