Between October 9, 1991 and October 16, 1991 a fire burned 577 hectares in the Columbia River Gorge near the west end on the Oregon side. All of the area burned consisted of second growth Pseudotsuga menziesii and the accompanying understory. This was the first disturbance of this magnitude in this part of the Columbia River Gorge since 1902. The purpose of this study was to examine the pattern of understory recovery in the first two years following the fire. This study also sought to learn: 1) how Pseudotsuga menziesii seedlings are recruited into the population, 2) how quickly the litter layer is a reforming, and 3) how quickly snags and downed logs are recruited into the understory. Four 800 square meter circular plots were established within the burned area of the Columbia River Gorge. Two plots were designated sun plots since the fire had killed the overstory. The other two were designated shade sites since the canopy over them was still intact. Twenty five randomly placed sample units (20 x 50 centimeters) were placed in each main plot. The plots were then sampled at approximately onemonth intervals from May through September of 1992 and 1993. The frequency and percentage of cover was recorded for all plant species that occurred in each sample unit. The data from 1992 and 1993 were compared by date of visit and type of plot, either (sun or shade) using the Pearson Goodness-of-Fit Test to examine and compare differences in the extent of cover and distribution of understory species. No significant differences were found. An increase in species richness and relative abundance of understory species was noted between pre-fire data collected by the US Forest Service and what I found. However, statistical analysis was not possible because of the limited data collection in the pre-fire sample.

As the second largest river in the U.S., the entrance to the Columbia River is home to some of the most extreme wave conditions on the Pacific Coast. Winter storms commonly generate waves 6-8 m in height, which in combination with strong tidal currents, can produce dangerous navigation conditions. To improve understanding of the wave dynamics in this complex setting, the SWAN model is applied; 2 hindcasts are conducted and an operations forecast is developed. The model is forced with offshore wave heights obtained from a buoy located in 134 m water depth (for the hindcasts) and a specialized WaveWatchIII forecast (for the forecast). In both cases tidal currents are obtained from SELFE, a circulation model of the Columbia River. The hindcasts are validated through measurements obtained from an inshore buoy located in 25 m water depth, a 4-week field experiment and remote sensing methods. The model performs best at the location of the buoy, with a normalized root-mean-squared error (NRMSE) of 11%, primarily because it is outside the area of strong tidal currents. Within the river mouth, the model is able to predict the changes in the wave field due to currents, but its performance is limited by errors in velocity estimates and strong shears in the tidal current profile. From the modeling work, it is evident that wave transformations at the mouth of the river are dominated by the tidal currents. The forecast has been operational since August 2011 and provides 45-hours of predictive wave information. In comparison with measured wave heights at the buoy, the forecast performs well, with a NRMSE of 16%. The majority of errors are caused by errors in the input conditions, since they themselves are forecasted. Additional errors arise from phase-resolved properties in the wave field that the model is unable to produce; these errors are also present in the hindcasts. Despite the limitations, this forecast provides valuable information to bar pilots since it includes the effects of the tidal currents. / Graduation date: 2013

Natural and anthropogenic processes over the past 150 years have altered the bathymetry of the Lower Columbia River (LCR) and have changed the long wave propagation of tides and floods. Possible causes for the increase in tidal amplitudes (+7% in tidal range in Astoria) are decreases in river discharge, lengthening of the river channel due to the construction of jetties at the mouth, dredging and deepening of the shipping channel, and reduction of the tidal prism due to the filling and diking of tidal wetlands. In this study, changes in the characteristics of long waves are elucidated by developing two hydrodynamic models of the LCR which reflect historical and modern bathymetric conditions and forcing. The historic model simulates late 19th century conditions and is extensively validated using recently recovered tide records along the LCR (e.g., Astoria, 1853-1876) and river stage measurements (e.g., Portland, 1876-1964). Results suggest that water levels in Portland at low river discharge are up to 0.5-1.0m lower than in the past. However, historical water levels during a flood scenario based on the 1880 spring freshet are similar to modern water levels. Since tidal range in the modern scenario is persistently higher at all locations, the flood risk in many locations along the LCR has increased for the same boundary conditions. The results are explained by considering the governing equations of momentum and mass-conservation. At low river flow, greater depth leads to reduced frictional effects, producing amplified tidal range and tidal velocities but a decreased river slope (and lower Portland water levels). At high flow, the modern flood is confined by dikes and the loss of wetlands, which counteracts the effect of decreased friction. Nonetheless, the high friction of the historical wooded floodplain also confined the historical flood path. Hence, historical and modern flood heights are surprisingly similar, though scaling analysis suggests that the historical flood wave was more diffusive.

