Loading...
HomeMy WebLinkAboutAlaska In_River Hydrokinetic Energy Resources FINAL REPORT July 3 2014 2014 Tom Ravens, Ph.D. University of Alaska, Anchorage 7/3/2014 Alaska In-River Hydrokinetic Energy Resources 2 Acknowledgements The Alaska In-River Hydrokinetic Energy Resource Assessment was conducted by a large number of student research assistants. The students arranged for logistical support during the field visits, operated the equipment, analyzed and processed the data, and wrote a good portion of the final report. The students who contributed to this project are listed below. The contributions of Garrett Yager, in particular, should be noted as he contributed expertise in the area of surveying, the operation of oceanographic equipment, and data processing. Garrett also led the field team for a good portion of the project. Maria Kartezhnikova excelled at the numerical modeling of river hydraulics. Michael Ulmgren excelled at hydrographic surveying and velocity measurement. I would also like to acknowledge the assistance of our numerous boat operators and village collaborators. Kyle Albert Aaron Butterer Erin Eggleston Taylor Harper Patrick Hoosier Curtis Huenefeld Maria Kartezhnikova Daniel King Jennifer Mills John Oldfield Donald Richardson Michael Ulmgren Garrett Yager 3 Executive Summary The University of Alaska Anchorage (UAA) conducted a statewide assessment of the in-river hydrokinetic energy resource in Alaska. The assessment featured visits to 31 rural Alaska village sites; the surveying of proximal river segments for bathymetry, topography, velocity, and water surface elevation; the construction of hydraulic and hydrologic models based on the data; and the estimation of the hydrokinetic power density (W/m2) at the 25, 50, and 75-percentile flow rates of the open water period at each village site. Site assessments were conducted for villages on the Yukon, Kuskokwim, Tanana, Copper, Susitna, and Talkeetna Rivers. Figure ES-1 below provides example results for the power density (W/m2) and velocity (m/s) at four flow rates for the village of Red Devil on the Kuskokwim River. Figure ES-1. Hydrokinetic power density (W/m2) and velocity (m/s) in the Kuskokwim River proximal to Red Devil for the flow rate at the time of measurement and for the 25, 50, and 75- percentile flows of the open water period. 4 In addition, the project produced a statewide overview of the hydrokinetic resource based on the median open-water-period discharge and the cross-section average velocity of representative transects, at the selected sites (Figure ES-2). The statewide overview indicates that the sites on the Copper River tended to have the highest velocities and power densities. Conversely, the sites on the lower Yukon had relatively low power densities. The site with the lowest calculated power density was Teller, which was the one tidal site considered. All of the sites studied have sufficient velocities to be of interest to developers of hydrokinetic energy systems. Figure ES-2. Statewide distribution of in-river hydrokinetic power density. For quantitative data, see Table 2 on page 33. This report provides detailed information about the site assessment methodology as well as summary information on the resource at individual sites. Readers interested in detailed information from particular sites are directed to the “Site Investigation Reports” that were developed for individual sites. These Site Investigation Reports accompany this Final Report. Finally, the reader should note that this report provides data on the theoretically recoverable resource. Practical considerations, such as insufficient water depth or insufficient velocity, might mean that the resource is not recoverable with existing technologies. EPRI (2012) has produced a companion assessment for the technically recoverable as well as the theoretical resources in Alaska and in the “Lower 48”. They address the distinction between the theoretically available resource and the technically recoverable resource in detail. 30 - 50 W/m2 5 Table of Contents Section 1: Background, Objectives and Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.1 Background and overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.2 Objectives and tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . 10 1.3.1 Identifying candidate village sites for resource assessment . . . . . . . . . . . . 10 1.3.2 Logistics of travel to village sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10 1.3.3 Equipment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.3.4 Planning the data collection effort. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 1.3.5 Equipment trials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 1.3.6 Data acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 1.4 Selected sites for data acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Section 2: Data Analysis (post processing) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.1 Control and Adjustments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.1.1 OPUS Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.1.2 Adjustments to upland topography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.1.3 Adjustments to bathymetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.1.4 Determine river slope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.1.5 Determine the Manning roughness coefficient (n) . . . . . . . . . . . . . . . . . . . 30 2.1.6 Stage adjustments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.2 Modeling with the Manning equation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31 2.2.1 Modeling of discharge for observed stages. . . . . . . . . . . . . . . . . . . . . . . . . 31 2.2.2 Modeling of discharge versus cross-section average velocity . . . . . . . . . . 31 2.3 Validation and application of models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33 6 2.3.1 Validating model output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 2.3.2 Time of travel and gauge factor estimation . . . . . . . . . . . . . . . . . . . . . . . . .33 2.3.3 Estimation of discharge, velocity, and power . . . . . . . . . . . . . . . . . . . . . . .35 2.4 Numerical modeling of river hydraulics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 2.4.1 Data preparation for modeling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 2.4.2 Description of available hydraulic models . . . . . . . . . . . . . . . . . . . . . . . . .39 2.4.3 Assessment of available models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 2.4.4 Model verification at specific sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Section 3: Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 3.1 Map of statewide power density distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 3.2 Power and velocity contour plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Section 4: Error Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 4.1 Uncertainty in the calculation of statewide distribution of power density . . . . . . . . 44 4.2 Uncertainty in the calculation of spatial distribution of power density . . . . . . . . . . 44 References . