geomorphic modeling and routing improvements for gis-based watershed assessment in arid regions...
TRANSCRIPT
-
Geomorphic Modeling and Routing Improvements for GIS-Based Watershed Assessment in Arid RegionsDarius J. SemmensPh.D. Candidate, Watershed ManagementMarch 5, 2004
-
AcknowledgementsUSDA-ARS Southwest Watershed Research CenterDavid Goodrich, Scott Miller, Carl UnkrichUSGS Waite OsterkampU of AZPhil Guertin, Richard Hawkins, Vicente Lopes, Craig WisslerU.S. EPA, Landscape Ecology BranchBill Kepner, Bruce JonesBetsy Semmens
-
IntroductionHydrologic and geomorphic systems are defined and linked by the movement of water on the Earths surface Management and planning for land and water resources is facilitated by watershed modelsRecent improvements to watershed models have been primarily focused on humid environmentsArid regions characterized by processes operating at different temporal and spatial scales, thus require specialized conceptual modelsThis research addresses two limitations of arid-region distributed watershed models that hinder their use as assessment and planning toolsLimitation of spatial scale (hydrologic model)Inability to simulate geomorphic response to landscape change (geomorphic model)
-
Problem Statement Hydrologic ModelSmall-watershed models designed to simulate short-duration ephemeral flowsPerformance declines when applied to areas larger than about 100 km2Large-watershed models simulate longer-term water balancePerformance declines when applied to areas smaller than about 1,000 km2 Ephemeral runoff in medium-sized arid-region watersheds is best described by small-watershed modelsModifications are needed to improve the performance of small-watershed models at larger scales
-
Problem Statement Geomorphic ModelTo understand how an individual stream reach responds to external stresses it is necessary to study the channel network as a whole Geomorphic watershed models are thus necessary to evaluate long-term (years) impacts of landscape changeEvent-based watershed models simulate erosion and deposition based on assumption that channel geometry is static during the course of an event Prevents simulation of cumulative impacts from multiple eventsNo event-based watershed models that track cumulative adjustment of the channel network in terms of channel width, depth, and slope.
-
Identifying the Scale Gap in Watershed ModelingUrbanizationBMP implementationRange of characteristic space time scalesSmall WS Models (e.g. KINEROS2)Large WS Models (e.g. SWAT)EcosystemrestorationIntermediate-Scale WS Models (This Research)From: Bloschl and Sivapalan (1995)
-
Study AreaUSDA-ARS Walnut Gulch Experimental Watershed Semi-arid rangelandDesert scrub (brush) and grassland~150 km2Rainfall and runoff measured by a network of recording rain gauges, flumes, and weirs
-
nested subwatersheds and measuring devicesPrimary Watershed AreasLH104 (0.047 km2)
-
Hydrologic ModelHypothesis A significant source of error at intermediate scales results from the inability to account directly for diffusion of the flood wave as it is routed through the channel network ApproachImplement Variable Parameter Muskingum-Cunge (VPMC) routing in KINEROS2 (Smith et al., 1995)Compare with kinematic routing at multiple scales
-
MVPMC4 RoutingModified variable parameter Muskingum-Cunge routing with an iterative 4-point solution (Ponce and Chaganti, 1994) Nonlinear coefficient method that accounts for hydrograph diffusion based on the physical properties of the channel and flowMatches physical and numerical diffusionDownstream boundary condition not requiredCannot simulate backwater effects
-
Testing Hydrologic ModelDesign Event Simulations30-minute and 1-hour events with 5, 10, and 100-year return periodsWatershed areas of 4.7, 782, 9558, and 14664 haComparison of kinematic (K2) and MVPMC4 (K2MC) simulated hydrographs in terms ofOnset and peak timingPeak discharge Runoff volumeComparison Mass balance errorNon-parametric two-tailed Komolgorov-Smirnov (K-S) testing at a 5% level of significance
-
Results Hydrologic ModelDesign-Event SimulationsOnset timing identicalPeak timing most different for 5-year events on largest watershed (figure 1)Peak discharge for K2MC increasingly greater than for K2 with increasing watershed size (figure 2)Figure 1. Relative difference in timing of hydrograph peak vs. watershed areaFigure 2. Relative difference in peak discharge vs. watershed area
-
Results Hydrologic ModelDesign-Event SimulationsRunoff volume for K2MC increasingly greater than for K2 with increasing watershed size (figure 1)Mass-balance error decreases with increasing watershed size for K2MC, and visa versa for K2 (figure 2)Figure 1. Relative difference in runoff volume vs. watershed areaFigure 2. Mass-balance error vs. watershed area
Chart1
2.5566516667-0.19641416672.7530658333
2.08282333330.65080866671.4320146667
1.20498183330.71536521670.4896166167
1.06000016671.0779835-0.0179833333
K2MC Ave.