Pacific lamprey (Entosphenus tridentatus), an endemic species to the Columbia River Basin, U.S.A, has experienced staggering decreases in returns to spawning territories in recent decades. As lamprey are threatened severely by a lack of passage at mainstem dams, lamprey specific passage structures have been designed and constructed to address the problem. The Cascades Island Lamprey Passage Structure (LPS) at Bonneville Dam is the longest and steepest structure of its type, following the addition of an exit pipe which allows lampreys to travel from the tailrace of the dam to the forebay. The intent of this study was to assess lamprey use of the structure and whether the structure hinders lamprey migration to subsequent dams. The study was carried out during the 2013 migration season. The study used three different treatment groups of lampreys released on five dates spanning the migration season (n=75 lamprey). Two of these groups (n=50), with different tagging methods, were released directly into the LPS to assess passage success, travel time, and tagging effect. The third group (n=25) was released into the forebay to test whether the structure impedes migration upstream. Fish were monitored via receiver arrays on the LPS and at dams on the river system. Overall passage efficiency was 74% (37 of 50 used the CI LPS successfully). Mean travel time to navigate the structure was 12 h. Fish size had no significant effect on travel time in the LPS. Water temperature had a significant effect on travel time in the LPS. There was no statistically significant effect of tagging on passage efficiency or travel time. The groups that used the LPS performed slightly better migrating upstream to the next dam than the group that bypassed the structure, but the difference was not significant. The groups that used the LPS traveled to more subsequent dams upstream than did the group that bypassed the LPS. It can be concluded that lamprey passed the structure successfully. Temperature (proxy for seasonality) had an effect on travel time in the LPS; however fish size and tagging had no effect. The LPS does not affect the ability of migrating lampreys to continue migration to subsequent dams. Such findings have important implications for management of lamprey in the region.

The multi downscaled-scenario products allow us to better assess the uncertainty of the variations of precipitation and temperature in the current and future periods. Joint Probability distribution functions (PDFs), of both the climatic variables, might help better understand the interdependence of the two, and thus in-turn help in accessing the future with confidence. In the present study, we have used multi-modelled statistically downscaled ensemble of precipitation and temperature variables. The dataset used is multi-model ensemble of 10 Global Climate Models (GCMs) downscaled product from CMIP5 daily dataset, using the Bias Correction and Spatial Downscaling (BCSD) technique, generated at Portland State University. The multi-model ensemble PDFs of both precipitation and temperature is evaluated for summer (dry) and winter (wet) periods for 10 sub-basins across Columbia River Basin (CRB). Eventually, Copula is applied to establish the joint distribution of two variables on multi-model ensemble data. Results have indicated that the probabilistic distribution helps remove the limitations on marginal distributions of variables in question and helps in better prediction. The joint distribution is then used to estimate the change in trends of said variables in future, along with estimation of the probabilities of the given change. The joint distribution trends are varied, but certainly positive, for summer and winter time scales based on sub-basins. Dry season, generally, is indicating towards higher positive changes in precipitation than temperature (as compared to historical) across sub-basins with wet season inferring otherwise. Probabilities of changes in future, as estimated by the joint precipitation and temperature, also indicates varied degree and forms during dry season whereas the wet season is rather constant across all the sub-basins.