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Appendix A. Assessment of Manning Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Appendix B. Assessment of Hydraulic Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Appendix C. Average Dates of Open Water Period at Selected Communities . . . . . . . . . . 50 7 Section 1 Background, Objectives and Methodology 1.1 Background and Overview Alaska has over 300 rural villages that are located near a river. The Electric Power Research Institute (EPRI) has performed conceptual feasibility studies for a few Alaska sites which showed that the economics for River In-Stream Energy Conversion are quite promising, potentially reducing the cost of electricity by 50-70% (EPRI 2008). In this project, the hydrokinetic energy resource was assessed at 31 rural sites in Alaska. The sites included: 16 sites on the Yukon River (Rampart, Stevens Village, Beaver, Tanana, Napaimute, Galena, Koyukuk, Nulato, Kaltag, Grayling, Anvik, Holy Cross, Marshall, Pilot Station, St. Mary’s, and Mountain Village ), 9 sites on the Kuskokwim River (Bethel, Lower Kalskag, Upper Kalskag, Aniak, Chuathbaluk, Stony River, Sleetmute, Red Devil, Crooked Creek), 1 site on the Tanana River (Whitestone), 3 sites on the Copper River (Gakona, Copper Center, Chitina), and 1 site on the Susitna and Talkeetna Rivers (Talkeetna). In addition, there was one site in a tidal inlet (Teller). Due to a lack of pre-existing data with which to do an assessment at the selected sites, a UAA team traveled to each site and collected data on river bathymetry, river bank topography, velocity and discharge, water surface elevation and river slope. Based on the field measurements, hydraulic and hydrologic models were developed for each village site. Based on the models, the spatial distribution of depth-averaged velocity and hydrokinetic power density (PD, W m-2) was calculated (equation 1). 𝑃𝐷= 𝜌2 𝜀𝑉3 1 where ρ is water density (kg m-3), 𝜀 is device efficiency, and V is velocity (m/s). In this report, the efficiency (also referred to as the coefficient of performance) was taken to be 0.30. The calculations of velocity and power density were made for the 25, 50, and 75 percentile flows for the open water period as well as for the flow rate that prevailed at the time of measurement. Model calculations of velocity distribution were validated by comparing calculated and measured water velocity for the flow conditions that prevailed at the time of measurement. The comparison of measured and calculated velocities allowed for an estimate of the uncertainty in the velocity and in the power density calculations. A key hydraulic equation used throughout the project was the Manning Equation, which is presented below in two forms: 8 𝑉= 1𝑛𝑅2/3𝑆1/2 or 𝑄= 𝐴𝑛𝑅2/3 𝑆1/2 2 where 𝑉= cross-section averaged velocity (m/s); 𝑛= Manning roughness coefficient (s/m1/3); 𝑅= hydraulic radius (cross-sectional area/wetted perimeter, m); 𝑆= slope; 𝑄= discharge (m3/s); and 𝐴= cross-sectional area (m2). The Manning Equation directly relates the velocity or flow rate to the river cross-section dimensions and slope. 1.2 Objectives and Tasks Successful completion of the project required that a number of objectives be met and tasks be undertaken. They included: 1. Reach out to villages, governmental organizations, and the hydrokinetic community to obtain a list of sites that could potentially be developed to generate hydrokinetic energy. 2. Plan the trips to the village sites in order to minimize the overall project expenses, ensure that project tasks are conducted safely, and maximize community involvement in the project. The critical components included finding a suitable boat and boat operator, and finding a local contact, who could help the team find lodging and logistical support while at the village. 3. Design a system of instruments for collecting the data necessary for the assessment. 4. Prior to heading out into the field, select 6 to 10 cross-river transects that should be surveyed for bathymetry, bank topography, and velocity. Make adjustments in the field as necessary. 5. At each village site, set up a system of temporary survey benchmarks with which RTK-GPS surveying can be conducted. 6. Select a team of researchers including personnel for the hydrographic surveying and for the land surveying. Usually, the hydrographic surveying team will include a UAA researcher with expertise in Acoustic Doppler Current Profiler (ADCP) operation and the boat operator. The land survey team will include a researcher with expertise in RTK-GPS surveying and an assistant. The land surveying team (in particular) needs to be aware of their environment at all times and be prepared for potential bear encounters (with bear spray). The boat operator needs to meet university certification requirements. 9 7. Collect bathymetric, topographic, and velocity data along each cross-river transect. Run one or more down-river transects using RTK-GPS equipment to record slope of the river surface. Deploy a water level sensor (e.g., a HOBO) from the river bank near one of the cross-river transects. Arrange to have someone return the HOBO to UAA after one or two months so that a record of water level at the site is obtained. 8. Post-process the survey data including obtaining OPUS solutions and integrating the bathymetric and topographic data using Trimble Geomatics Office. Use velocity and survey data and “Winriver” software to compute the discharge across each transect. 9. Use the Manning Equation in conjunction with data on discharge, river profile (i.e., cross- section), and river slope to determine the Manning roughness coefficient at each cross-river transect. 10. Use the water level data (from the HOBO) to estimate the discharge (for the period of time that water level data is available. This was done using the Manning Equation in conjunction with data on the river profile, river slope, and knowledge of the Manning roughness coefficient. 11. Identify the nearest USGS gauge, account for the time of travel between the USGS gauge and the village site, and determine the ratio of discharge at the village site to discharge at the USGS reference gauge. This ratio was referred to as the Gauge Factor and it allowed the use of the long term USGS gauge data as a proxy for the village gauge data. 12. At each village site, use long term discharge data at the reference USGS gauge site, in conjunction with the estimated time of travel between the gauge site and the village site, data on the beginning and end date of the open water period (Appendix C), and the gauge factor to compute a cumulative distribution function (CDF) for open water discharge at the village site. Based on the CDF, determine the 25, 50, and 75-percentile flows at the village site. 13. Develop a 2D (depth-averaged) hydraulic model of flow for each village site. The hydraulic model CCHE (developed by the National Center for Computational Hydroscience and Engineering at the University of Mississippi), was used. The model was built using the bathymetric and topographic data collected during the site visits. The model was “forced” using discharge data and using data on water surface elevation at the inlet and at the outlet of the model domain. Model calculations were validated by running the model with the discharge and boundary water surface elevation data from the site visit and comparing calculated velocity with measured velocity. 