K2 Ave.
Difference
Area (ha)
Obs_Ave_Graphs
Average Values as a function of scale - Efficiency
Area (ha)K2K2MC
4.740485-3.0166262-2.56910888
782.016-0.6055938-0.5883822
9557.6370.1471568420.278136236
14664.3-0.0985745143-0.0844701714
Average Values as a function of scale - Onset
Area (ha)K2K2MC
4.740485-0.0602884158-0.0602884158
782.0160.0262009017-0.1067605588
9557.6370.36849290780.2209692671
14664.30.6643491650.0673600165
Average Values as a function of scale - Peak
Area (ha)K2K2MC
4.740485-0.0306095501-0.0306095501
782.0160.03094505380.0005136039
9557.6370.08397292410.015890352
14664.31.84962238540.2450663019
Obs_Ave_Graphs
00
00
00
00
K2
K2MC
Area (ha)
Average Model Efficiency
Obs_Stdev_Graphs
00
00
00
00
K2
K2MC
Area (ha)
Average Onset Error (%)
Observed_Summary
00
00
00
00
K2
K2MC
Area (ha)
Average Peak Error (%)
Desing_Summary
Standard deviation as a function of scale - Efficiency
Area (ha)K2K2MC
4.7404856.45219213655.5962971866
782.0161.99718259532.0453376077
9557.6370.30051365270.3356297181
14664.30.20155604180.1944119929
Standard deviation as a function of scale - Onset
Area (ha)K2K2MC
4.7404850.22780686170.2278068617
782.0160.53557083580.3460264722
9557.6370.16464884980.1246822944
14664.31.11803240890.3244388189
Standard deviation as a function of scale - Peak
Area (ha)K2K2MC
4.7404850.22490832420.2249083242
782.0160.26241225630.2515058013
9557.6370.1746084170.151266561
14664.32.28108529350.3780694877
Desing_Summary
00
00
00
00
K2
K2MC
Area (ha)
Standard Deviation of Model Efficiency
Design_Qtot
00
00
00
00
K2
K2MC
Area (ha)
Standard Deviation of Onset Error (%)
Design_Err
00
00
00
00
K2
K2MC
Area (ha)
Standard Deviation of Peak Error (%)
K-S Test
WSEventModelEfficiencyOnsetPeakOnsetErrPeakErr
lh104780801Observed-110113
lh104780801K2-1.3032797112-0.1181818182-0.0088495575
lh104780801K2mc-1.225397112-0.1181818182-0.0088495575
lh104810728Observed-60138
lh104810728K2-14.4921701470.16666666670.0652173913
lh104810728K2mc-12.5064701470.16666666670.0652173913
lh104830910Observed-5666
lh104830910K20.3729234662-0.1785714286-0.0606060606
lh104830910K2mc0.430274662-0.1785714286-0.0606060606
lh104860810Observed-4663
lh104860810K2-0.0618453039-0.347826087-0.380952381
lh104860810K2mc0.03387863039-0.347826087-0.380952381
lh104860829Observed-102112
lh104860829K20.4011611201380.17647058820.2321428571
lh104860829K2mc0.4220071201380.17647058820.2321428571
wg11760727Observed-2646
wg11760727K2-0.19882327630.03846153850.3695652174
wg11760727K2mc-0.14584927610.03846153850.3260869565
wg11780801Observed-76100
wg11780801K20.65265781070.02631578950.07
wg11780801K2mc0.642527781060.02631578950.06
wg11800804Observed-6887
wg11800804K20.4293816593-0.04411764710.0689655172
wg11800804K2mc0.5201046286-0.0882352941-0.0114942529
wg11860829Observed-58166
wg11860829K20.2227331051680.81034482760.0120481928
wg11860829K2mc0.248247691670.18965517240.0060240964
wg11900801Observed-6082
wg11900801K2-4.133911852-0.7-0.3658536585
wg11900801K2mc-4.206941851-0.7-0.3780487805
wg1760727Observed-64203
wg1760727K2-0.337501721380.125-0.3201970443
wg1760727K2mc-0.333063711260.109375-0.3793103448
wg1780801Observed-218227
wg1780801K20.01083134855951.22477064221.6211453744
wg1780801K2mc0.01256162463490.1284403670.5374449339
wg1800804Observed-308374
wg1800804K20.15067300-1-1
wg1800804K2mc0.1618163544000.14935064940.0695187166
wg1810730Observed-95418
wg1810730K2-0.33365390558-0.05263157890.3349282297
wg1810730K2mc-0.28163790505-0.05263157890.2081339713
wg1860817Observed-8098
wg1860817K2-0.2279231036710.28755.8469387755
wg1860817K2mc-0.2199361031720.28750.7551020408
wg1860829Observed-355393
wg1860829K20.0967043326796-0.08169014081.0254452926
wg1860829K2mc0.102967148413-0.58309859150.0508905852
wg1900801Observed-178190
wg1900801K2-0.04915326206822.48314606742.5894736842
wg1900801K2mc-0.03399982552800.43258426970.4736842105
wg6760727Observed-94143
wg6760727K2-0.1056371541570.63829787230.0979020979
wg6760727K2mc0.005430981191360.2659574468-0.048951049