Impact pile driving can produce extremely high underwater sound levels, which are of increasing environmental concern due to their deleterious effects on marine wildlife. Prediction of underwater sound levels is important to the assessment and mitigation of the environmental impacts caused by pile driving. Current prediction methods are limited and do not account for the dynamic pile driving source, inhomogeneities in bathymetry and sediment, or physics-based sound wave propagation. In this thesis, a computational model is presented that analyzes and predicts the underwater noise radiated by pile driving and is suitable for shallow, inhomogeneous environments and long propagation ranges. The computational model uses dynamic source models from recent developments in the technical literature. Pile source models are coupled to a broadband application of the range-dependent acoustic model (RAMPE), a standard parabolic equation (PE) propagation code capable of modeling wave propagation through complex, range dependent environments. Simulation results are shown to be in good agreement with several observations of pile driving operations in the Columbia River between Portland, Oregon and Vancouver, Washington. The model is further applied to extend sound level predictions over the entire river and study the effects of sediment and bathymetry on the underwater sound levels present in the environment.

A survey of small mammals on eight islands in the Columbia and Willamette Rivers near Portland, Oregon was conducted in 1974. The islands were Govermnent , Sand, Lemon, McGuire and Sandy Islands in the Columbia River and East, Ross and Hardtack Islands in the Willamette River. The objectives of the study were to ascertain and compare the kinds, distribution and relative densitites of small mammals . A variety of traps was used to capture the animals including Museum Special Rodent Traps , back- break mouse traps, scissor and guillotine- type stab mole traps , Sherman traps , modified Young traps and pitfalls . Most traps were set in linear transects . Trapping was conducted with the objective of sampling the major habitats on each island. Observations of scats, tracks , burrows and dead animals were also noted. One species of small marmnal , Sorex vargrans , was found on all eight islands . Microtus was found on five islands and Peromyscus maniculatus on four islands . Two islands yielded Scapanus townsendii. Glaucomvs sabrinus was found on only one island. I hoped that evidence would be found to prove or disprove that current theories of island biogeography would apply to river islands. However, data in this study were inadequate to allow calculation of precise indices of population densities . Therefore, I was unable to determine if theories of island biogeography do apply to these river islands.

Monthly samples of nutrients, phytoplankton and zooplankton were taken in the Columbia River estuary over a period of 16 months in order to determine distribution with season and salinity, and interrelationships between plankton and nutrients. Nitrate and phosphate levels in the river water entering the estuary are high in the winter and show depletion during the summer. Silicate levels are high in the river water at all seasons. Nitrate and phosphate levels are high in the entering ocean water during summer offshore upwelling. The nutrient levels in the estuary generally show a linear relationship with salinity, resulting from the levels of nutrients in the entering river and ocean water. Superimposed upon this linear relationship is a tendency for the nutrients to be enriched in the bottom waters of the central part of the estuary. The estuary phytoplankton are primarily composed of freshwater forms, and probably represent a downstream extension of the river flora. Regression analysis of phytoplankton levels vs. light, nutrients, and river flow indicates that light probably limits phytoplankton abundance on most dates. The zooplankton of the estuary are composed of three groups, preferring fresh, oligohaline, and polyhaline waters respectively. Regression analysis indicates a strong correlation between abundance of the freshwater group and river temperature. The factors controlling the abundance of the oligohaline and polyhaline groups are less obvious. The oligohaline group, principally Eurytemora affinis, reaches the greatest population density (100,000/m³ or more). Regression analysis indicates a close correlation between Eurytemora abundance and phosphate levels. This indicates a strong potential for zooplankton regeneration of phosphate necessary for phytoplankton growth. / Graduation date: 1970

The spawning habitat associated with fall chinook salmon (Oncorhynchus tshawytscha) redd clusters was investigated in the Hanford Reach of the Columbia River. A conceptual spawning habitat model is proposed that describes how geomorphic features of river channels create hydraulic processes, including hyporheic flows, that influence where salmon spawn in unconstrained reaches of large mainstem alluvial rivers. Spatial point pattern analysis of redds showed that redd clusters averaged approximately 10 hectares in area and their locations were consistent from year to year. The tendency to spawn in clusters suggests fall chinook salmon's use of spawning habitat is highly selective. Hydraulic characteristics of the redd clusters were significantly different than the habitat surrounding them. Velocity and lateral slope of the river bottom were the most important habitat variables in predicting redd site selection. While these variables explained a large proportion of the variance in redd site selection (86 to 96%), some unmeasured factors still accounted for a small percentage of actual spawning site selection. Further investigation showed that the magnitude and chemical characteristics of hyporheic discharge were different between and within two spawning areas. Apparently, fall chinook salmon used chemical and physical cues from the discharge to locate spawning areas. Traditional spawning habitat models could be improved if they: used spawning area-specific, rather than river-specific; spawning characteristics; incorporated hyporheic discharge measurements; and gave further consideration to the geomorphic features that are present in the unconstrained segments of large alluvial rivers. Ultimately the recovery of endangered fall chinook salmon will depend on how well we are able to recreate the characteristics once common in alluvial floodplains of large rivers. The results from this research can be used to better define the relationship between these physical habitat characteristics and fall chinook salmon spawning site selection, and provide more efficient use of limited recovery resources. / Graduation date: 1999