14. At each site, use the validated CCHE model to calculate the velocity and power density in the river segment (i.e., model domain) for the 25, 50, and 75-percentile flows. Use the Manning 10 equation and the river profiles at the inlet and outlet of the study area to determine water surface elevation at the inlet and outlet. 15. Produce a “snapshot” of the statewide distribution of the in-river hydrokinetic resource based on a number of representative cross-river transects at each village site and based the cross- section averaged velocity at the 50-percentile (median) flow rate (for the open water period). 1.3 Methodology 1.3.1 Identifying candidate village sites for resource assessment The UAA team worked to identify candidate village sites for resource assessment. Dr. Nyree McDonald, formerly an Assistant Professor for the UAA-SOE, attended the Galena Renewable Energy Conference on April 2nd through the 4th of 2009 to establish local relationships and identify villages with an interest in developing hydrokinetic energy. At this conference, flyers were circulated informing representatives for each attending community of the intent of the research. In addition, surveys were distributed at the Galena Conference and at other conferences in an effort to characterize the interest in hydrokinetic energy. This survey covered such topics as the local energy cost, level of interest, and possible concerns of the community. These surveys coupled with the Galena conference and the rural connections that were forged gave the research a point of origin. A list of prospective research sites was compiled and evaluated, based on but not limited to: the site’s proximity to a power grid, anticipated river depth and velocity, the local need for alternative energy, and general community interest. 1.3.2 Logistics of travel to village sites The UAA team realized early on that it would be most efficient and cost effective to pick sites that were sequentially located along a river that was easily accessible by boat. Initially, Erin Eggleston, a Research Technician for the UAA-SOE, had correspondence with Jody Malus of the Bethel Alternative Energy Council to begin making arrangements for studies on the Kuskokwim River. Another key contributor to the logistical process for the Kuskokwim project was David Griso, the Executive Director of the Kuskokwim Watershed Council. He was instrumental in the selection process for locations of potential research, offering his recommendations as well as providing the necessary contacts for communication in the various villages. 11 The inflated expense of travel to rural Alaska made it necessary to make as few roundtrips to and from Anchorage as possible. The result was an effort to orchestrate arrangements to visit 6 sites from Bethel to Napaimute over a span of 14 days, 7 sites from Galena to Holy Cross in 16 days, and 4 sites from Marshall to Mountain Village in 8 days. Not only was this the most economical solution for personnel transport, but also for the 1000 plus pounds of equipment that the project required. Because of the relative inaccessibility of many of the sites, the teams’ main cargo handler, Northern Air Cargo (NAC), could only deliver to certain sites on certain days of the week. This played a vital role in determining the order of site visits, as it was necessary for the equipment to be awaiting the arrival of the research team. Other options for equipment transport were considered and utilized as well. Knowing that the research team was going to visit Marshall, AK after a short return to Anchorage, Teddy Heckman, the vessel operator for the Lower Yukon research, was contracted to transport equipment from Holy Cross to Marshall. This proved a very efficient and economical decision as opposed to shipping the gear back to Anchorage, as NAC does not directly operate through Marshall. Once the project sites were determined, it was necessary to begin arranging for accommodations, travel arrangements, and vessels with skilled operators to perform survey work and provide transportation. The key requirements for accommodations at each site included relative close proximity to the project site (the river) and electricity for charging the numerous batteries that powered the instruments. The high value associated with the equipment being used necessitated proper storage on a nightly basis. Hand carrying large volumes of equipment to and from lodgings each day was a last resort option, as it was time consuming and generally inefficient. For this reason, it was necessary to also secure ground transportation within each village, where possible. In addition to accommodations, travel between the project sites was considered. Due to the high volume and weight of the equipment necessary to perform the research, flying between sites had to be minimized, for cost considerations. This meant ensuring that travel by boat was obtainable to shuttle equipment from village to village, in the required time frame. Ideally, the aim was to utilize a minimum number of vessel operators, thus using one boat and captain for as many consecutive project sites as possible. Not only does this ensure that the research team has reliable transportation available, but it reduces the necessary training time involved in acquainting each operator with the navigational system and the overall process of collecting bathymetric data. Ultimately, it was necessary to use air transportation for personnel and equipment between Kaltag and Grayling, because the distance between the two villages made boat travel impractical and uneconomical. 12 1.3.3 Equipment High precision equipment is necessary to accurately measure river velocity and surface elevation, bathymetry, and topographic data. River velocities and bathymetry were measured using an ADCP. Navigation and positioning was conducted using Real Time Kinematic GPS, as was the upland topography of the river banks. The integration of these systems requires researchers to become well acquainted with equipment configuration settings, as well as various data interchange formats. Equipment trials were conducted prior to field work to ensure equipment compatibility and operator experience. 1.3.2.1 Acoustic Doppler Current Profiler The UAA team decided to use an Acoustic Doppler Current Profiler (ADCP) to collect bathymetric data and to collect velocity data. There are several ADCP’s available on the market that all promise accurate readings when used according to their respective design specifications. There are subtle differences in each manufacturer’s design that may suit the varied needs of the research to be conducted. For this particular hydrokinetic assessment, two ADCP models offered by two separate manufacturers were considered: the SonTek M9 RiverSurveyor and the Teledyne RDI Rio Grande Workhorse. The SonTek M9 River Surveyor features nine transducers, with a dual 4-beam 3.0MHz/ 1.0MHz Janus arrangement. The remaining transducer is a .5MHz vertical beam. This instrument claims to have a velocity profiling range from 0.06m to 30m, with a resolution of 0.001m/s. In addition, it has a depth measurement range of 0.20m to 80m, with a resolution of 0.001m. It has an internal memory capacity of 8GB and is available with an optional SonTek RTK GPS package, thus eliminating the need for more costly equipment, when the use of GPS is necessary. Teledyne’s 600kHz Workhorse Rio Grande ADCP has a single 4-beam Janus transducer array that is capable of measuring velocity profiles between .7m to 75m with a comparable resolution of 0.001m/s. It is capable of making depth measurements up to 100m for typical river water conditions. However, the Rio Grande has only 2GB of internal memory and does not come with the RTK GPS option. The final decision on which ADCP to choose was not determined by the available options that each instrument offered, but rather the instrument’s availability. SonTek could not deliver the Figure 1: Teledyne RDI Rio Grande ADCP (Teledyne, 2009). 