wg6780801Observed-135156
wg6780801K20.07462671711730.26666666670.108974359
wg6780801K2mc0.2785211561680.15555555560.0769230769
wg6800804Observed-120122
wg6800804K2-0.006051491611640.34166666670.3442622951
wg6800804K2mc0.04095621401510.16666666670.237704918
wg6860829Observed-120259
wg6860829K20.1088451662250.3833333333-0.1312741313
wg6860829K2mc0.2245271702160.4166666667-0.166023166
wg6900801Observed-8099
wg6900801K20.66400197990.21250
wg6900801K2mc0.84124688970.1-0.0202020202
Total Average Values for K2-0.82115169950.2694343790.5485369739
Total Average Values for K2MC-0.68127570090.03368734450.0747470974
WG1_obs
WSEventModelOnset_TPeak_TPeak_QOT_ErrPT_ErrPQ_ErrRelative Error Table - Onset Time
lh104100_1K27401.23048Area (ha)5_305_110_3010_1100_30100_1
lh104100_1K2mc7401.1817300-3.96186853914.74048500-1000
lh104100_30K22201.96785782.01060000-10
lh104100_30K2mc2201.8426300-6.36328988499557.6370-10-10-1
lh10410_1K212400.76854514664.3-100000
lh10410_1K2mc12400.74224600-3.4219206423
lh10410_30K24201.26064Relative Error Table - Time to Peak
lh10410_30K2mc3201.17973-10-6.4181685493Area (ha)5_305_110_3010_1100_30100_1
lh1045_1K214400.5995134.740485000000
lh1045_1K2mc14400.57996700-3.260312954782.0106-3.125-1.5151515152-2.27272727270-2.63157894740
lh1045_30K28200.6543599557.637-10.6796116505-5.737704918-3.488372093-3.0612244898-1.35135135140
lh1045_30K2mc8200.6168900-5.726061687914664.3-62.9370629371-12-4.5454545455-3.3057851246.7307692308-0.8403361345
wg11100_1K21653112.236
wg11100_1K2mc1653107.68200-4.0575216508Relative Error Table - Peak Discharge
wg11100_30K2838101.245Area (ha)5_305_110_3010_1100_30100_1
wg11100_30K2mc73797.9564-1-2.6315789474-3.2481604034.740485-5.7260616879-3.260312954-6.4181685493-3.4219206423-6.3632898849-3.9618685391
wg1110_1K2206045.5332782.01061.4459496989-4.5832209893-3.1717053697-5.0716839581-3.248160403-4.0575216508
wg1110_1K2mc206043.223900-5.07168395819557.63738.23025988493.0287193862-3.8938604274-6.4253586646-1.135098047-2.1115363442
wg1110_30K2104441.08214664.310057.653090968418.087558621217.3644520873-1.94478399960.5834305718
wg1110_30K2mc104339.7790-2.2727272727-3.1717053697
wg115_1K2216624.1075
wg115_1K2mc216523.00260-1.5151515152-4.5832209893
wg115_30K211644.93309
wg115_30K2mc11625.004420-3.1251.4459496989
wg1100_1K224119256.243
wg1100_1K2mc24118257.7380-0.84033613450.5834305718
wg1100_30K212104153.796
wg1100_30K2mc12111150.80506.7307692308-1.9447839996
wg110_1K22612114.9449
wg110_1K2mc2611717.540-3.30578512417.3644520873
wg110_30K21311011.3227
wg110_30K2mc1310513.37070-4.545454545518.0875586212
wg15_1K2271503.43064
wg15_1K2mc271325.408510-1257.6530909684
wg15_30K2174290.0092007
wg15_30K2mc161590.331149-1-62.93706293713499.1718021455
wg6100_1K22586301.534
wg6100_1K2mc2486295.167-10-2.1115363442
wg6100_30K21374201.128
wg6100_30K2mc1373198.8450-1.3513513514-1.135098047
wg610_1K2269850.6462
wg610_1K2mc259547.392-1-3.0612244898-6.4253586646
wg610_30K2138634.0382
wg610_30K2mc138332.71280-3.488372093-3.8938604274
wg65_1K22912212.6291
wg65_1K2mc2811513.0116-1-5.7377049183.0287193862
wg65_30K2151030.675376
wg65_30K2mc15920.9335740-10.679611650538.2302598849
WG1_obs
000000
000000
000000
000000
5_30
5_1
10_30
10_1
100_30
100_1
Area (ha)
Relative Error (%)
WG6_obs
000000
000000
000000
000000
5_30
5_1
10_30
10_1
100_30
100_1
Area (ha)
Relative Error (%)
WG11_obs
Total Runoff Discharge ValuesDifferences
WSArea (ha)Model5_305_110_3010_1100_30100_15_305_110_3010_1100_30100_1
lh1044.740485K2400.2794743.095882.0351010.0771457.4071729.482
lh1044.740485K2MC382.3582715.685840.76972.6461385.8151657.762
lh1044.740485rel. error (%)-4.6870186124-3.8298972313-4.9092487749-3.8483682655-5.1660575185-4.3263146338-17.9212-27.41-41.275-37.431-71.592-71.72
2-elem5K2194.9321445.467639.859699.3721217.1971433.428
2-elem5K2MC201.8431451.129641.606709.4531226.3881436.302
2-elem5rel. error (%)3.42394661991.25507338260.27228548361.42095388980.74943655680.2000971946.9115.6621.74710.0819.1912.874
