Fluvial Change Processes during an Exceptional Drought Punctuated by Atmospheric Matthew D. Weber1, Gregory B. Pasternack1, Duane A. Massa2 1Department of Land, Air and Water Resources, Univ. of California at Davis, Davis, CA 95616; 2Yuba Accord Management Team, Marysville, CA

Introduction Example DEM Uncertainty Analysis* Budgets and Topographic Change Processes (TCP’s) High-resolution (meter-scale) repeat topographic surveys help improve our understanding of watershed retreating in Timbuctoo Bend sediment budgets, geomorphological processes, and their links to ecosystem services, but in order to Migration in Timbuctoo Bend Reach • Results show a series of crests develop accurate assessments of topographic change, results must be adjusted based on the 1. Classify *Note: Full 37.5 km retreating through knickpoint migration in uncertainty associated with each raster-based digital elevation model (DEM). This research develops Point Cloud Timbuctoo Bend Reach (TBR). a new method for scaling confidence thresholds (i.e. Levels of Detection) for a raster-DEM stretch of LYR was • This is in contrast to long-term and large- based on point density, surface variability, and land cover type. The results are applied to a analyzed. results which show the stability of topographic change analysis for the Lower Yuba River from 2008-2014, a period that riffle crest locations in TBR. experienced several moderate from atmospheric river events. • These moderate flood events may cause riffles to erode without the replacement of Study Area 2. TIN to 3. Create Land material that would be scoured from pools Raster DEM Cover Raster during larger events. The 37.5-km lower Yuba River (LYR) Zr2014 RMSELC A is a regulated wandering gravel-bed Out of In Net river with a thick mixed coarse Reach 3 Channel 3 [m /yr] 3 [m /yr] sediment alluvial fill (Dmean ~ 100 [m /yr] mm), significant variations in Englebright 949 -182 767 channel and width, and a Timbuctoo Bend -282 -2091 -2373 dynamic flood regime with frequent Parks 1260 -4789 -3528 overbank floods and a high B Area = 3480 km2 Dry Creek 2167 -7306 -5139 probability of large floods.. Daguerre Point 1040 -7604 -6564 Objectives Hallwood 14844 -16487 -1642 4. Ground Point Marysville 2819 -6064 -3245 1. Create a robust DEM 5. Create a standard deviation raster Count Raster C All LYR 22797 -44521 -21725 푆퐷2014 = 푆퐷_푍2014 + 푅푀푆퐸퐿퐶 uncertainty analysis N2014 2. Quantify topographic Reach Statistics • Englebright Dam reach shows fill from a change across segment, gravel augmentation project. reach, and morphological • The Timbuctoo Bend reach is the only reach unit scales with net scour for both in-channel and out-of- channel areas. This is due to the prominence of knickpoint migration in that Lateral Migration in Dry Creek Reach reach. • Other reaches are dominated by lateral Data Collection channel migration which leads to in-channel fill and scour as the channel moves. 2008 Map Survey Point Statistics • Airborne Near-Infrared (NIR) LiDAR Scour Fill Net Methods: 6. Calculate Standard Methods (continued): 2008 MU's • Singlebeam Sonar 2008 Map 2014 Map 5. Estimate the variability in the [m3] [m3] [m3] Land Cover 2 2 1. Classify all survey points Error of the Mean • RTK GPS/Total Station surveys [pts/m ] [pts/m ] 푆퐷2014 chute -7276 19683 12407 into ground, vegetation, 푆퐸푀 = data, SDYEAR, which is the 2014 푁 Bare earth 5.71 13.17 building, or noise data. 2014 sum of the standard deviation fast glide -9538 52845 43307 2014 Map of heights in each raster cell, pool -16208 82660 66453 Water 0.59 5.12 2. Use the ground points to • Airborne NIR and Green (bathymetric LiDAR) SD_Z , and the RMSE for riffle -34145 32145 -2000 create a Triangulated 2014 LC • Multibeam Sonar for deep pools Vegetated 1.37 3.05 the land cover class. riffle transition -19546 31961 12414 ground Irregular Network (TIN). • Limited gap-fill by other methods: Singlebeam 6. Calculate the Standard Error run -7813 40298 32484 Create a raster DEM from Lateral Migration Sonar and RTK-GPS surveys of the Mean, SEM , using slackwater -34629 26571 -8058 that TIN where ZrYEAR is the YEAR • Lateral migration was a dominant mechanism outside of Timbuctoo Bend. the standard deviation raster, slow glide -20593 28193 7600 Study Period height in each raster cell. • Abundant non-cohesive bank material and frequent over-bank flows allow the LYR to adjust its SDYEAR, and the number of 3. Create a land cover raster channel with these moderate events. Morphological Unit Results • Bankfull for the LYR is ground returns per raster cell, and calculate the RMSE for • Because the dominant TCP’s were due to approximately 142 m3/s with a NYEAR. This represents the each land cover type. Lateral Migration in Hallwood Reach lateral channel migration, the results D return period of ~1.25 years. confidence band for the DEM Calculated RMSELC: 7. Calculate 95% Conf. Level of Detection classified by in-channel MU’s are in each raster cell. • Bare earth – 0.038 m 퐿푂퐷 = (1.96 ∗ 푆퐸푀 )2 + (1.96 ∗ 푆퐸푀 )2 dominated by fill. C • filling discharge is 95 2008 2014 7. Calculate the 95% confidence B • Water – 0.073 m • This does not allow us to answer the A approximately 600 m3/s with a level of detection, LOD , by • Vegetated – 0.091 m 95 question of whether pools or riffles are return period of ~2.5 years. combining the SEM’s from 4. Count ground points for aggrading or scouring. each year’s DEM. • In order to answer that question, MU • During the study period, four each raster cell, NYEAR. 8. Add or subtract the LOD movements must be tracked for each floodway filling events (A-D on the 95 from the raw difference of survey period and then analyzed. hydrograph) occurred ranging DEM’s to get the net scour from 5.9-8.8 times bankfull and fill rasters. discharge or ~2.5-5 year events.

Conclusions Example Observed Channel Change 8. Adjust the Raw Topographic Change by the LOD 95 • The uncertainty analysis effectively eliminates noise in the data and scales the level of detectable change by ∆푍 = 푍푟2014 − 푍푟2008 ± 퐿푂퐷95 (4) 2008 Channel 2014 Channel accounting for point density, surface variability, and land cover. Raw Topographic Change Statistically Significant Topographic Change • Moderate floods caused the retreat of riffle crests through knickpoint migration in Timbuctoo Bend Reach (TBR). This result is in contrast to the observed long-term trend of persistent riffle locations in TBR. • Lateral migration dominated elsewhere with non-cohesive bank retreat and in-channel fill being dominant processes. • The Lower Yuba River as a whole is net erosional with an estimated export of ~22,000 m3/yr of sediment.

Contact: Matthew D. Weber Email: [email protected]