13 instrument within a two month timeframe, whereas Teledyne RDI could have the instrument assembled and shipped in a matter of weeks. Due to the anticipation of high sediment conditions in the Alaskan rivers, the 600 kHz ADCP was preferred over the 1200 kHz model that Teledyne RDI offers. This is due to the fact that the lower frequency ADCP’s have a greater capacity to penetrate water with a high concentration of suspended particulates. There are, however, a number of drawbacks to the lower frequency model. Although the lower frequency ADCP will measure deeper than the high frequency models, they require larger depth cells and have a larger blanking distance. This may become inconvenient in waters that are relatively shallow (less than 1m). The Rio Grande features a user specified input which allows for the modification of critical settings for data acquisition when default settings result in unsatisfactory results. This input, the WS mode, modifies the bin size that the ADCP will collect while the instrument is pinging. River conditions in which the water is relatively deep will require a much larger bin size than shallow water. As a consequence, choosing the proper WS mode is very site specific. Rivers with long sweeping banks followed by deep channels are especially challenging. In order for the ADCP to collect discharge measurements in the shallow banks, the bin size must be relatively small. Conversely, the deep channel requires a greater bin size. Since the Teledyne Rio Grande does not have the capacity to automatically adjust the bin size during a transect, often several test transects must be completed in order to determine the appropriate WS mode. If the bin size is too small, discharge measurements will not be recorded in the deeper portion of the channel. Ultimately, it was the responsibility of the ADCP operator/researcher to ensure that the data collected is of the quality that is required. 1.3.2.2 Global Positioning System Relative positioning of the hydrography and upland topography were critical in accurately defining the river cross-sections and geo- referencing the velocity data. Originally, several pieces of survey equipment were considered for positioning and navigation. A survey level was ruled out. Although it can be quite accurate for vertical control, it does not collect horizontal data nor provide positioning for the hydrographic work. A total station was also considered for the project, but for the volume of work to be completed in a small window of time, it was deemed inadequate. In addition, neither of these instruments allows their direct integration into the ADCP. All positioning data for this assessment was completed using Trimble 5700, 5800 and R8 models, utilizing Real Time Kinematic (RTK) GPS mode. The 5700 and the 5800 receivers as well as the necessary ancillary equipment was provided by the Geomatics Department of the Figure 2: Trimble 5700 GPS Receiver (Trimble, 2009). 14 UAA-SOE, while the R8 model was purchased for this project. For the required precision needed for this project, standalone GPS was not an option, as this mode only provides for accuracies typical of handheld GPS receivers. RTK systems use a single base station receiver and a number of mobile roving receivers. The base station re-broadcasts the phase of the carrier that it measures, and the mobile units compare their own phase measurements with the ones received from the base station. This differential in carrier phase allows for accuracies on the order of a few centimeters both horizontally and vertically. In addition to their precision capabilities, integration of RTK GPS was desired for ADCP operations. By streaming the National Marine Electronics Association’s (NMEA) 0183 string into the ADCP software, it was possible to geo-reference all the collected bathymetric and hydrographic data to the upland topography. This was imperative for interpreting data at a later time. Although integrating the GPS receivers with the ADCP is not always necessary, certain flow conditions make it preferable. When not using GPS as navigational reference, the ADCP computes observed velocities using “Bottom Tracking” (BT). As a consequence of the instrument vessel moving as the ADCP records measurements, the software must subtract the speed of the boat to obtain accurate velocity measurements. Under some conditions BT is an acceptable method for this process. However, if the river bed is dynamic (moving bed), the vessel velocities observed by BT will be biased low, resulting in inaccurate water velocity measurements. Due to the uncertainty of such conditions along the Yukon and Kuskokwim Rivers, it was decided to err on the safe side and integrate GPS into the ADCP software, using the GGA data as a reference to accurately compute flow velocities. 1.3.2.3 Hypack - navigation A majority of the transects were pre-planned in Anchorage using geo-referenced tiff images. This allowed the research team to target the areas where the highest flow velocities were anticipated. These rivers can easily surpass one mile in width, requiring a navigational aid to stay true-to-course. HYPACK was selected as the navigational tool and was integrated with the GPS rover receiver to provide RTK quality navigation of the vessel. Although HYPACK is typically used for port and harbor applications, it is a very versatile program that works exceedingly well for its intended use for this project. As was the case with most of the GPS equipment, the UAA-SOE has access to a version of this software, making it the most economical option for navigation. Use of this software is instrumental in ensuring that the vessel operator does not venture far from the pre-planned transects, as it provides a real time location of the vessel and its proximity to the transect being studied. Not only does this Figure 3: Trimble R8 GPS Receiver (Trimble, 2009). 15 streamline the collection of bathymetric data, it also ensures that the data is as close to a linear transect as possible. 