3-elem10K2539.5751125.8621524.9661686.4322752.7253224.324
3-elem10K2MC500.0791056.4111452.3781602.4862641.5423088.585
3-elem10rel. error (%)-7.8979521236-6.5742405181-4.9978724547-5.2384857028-4.2090188231-4.3948604296-39.496-69.451-72.588-83.946-111.183-135.739
4-elem15K2823.5571709.5922310.1822554.4014152.8364872.251
4-elem15K2MC802.4671682.7982273.1192524.1214102.2544828.004
4-elem15rel. error (%)-2.6281454564-1.5922291327-1.6304909686-1.199625533-1.2330294516-0.9164656864-21.09-26.794-37.063-30.28-50.582-44.247
11-elem40K21931.0674189.845712.676349.3910477.5812325.63
11-elem40K2MC1848.4594057.395459.136173.679977.5312004.83
11-elem40rel. error (%)-4.4690198701-3.2644138227-4.6443297742-2.8462810613-5.0117614279-2.6722577496-82.608-132.45-253.54-175.72-500.05-320.8
wg11782.0106K212405.945104.868892.880108151408.3185300.4
wg11782.0106K2MC12029.842182.365438.575707.2144502.7177447.5
wg11782.0106rel. error (%)-3.1264027665-6.9282613798-5.2786967916-5.8129213602-4.7788726439-4.4254779583-376.1-2922.5-3454.3-4400.8-6905.6-7852.9
wg69557.637K23216549221308201860627551001068190
wg69557.637K2MC4453559151293891833907266281024771
wg69557.637rel. error (%)27.77902537621.7759098632-1.105967277-1.4570041987-3.9183736382-4.23694659591237993-1431-2672-28472-43419
wg114664.3K2571583855778921376959611128207
wg114664.3K2MC9642365063676977016751521085149
wg114664.3rel. error (%)94.087136929533.031712473612.40341730015.6949263569-3.0821207669-3.9679343574907781278985564-20809-43058
Negative values mean that K2MC predicted less runoff than K2
Relative Error Table
Area (ha)5_305_110_3010_1100_30100_1
4.740485-4.6870186124-3.8298972313-4.9092487749-3.8483682655-5.1660575185-4.3263146338
782.0106-3.1264027665-6.9282613798-5.2786967916-5.8129213602-4.7788726439-4.4254779583
9557.63727.77902537621.7759098632-1.105967277-1.4570041987-3.9183736382-4.2369465959
14664.394.087136929533.031712473612.40341730015.6949263569-3.0821207669-3.9679343574
53.42394661991.25507338260.27228548361.42095388980.74943655680.200097194
10-7.8979521236-6.5742405181-4.9978724547-5.2384857028-4.2090188231-4.3948604296
15-2.6281454564-1.5922291327-1.6304909686-1.199625533-1.2330294516-0.9164656864
40-4.4690198701-3.2644138227-4.6443297742-2.8462810613-5.0117614279-2.6722577496
WG11_obs
000000
000000
000000
000000
5_30
5_1
10_30
10_1
100_30
100_1
Watershed Area (ha)
Relative Error in Total Diascharge Volume (%)
LH104_obs
WSModel100_1100_3010_110_305_15_30
lh104K2
-
Results Hydrologic ModelDesign-Event SimulationsK-S testing showed significant differences between hydrographs simulated by K2MC and K2 for WG6 (9,558 ha) and WG1 (14,664 ha) for the 5-year return period eventsNull hypothesis: simulated discharges for MVPMC4 and kinematic routing have the same continuous distribution
WatershedArea (ha)5_305_110_3010_1100_30100_1LH1044.7AcceptAcceptAcceptAcceptAcceptAcceptWG11782AcceptAcceptAcceptAcceptAcceptAcceptWG69558RejectRejectAcceptAcceptAcceptAcceptWG114664RejectRejectAcceptAcceptAcceptAccept
-
Testing Hydrologic ModelObserved-Event SimulationsWatershed areas of 4.7, 782, 9558, and 14664 ha Five mid-sized events selected for eachUncalibrated comparison of kinematic and MVPMC4 simulated hydrographs in terms ofOnset and peak timingNash-Sutcliffe (1970) model efficiency (shape)Relative performance only
-
Results Hydrologic ModelObserved-Event SimulationsTiming of flow onset worsens with increasing watershed area for K2Timing of flow onset for K2MC shows no consistent relationship to watershed areaTiming of peak flow worsens with increasing watershed area for both models, but less so for K2MCFigure 1. Average error (%) in the timing of flow onset vs. watershed areaFigure 2. Average error (%) in the timing of peak discharge vs. watershed area
-
Results Hydrologic ModelObserved-Event SimulationsModel efficiency slightly greater for K2MC than K2 at all scales
-