The mechanisms of mortality during critical life stages of fish are not well-understood and, for many species, it is not clear if the mechanisms are similar for naturally and artificially propagated individuals. For Chinook salmon (Oncorhynchus tshawytscha), natural fish potentially face negative interactions, such as competition, and survival disadvantages, such as smaller size, that may limit survival when in association with hatchery fish. To better understand the mechanisms of mortality for hatchery and natural Chinook salmon during the critical early marine residence stage, I: (1) developed a model to discriminate between hatchery and natural juveniles using otolith structure; (2) directly compared migratory patterns of hatchery and natural juveniles; and (3) determined if there was evidence for selective mortality during early marine residence. I followed two cohorts through space and time by collecting juveniles from May-September in the Columbia River estuary and off the coast in September of 2010 and 2011. I compared attributes of those juveniles when they firstentered marine waters with those of survivors after their first summer at sea. I used a combination of genetic stock identification, otolith chemistry and structure, and physical tags to determine stock of origin, size at and timing of freshwater emigration, marine growth, and production type (hatchery or natural). I focused on the subyearling life history of a federally managed genetic stock group (upper Columbia River summer and fall Chinook salmon, UCR Su/F) because: 1) it is an abundant stock group; 2) subyearlings may be more vulnerable to size-selective mortality than yearlings; and 3) it is currently impossible to assess impacts of hatchery production due to low rates of marking the hatchery fish within this stock group. The classification model included two metrics, the presence or absence of a previously unreported transfer check associated with hatchery rearing and variability in otolith increment width, and predicted production type with a 92% jack-knifed accuracy. Overall, timing of marine entry was similar for hatchery and natural UCR Su/F juveniles, which entered marine waters from May-September with a peak in July and August in both years. Estuarine residence times were brief: 80% of the individuals captured in the estuary had resided in saline waters for < 3 days and mean estuarine residence was significantly greater (7 ± 1.3 d) in 2010 than 2011 (1 ± 0.3 d). The only clear difference was that natural individuals captured in the estuary in 2011 migrated to saline waters earlier (July 13th ± 4 d) than hatchery conspecifics (August 10th ± 6 d). However, the timing of marine entry was similar (July 27th ± 1 d) between hatchery and natural fish collected later in the ocean. This observation could be due to differential survival related to the timing of marine entry. Alternatively, estuarine collections may not have adequately represented the emigrating population due to rapid emigration. I documented clear spatial overlap between production types during early marine residence but no difference in median size at marine entry (100 ± 3.5 mm), size at capture (152 ± 4.0 mm), or marine growth (0.94 ± 0.1 %b l d-1). There were also no significant differences in size at marine entry between estuary and ocean collections, which indicates that size-selective mortality had not occurred. Based on both external tags and the otolith classification model, the mean percentage of natural fish in ocean collections was 17% (± 4.8) greater than in the Columbia River estuary; this finding may indicate that estuarine collections are biased to hatchery fish or, more likely, that natural fish survived at higher rates than hatchery fish. Increased survival of natural fish may be related to greater selection pressure during freshwater rearing and prior experience with predators. This study provides the first direct stock-specific comparison of juvenile migratory behavior in natural and hatchery juvenile Columbia River Chinook salmon during early marine residence. Further research is needed to determine if natural fish consistently survive better than hatchery conspecifics and, if so, determine the specific traits and behaviors that afford a survival advantage. / Graduation date: 2013