1.3.2.4 HOBO water level gauges In order to monitor the discharge at the village sites for a significant portion of the open water period and to determine the relationship between discharge at the village sites and discharge at a reference USGS gauge site, the team deployed HOBO water level sensors at the majority of the sites studied. The procedure for estimating discharge based on water level is described in section 2 below. A total of 15 HOBO gauges manufactured by Onset were set during the summer of 2009. Gauges were not deployed at Pilot Station and Upper Kalskag. Initially, each gauge was inserted through the end of a 25-30 foot PVC stilling well that was assembled at the project site. The gauge was tethered to a stationary object on the river bank using vinyl coated non-stretch steel wire. Typically, a large angular rock was tied to the end of the stilling well meant to be submerged in the river, so as to resist movement throughout the data collection period. The goal of the process was to deploy the device in an area where it was unlikely to be disturbed, and in an area of sufficient depth so as to record data for the entirety of the season. Being that these rivers are dramatically dynamic, their water level may drop as much as 20 or more feet from the spring months to the beginning of ice formation. Reports from various communities began streaming in that there was very little water covering the stilling wells as early as late July and the method above was deemed insufficient. A revised method was designed to deploy these gauges involving creating a “basket” using PVC as a frame. The PVC was then wrapped with wire mesh and the basket was filled with rocks. The water gauge was then attached to the outside of the basket and the entire apparatus was tethered to shore using the same wire as mentioned above. This made deployment much easier, as it enabled the team to drop it from the side of the boat some distance from shore, ensuring sufficient depth for deployment. The basket design also significantly reduced the amount of PVC being transported to the sites, and proved to be a much lighter, efficient system. At each site, contacts were made for retrieval of these devices before ice formed, which would make recovery unlikely using the described methods. Shipping materials including padded shipping boxes and prepaid shipping labels were given to each volunteer who were then advised on the methods for retrieval and told to await word from the research team for when to remove the gauges. This process was overall the most cost efficient for the water level data that is required. Figure 4: HOBO Water Level Logger manufactured by Onset Computer Corporation (Onset, 2009). 16 1.3.2.5 Laptop In order to integrate all three systems (GPS, HYPACK, ADCP), it was necessary to purchase a computer that was adequately robust for field work as well as possessing a parallel port. HYPACK requires that a hard-lock key be inserted into the parallel port of the computer being used for the software to run. Unfortunately, most modern laptops no longer come equipped with this feature. In addition to the parallel port, the field computer had to possess serial, VTG, and USB ports. With the need of all of these features, it made it necessary to purchase a port replicator to satisfy all of these requirements. The VTG port was essential to enable the use of a second monitor for vessel navigation. HYPACK was displayed on the second monitor for the vessel operator to navigate the preplanned transects, while the researcher simultaneously monitored the status of the ADCP. The serial port was required for the uplink to the ADCP, to enable data acquisition. Finally, the 2 USB ports were required to stream the GPS data into HYPACK and WinRiver II simultaneously. Due to the nature of the work, it was necessary to purchase a Panasonic Toughbook, which is both impact and water resistant. The process of shuffling equipment on a daily basis as well as the fact that work was being performed mostly in open hull vessels on open water necessitated these features. 1.3.2.6 ADCP Mount When deploying an ADCP there are several options available. One method for data collection is the bow-mounted approach. This method puts the instrument at greater risk, being as it will undoubtedly be the first thing to hit bottom when traversing a transect. This consideration along with the fact that the bow-mount mount design would be relatively complicated encouraged further alternatives. Another possible method is the stern-mount concept. This method was quickly abandoned for two reasons: (1) The cavitations created by the prop on the motor was likely to interfere with data collection and (2) the battery for the vessel can potentially cause magnetic interference, thus fouling data. The option that seemed to best suit the needs of the research team was the side-mount deployment. Figure 5: Panasonic Toughbook (shown without port replicator) (Panasonic, 2009). 17 After deciding on a concept, research revealed that the mount was required to be made entirely from non-ferrous material, as to not create a local magnetic field that would interfere with the instrument. The design was conceived by the research team and was fabricated free of charge by the UAA Welding Department. The mount featured a telescoping cross beam that allowed for adjustment to vessels of a broad range of widths. This was an essential feature, as the research team was to use several different vessels throughout the course of the project. The ADCP was mounted to the bottom of a vertical post that was joined to the cross beam via a “T” fitting, and was allowed to slide freely for both easy deployment and removal. In addition, a Zephyr GPS antenna was co-located on a length of all-thread fitted to the top of the post to reduce the need for measuring horizontal offsets for post processing considerations. Finally, the entire apparatus was fixed to the gunwales of the vessel via two large clamps and a tag line running from the top of the vertical post to the stern, to resist rotation. This design proved to be very reliable and was quite cost efficient. Figure 6: UAA-SOE ADCP side mount fitted to research vessel (Yager, 2009). 18 Figure 7: Construction schematic of ADCP boat side mount (Harper, 2009). 19 1.3.4 Planning the data collection effort Aside from coordinating lodging, transportation, etc., there was a great deal of time spent analyzing aerial photography of the selected sites. Given the fact that very little, if any data existed for these sites, decisions had to be made as to where the highest velocities were likely to be observed. Ideally, it would be most beneficial to plan transects very close together in order to capture a denser set of data. However, a process such as this would require a great deal of time, and much more than the research team would be able to spend at each site. Following this realization it was determined that the aim of the work should be to cover a longer reach of the river, thus producing more generalized results. One of the general criteria for choosing a set of transects for each site was that they had to be in relatively close proximity to one another. This was mainly due to the fact that the GPS base station radio has a limited range to which it can broadcast GPS corrections, roughly 1.5 to 2 miles with limited obstructions. Using the aerial maps, obvious locations for problematic transects were immediately ruled out based on this criteria. Another criterion for choosing transects was that they also had to be relatively close to the actual village, for transmission line considerations. At many locations, islands and sand bars are clearly visible from the imagery and where possible, placing transects that traversed straight through these obstructions was discouraged, mostly to ensure that the team maintained a signal from the base station. This is not to say that these types of transects weren’t decided on, because at several locations there was no other option. In order to collect the densest set of data, often several river channels were surveyed. On occasion, transects were placed based on the recommendations of the local townsmen, as they were more familiar with the sites. After relative locations for transects were decided, there was discussion as to how many could be traversed in the time allotted for each site. Taking into consideration setup time and allowing for unforeseen complications, it was determined that 7 to 10 transects could conceivably be completed in a single work day, while yielding the desired quality. With the transects drawn onto Figure 8: Preplanned transects for Aniak, AK (UAA-SOE, 2009). 