Conclusions Hydrologic ModelHydrographs simulated by kinematic and VPMC routing were statistically different at the 5% level of confidence for events with a 5-year return period or smaller on watersheds of 9558 ha (95.6 km2) and larger Model mass-balance error decreased with increasing watershed area when VPMC routing is used opposite true for kinematic routing Outflow hydrographs simulated with VPMC routing more closely represented observed hydrographs than those simulated with kinematic routing at all scales, and performance gains increased with increasing watershed size VPMC routing more suitable than kinematic routing for conditions where flood-wave diffusion is most pronounced: for small to moderate events in watersheds larger than about 100 km2
-
Future Research Hydrologic ModelCompound channels were not implemented in the hydrologic model No event-based models using VPMC routing have implemented compound channelsCompound channels necessary for geomorphic modeling because Floodplains represent an important reservoir of stored sediment Overbank flow reduces flow depth & erosion in main channelGeomorphic model necessarily used kinematic routing
-
Geomorphic ModelHypothesis A continuous-simulation, event-based geomorphic model describing channel width, depth and slope adjustments can predict reasonable geomorphic change in semi-arid watersheds ApproachImplement channel-geometry adjustments in KINEROS2 based on total stream power minimizationDevelop a GIS-based interface to facilitate model parameterization, multiple-event simulations, and results visualizationEvaluate generalized model behavior in absence of observed channel-geometry changeSensitivity to initial channel geometryResponse to different precipitation recordsResponse to land-cover change
-
KINEROS2 Geomorphic Model (K2G)Width and depth adjusted to minimize total stream power at end of each time step Depth adjustmentsMaximum erodible depthBank failure Width adjustmentsCompound channelsDepth
-
AGWA-GGIS-based interface for K2G, customized version of AGWA Watershed delineation and discretizationLand cover and soils parameterizationCoordinates multiple consecutive simulations and tracks cumulative outputsResults visualizationDifferencing results from two simulations relative assessment
-
Profile SmoothingReaches treated independently in K2G, so slopes adjust independently to convey the inflowing sedimentExternal profile smoothing is thus required to maintain reasonable channel profiles during batch simulationsWeighted average elevation computed for each channel junctionEffectively transfers some sediment from downstream reach back to lower end of upstream reach(es) during deposition and visa versa during erosionUnsmoothed channel profileSmoothed channel profile
-
Geomorphic Model TestingObserved, distributed precipitation inputSSURGO SoilsHydraulic-geometry and observed-geometry channelsFour land-cover scenariosCompare results for 1964, 1977, & 1978 monsoon season on WG111973Part urbanAll urban1997DiscretizationSoilsRain GaugesElevation
-
Simulation InputsSediment grain-size distributionsLand-cover scenariosPrecipitation record characteristics for the 1964, 1977 and 1978 monsoon seasons
Year196419771978Number of events524741Total Precip. (mm)483.3369.241.9Ave. event depth (mm)9.37.91.0Max. event depth (mm)51.940.41.8Standard deviation12.48.90.4
Mesquite WoodlandsGrasslandDesertscrubUrbanNALC 19730.054.845.20.0NALC 19974.953.042.10.0Part Urban 19971.234.428.935.5All Urban0.00.00.0100.0
-
ResultsHydraulic-geometry channels1997 land coverWet (top), intermediate (middle), and dry (bottom) year simulation resultsDepth changes mapped on the left, width changes mapped on the rightErosion during wet year, and deposition during dry yearDecreasing Precipitation196419771978
-
ResultsHydraulic-geometry channelsPartially urbanized land coverDifferences from 1997 land cover not obviousLess erosion within, and more deposition and downstream of urbanized tributaryDecreasing Precipitation196419771978
-