20 the geo-referenced tiff images, the coordinates could then be keyed into the HYPACK software for navigation at each site. There were occasions when the preplanned transects were either impossible to complete or finished ahead of schedule. In the event that transects were finished early, it became the responsibility of the research team to determine locations for additional transects. As was often the case, new transects were placed between the preplanned lines to collect a more dense set of data. Some reasons for the decision to abandon certain transects included water too shallow for boat navigation, loss of base station radio contact, and inclement weather that made conditions unsafe on the river. Also considered during the preplanning phase were the potential locations for the water level gauges. This was difficult to plan from Anchorage, as there was little data available for review that would potentially aid this process. Ultimately, the placement of the HOBO’s was at the discretion of the research team while in the field and the decision was made after evaluating each project site. During the preplanning phase the research team investigated the location of any and all NGS (National Geodetic Survey) monuments that had been placed at the various sites. The hope was to utilize these control points as a location for the base station, being that their location was well documented and of control quality. Upon arriving at each site, it quickly became obvious that most if not all of the NGS monuments would not aid the research team. Most of the control points were located at great distance from the project site (the river), thus making it an impractical location for a base station due to the range limitations of the base station radio. One particular monument in Aniak that was potentially within range of the project site was located beneath heavy tree cover, making it virtually unusable for GPS applications. As part of the control search, researchers obtained magnetic declination adjustments for each site that was to be visited. This was important for the ADCP measurements as the internal compass of the instrument must be calibrated at each site. Additionally, at the start of each measurement WinRiver II (software for controlling the ADCP equipment) prompts the user for a local magnetic declination as to accurately interpret the vessel course as well as the direction of river flow. 21 1.3.5 Equipment trials When integrating various survey systems, it is paramount to test all equipment before any field work is commenced. This is even more important for research in rural Alaska, as replacement parts generally will have to be flown out to the site, which is very costly and inefficient. With the added uncertainty of communication with technical support, it is essential to become as familiar as one can be with the various systems in order to significantly reduce the need for such measures. As preplanning operations were taking place, the survey team began testing all the various GPS components from wires to receivers. Researchers spent several days on the phone with technical support or in the school parking lot running tests and familiarizing themselves with the equipment. This was instrumental in both troubleshooting procedures and equipment malfunctions. Several times throughout the course of the research the survey team was required to rely on knowledge obtained from these very trials, undoubtedly saving both time and money. The next task for the research team involved integrating the GPS data with both the ADCP software and the navigational software. Using a base station located on the roof of the UAA Engineering Building, the team first tested the navigational software. This was done by attaching a Trimble 5700 receiver and antenna to a rolling cart serving as the “vessel”. The HYPACK software could then verify its position at a certain point, which was then compared to the land survey team’s RTK rover. This allowed the team to be certain that the navigational software was accurate and could be used as part of the project. Being that none of the research team was familiar with the Rio Grande ADCP, it was decided that testing the device in the UAA pool was possibly the best way to become acquainted with both its hardware and software. A mount was specifically constructed for this test which would allow a research member to pull the ADCP from one end of the pool to the other. At the same time another member wheeled a cart with the battery and computer for the ADCP alongside and ensured that data was being collected and that all systems were running properly. This test was very useful for the researchers performing the bathymetric survey, as it acquainted them with the setup and user commands for the ADCP. Figure 9: Research members prepare the ADCP for a trial run in the UAA pool (Yager, 2009). 22 Finally, it was necessary to test all of the various components together as they would operate during the field work. The team first considered conducting a full scale test on a local Anchorage lake, but reconsidered due to the fact that no flow would be measured, nullifying one of the objectives to the equipment trial. The Knik River was deemed an appropriate test location for the scope of the research study, as it was the largest and most accessible river for the team. On June 5, 2009 the team set out with a 14 foot inflatable Zodiac and all the necessary equipment to perform a trial survey on the Knik River. The land survey team set up two monuments, one on either side of the Glenn Highway Bridge, and performed a static survey. At the same time, the hydrographic survey team fitted the vessel with the ADCP mount and all the navigational hardware, as well as the ancillary equipment needed to carry out the research. While the team was awaiting the results of the static survey, one of the HOBO water gauges was placed to ensure its proper functioning, as well as the function of the stilling well that it was housed in. The bathymetric research members quickly realized that something was not operating correctly with the ADCP. After exhausting all other troubleshooting methods, the manufacturer was contacted. The information that was conveyed to the research team was that WinRiver II, the ADCP software, has a default bin size setting for the depth at which measurements will be taken. Since the Knik River is relatively shallow (4 to 5 feet), custom user commands had to be input which would shrink the bin size down to a point where it would begin reading current velocities. This piece of information proved useful throughout the summer, as many of the sites had long, sweeping deposit banks that could not have been surveyed using the default setting. Overall, this Figure 10: Knik River equipment trial on June 5, 2009 (Mills, 2009). 23 trial run proved to be an enormous step forward and was very valuable for the research that was conducted throughout the summer. 1.3.6 Data Acquisition Upon arriving at each site, the first consideration was always where to locate the GPS receivers for the static survey so that not only did they have a clear view of the sky, but also a clear view of a relatively large reach of the river. Additionally, it was favorable to locate the receivers at a distance of several hundred meters apart, to achieve the longest baseline possible. For monuments at each site, two 2 inch aluminum rebar caps similar to the one shown in Figure 11 were used. The monuments were labeled with the project name, followed by the site and monument number and ending with the year. This method of monument was used because it required the minimum amount of materials when compared to digging holes and filling them with concrete as is needed for brass monuments. With the rebar driven and the caps placed, the GPS receivers were set up over the monuments and the static survey began. With the receivers mounted on their respective tripods, they were then leveled and the antenna heights were measured and recorded. For each control point, a static survey log sheet was filled out documenting the time, monument name, and approximate location, as well as several other required fields of information. With the receivers acquiring their position fix from the satellites over head, they were left to collect data for a minimum of two hours, although typically they were allowed roughly four hours to observe. Figure 11: 2 inch aluminum cap set flush to ground on rebar at Chuathbaluk, AK (Harper, 2009). 24 While the GPS receivers were conducting the static survey, typically the team then evaluated the most appropriate location for placement of the water level gauge. The gauge should be located relatively close to the project site, as to provide pertinent water level data to be used in post processing. Additionally, the location of the gauge must be well documented by taking a survey shot at its location immediately after deployment, as well as noting the time at which it was deployed. This survey shot will then serve as a reference water level for further analysis, to be conducted at a later time. When launching the HOBO prior to deployment, the individual whose task it was typically took a screen shot of the launch screen, not only as a time reference for post processing, but to ensure that all the required fields in the launch window were properly filled out. With the HOBO ready for deployment, the stilling well used to house the instrument was assembled from various lengths of PVC tubing using slip couplings and PVC cement. After approximately 10 minutes, the cement was dry enough to insert the instrument and place the apparatus into the river. Figure 12 shows the HOBO apparatus used in Bethel, AK, located on a sea wall that skirts the Kuskokwim River there. Contacts were made at each site occasionally for the sole purpose of retrieving these devices. A detailed set of instructions for removing the gauge was given to each contact, as well as prepaid shipping labels and padded boxes, to send them back to Anchorage for processing. This method proved quite effective and surpassed the expectations for the safe return of the instruments, as there were only two that were lost over the course of the summer. With the water gauges set and the static survey completed, the receivers were then connected to the computer and their Figure 12: HOBO water level gauge deployed on seawall at Bethel, AK (Yager, 2009). 25 information was downloaded into TGO. After verification that the static survey was a success the team was able to proceed with the scheduled work for that site. As the land survey team began conducting the topographic survey, the bathymetric crew fitted the vessel with the ADCP mount and prepared all necessary systems for the intended work. With the mount in place, the ADCP was securely fastened to the base plate of the mount via four nuts and bolts, paying close attention to make sure that the bolts were tight enough as to not risk losing the instrument during deployment. With the ADCP in place, the Zephyr GPS antenna was then screwed into its position at the top of the mount and all the proper cables were connected to their respective ports. At this point, the ADCP is lowered into position so that two important measurements can be taken. The distance from the water surface to the center of the transducer face on the ADCP must be measured and recorded so that the proper value may be entered when the configuration wizard prompts the user. Similarly, the distance from the water surface to the bottom of the GPS antenna must be accurately measured and recorded not only for river slope measurements, but for post processing considerations. All values are then recorded in the field book for quality assurance. After setting up the second monitor for HYPACK navigation, instrument calibrations could begin. Before the process of collecting transect data can begin, the ADCP must first run through a general diagnostic test followed by a compass calibration test. The diagnostic test ensures that all of the sonar beams are functioning properly as well as several other vital systems on the instrument. The compass must be calibrated at each site as well, for reasons Figure 13: Base station location at Kaltag, AK overlooking the Yukon River Project Site (Mills, 2009). 26 discussed previously. With these calibrations completed and after checking to make sure that the base station is communicating with both HYPACK and WinRiver II, a test transect can be run. The test transect provides several pieces of information that are used throughout data collection at each site. From this test, the researcher can observe the maximum depth, water speed, water temperature, and any problematic sandbars that may be present along the river. At the completion of this process, this data can be entered into the configuration wizard in WinRiver II and any user commands may be entered to adjust bin sizes where necessary. When beginning a transect it is of utmost importance to have the vessel operator maneuver onto the start of the preplanned line and maintain position in an area of sufficient depth until the ADCP has acquired 2 “good” bins. At this time, it is important to remain stationary with the ADCP pinging for at least 10 seconds observing those 2 bins so that the instrument can calibrate itself for flow conditions and make more accurate estimations of the discharge on the starting bank where the water is too shallow to take measurements. It is also necessary at this time to key in an “edge estimate” into WinRiver II. At