ResultsRunoff depth (mm) per unit contributing areaRunoff highest as flows coalesce in the headwaters, then decreases in the downstream direction because of channel infiltrationSignificant decreases occur further upstream for drier yearsDeposition occurs downstream of transitionDecreasing Precipitation196419771978
-
ResultsObserved-geometry channels1997 land coverWet (top), intermediate (middle), and dry (bottom) year simulation resultsDepth changes mapped on the left, width changes mapped on the rightReach adjustments more spatially varied with observed channelsDecreasing Precipitation196419771978
-
ResultsObserved-geometry channelsPartially urbanized land coverReach adjustments more spatially varied with observed channelsCan see preferential change on southern tributaryDecreasing Precipitation196419771978
-
Mass-Balance ErrorMass-balance error for modeled change in sediment storage (red) and equivalent mass of geometric adjustment (blue) for entire channel networkModeled cumulative magnitude of deposition/erosion (burgundy), and equivalent mass of geometric adjustment (cream)
Geometric adjustments conserve mass reasonably well for hydraulic-geometry simulations, but not for the observed-geometry simulationsMass-balance error (%)Magnitude of deposition/erosion (kg)
-
Hydrologic Impacts of Land-Cover ChangeHydraulic-Geometry ChannelsObserved-Geometry Channels
-
Relative AssessmentError in watershed modeling is substantial Even carefully calibrated models yield poor results when applied to events significantly larger or smaller than those used in the calibrationGeomorphic model is thus most useful for evaluating where in the watershed change is likely to be most significantAssuming the basic processes are represented accurately, and error is spatially uniform, it can be largely removed through differencing simulation resultsRelative assessment can thus identify general patterns of response to landscape change, even if the specific magnitude of that change is not correct
-
Results Relative AssessmentHydraulic-geometry channelsDifference in computed depth (left) and width (right) changes between PU and 97 simulation results for wet (top) and intermediate monsoon seasonsSignificant differences concentrated on urbanized tributaryErosion increases within urbanized area more pronounced for wet yearReduced erosion or increased deposition begins further upstream during drier yearAggradation downstream characterized by depth decreases and width increases
Decreasing PrecipitationDifference in depth changesDifference in width changes19641977
-
Results Relative AssessmentObserved-geometry channelsMagnitude of differences is different from the hydraulic geometry simulationsPattern of adjustment very similar to that for the hydraulic-geometry channels erosion in urbanized area and deposition downstreamSuggests that channel slope and discharge are the most important parameters governing channel response
Decreasing PrecipitationDifference in depth changesDifference in width changes19641977
-
ResultsWet monsoon cumulative runoff (mm), infiltration (m3/km), and sediment yield (kg/ha)Runoff increases from urbanization decrease in downstream directionInfiltration increases in downstream directionSediment yield increases from urbanization increase in downstream directionSpatial patterns very similar for both hydraulic and observed geometriesIncreased deposition downstream where stream power decreasesHydraulic-Geometry ChannelsObserved-Geometry ChannelsRunoffInfiltrationSed. YieldRunoffInfiltrationSed. Yield
-
ResultsIntermediate monsoon cumulative runoff (mm), infiltration (m3/km), and sediment yield (kg/ha)Runoff increase from urbanization dissipates more rapidly in downstream directionInfiltration increase peaks further upstreamSediment yield increase peaks further upstreamLocus of deposition shifts upstreamSpatial patterns very similar for both hydraulic and observed geometries
Hydraulic-Geometry ChannelsObserved-Geometry ChannelsRunoffInfiltrationSed. YieldRunoffInfiltrationSed. Yield
-
Conclusions Geomorphic ModelNetwork-wide mass conservation is reasonable when hydraulic-geometry channels are used, but needs work for more variable observed-geometry channels Erosion is most widespread during the wettest year, erosion and deposition mixed during intermediate year, and most widespread deposition for driest yearSpecific channel adjustments sensitive to initial channel geometry more uniform for hydraulic-geometry channels
Individual Batch Simulations
-
Conclusions Geomorphic ModelResults of the scenario-output differencing show the concentration of impacts within and downstream of the urbanized area, and no significant changes in the unaffected areas Geomorphic impacts of urbanization varied with the number and magnitude of precipitation events, but the general response was erosion in the urbanized area and deposition downstreamSpatial pattern of geomorphic response closely linked to changes in cumulative runoff and channel infiltrationSpatial pattern of geomorphic response relatively insensitive to initial channel geometry, suggesting that a suitable hydraulic-geometry relation may be sufficient for broad-scale application of the model
Relative Assessment
-
Future Research Geomorphic ModelModel validation need to demonstrate that simulated geomorphic adjustments are representative of observed adjustments Increase upper watershed size limit for K2G Diffusion-wave routingDiscritization of channel Link simulated geomorphic change and channel stability, or vulnerability to degradationEvaluate model behavior over broader range of precipitation records, and over longer periods of timeEvaluate model response to major disturbance, and whether response is persistent or transitiveLink simulated geomorphic change and riparian condition
watersheds thus represent the most convenient spatial entities within which they can be describedAnd their use as assessment tools(and assessment tools)
Upper limit of watersheds size to which arid-region overland flow models can be applied is sufficiently small to hinder their practical use for large-scale assessment and planning.Existing models can only simulate sediment yield - high spatial and temporal resolutionBegins to decline when small-watershed models applied to areas larger than about 10 km2, and beyond about 100km2 theyre useful only for the largest eventsLower temporal and spatial resolution for which they have insufficient spatial and temporal resolution to adequately resolve the hydrograph in arid regionsRelatively short-duration runoff from spatially-variable rainfall with little to no sub-surface flow - need high spatial and temporal resolution - Numerous reach-based, two- and three-dimensional geomorphic models, but these require water and sediment inflows. In the absence of this data, howeverThis figure illustrates hydrologic processes active over a range of spatial and temporal scales. In the realm of small watershed models (lower left) you can see that overland and channelized flow resulting from short-duration storms of limited spatial extent are well represented by simulations of 24 hours or less for areas of less than about 100 km2. Large watershed models, which can be run for much longer periods of time, are better suited to representing longer-duration rainfall over larger areasBetween these two scales, however, there is a fairly substantial gap, within which hydrologic and geomorphic systems are not well represented by either small- or large-watershed models. For arid regions in particular, it is necessary to represent channelized flow from short-duration events with a relatively high temporal resolution (hydrologic model). For geomorphic systems, it is necessary to simulate cumulative change from multiple short-duration events to extend the temporal scale and to address questions about geomorphic response to landscape change (geomorphic model).Define flood-wave diffusion!