that point in data collection it is merely a rough estimate. During post processing these estimates are corrected by using the topographic survey information coupled with the bathymetric data in TGO, yielding much more accurate outputs from the software. With this achieved, the vessel can then begin traversing across the river. In doing this it is equally important to monitor the vessel speed as indicated by WinRiver II to ensure that the boat is not traveling any faster than the velocities that are being observed by the instrument. Not only does this provide a more dense data set for a transect, but it makes velocity calculations more accurate in the software. Much like the process undertaken at the beginning of each transect, at the conclusion of a run the vessel must remain stationary for at least 10 seconds while observing 2 bins, as to better estimate the discharge in the section that cannot be observed by the ADCP. At the end of each transect WinRiver II also prompts the user for another “edge estimate” for the distance to the water line, and is handled precisely as stated above. It was typically the practice of the researcher to make two passes on a single transect to obtain an average for the sought values, as well as to identify any obviously erroneous data. After completing a transect and before moving on to the next, an ADCP log sheet is filled out and any notes were recorded for future review. Figure 14: River transects selected for Red Devil on the Kuskokwim River. 27 While the hydrographic survey is being conducted, the topographic survey was also being undertaken on the river banks. With the base station set over one of the two placed monuments, the survey team records an initial check shot over the unused monument. This will serve as quality check at the conclusion of the survey when the very same point is observed again and the two shots are compared and analyzed for any misclosures. With the initial check shot recorded in the data logger, the survey team set out to start observing on the river banks by staking out to the preplanned lines, first taking a shot at the water line to provide a reference water level and working up the banks from there. Typically 4 to 5 survey shots were taken for each bank on each transect observing any and all major grade changes and noting the height of the cut banks where present. This method was used throughout the summer and provided the topographic information that was needed. At each survey point, the name of the point and a brief description were recorded for cross-reference at a later time where needed. After completing one river bank, the survey team was shuttled across the river by the boat crew to begin work on the opposite side of the river. Where the depth of the river allowed, shots were taken in knee deep water in an attempt to complete the bathymetry that could not be recorded by the ADCP, due to shallow conditions. At some locations this was not possible due to the abrupt drop in river bed elevation, making it unsafe to wade into the river. Usually the final step in data collection involved “floating” the vessel downstream to acquire the data necessary for determining the river slope at each location. This was done by attaching the data logger to the Zephyr antenna atop the ADCP mount, keying in the vertical offsets, and setting it to record data points at 100 foot intervals. This was conducted over a distance of 7500 to 8000 feet with boat at idle speed to obtain enough data points to accurately determine the slope of the river. During this process, the occupants of the vessel were sure to be as still as possible as to maintain stability of the boat in the water so as to not disturb the height of the antenna during acquisition. With the topographic data collected, the survey team set out to locate landmarks and permanent structures to be used as control shots, in the event that a research team returned to the site in the future. This would allow any future data collected to be tied to the same coordinates and as well as serving as a check shot. Lastly, a final check shot on Figure 14: Survey team locating local permanent control monuments (Butterer, 2009). 28 the unoccupied monument set earlier was taken for quality assurance. At the conclusion of work at a particular site, the base station was taken down, the ADCP was dismounted and all equipment was stowed in its respective hard cases. Upon returning to the lodgings for the night, all GPS receivers and data loggers were downloaded and the information was backed up using an external hard drive. The ADCP data was likewise copied to the external hard drive and the various batteries used were plugged in to their respective chargers for work the following day. Throughout the process of both the land survey and the bathymetric survey, researchers were faced with many challenges. On the hydrographic side, long, sweeping, shallow deposit banks made it very difficult to set user commands that captured not only the shallow portions of the river, but the deepest portions as well. This can be overcome to a certain extent with the previously mentioned test transects, but ultimately it is left to the judgment of the researcher to determine an appropriate course of action for capturing the data that best represents flow at a particular location. Likewise, the land survey crew had many challenges on the river banks. Often times, the banks were nearly impassable due to very soft silty clay, making for very exhausting conditions and slow data collection. Also, the land survey team was occasionally forced to cut their own view of the sky due to a dense tree canopy. Additionally, a small number of transects had to be abandoned due to loss of base station radio contact as well as inclement weather that posed substantial risk to the equipment and research team. Using the methods described above, the research team was able to collect a wealth of information at each one of these sites, where little or none existed before. The following steps in the research involved the task of combining all the data and resolving it into a product that may be used for a variety of analysis. 1.4 Selected sites for data acquisition The Field data collection for this project took place on the shore and in the river at 31 locations. Table 1. Measurement sites. Site River Site River 1 Bethela Kuskokwim 17 Nulato Yukon 2 Lower Kalskag Kuskokwim 18 Kaltag Yukon 3 Upper Kalskag Kuskokwim 19 Grayling Yukon 4 Aniak Kuskokwim 20 Anvik Yukon 5 Chuathbaluk Kuskokwim 21 Holy Cross Yukon 6 Stony River Kuskokwim 22 Marshall Yukon 7 Sleetmute Kuskokwim 23 Pilot Station Yukon 8 Red Devil Kuskokwim 24 St. Mary's Yukon 9 Crooked Creek Kuskokwim 25 Mountain Village Yukon 10 Rampart Yukon 26 Talkeetna Talkeetna/ 29 Susitna 11 Stevens Village Yukon 27 Whitestone Tanana 12 Beaver Yukon 28 Gakona Copper 13 Tanana Yukon 29 Copper Center Copper 14 Napaimute Yukon 30 Chitina Copper 15 Galena Yukon 31 Tellerb Bering Sea 16 Koyukuk Yukon aBethel is a tidally influenced site and data analysis is ongoing as of the date of this report. bTeller Alaska was the one site village site that was cited on a coastal inlet as opposed to a river. Section 2 Data Analysis (post processing) 2.1 Control and Adjustments 2.1.1 OPUS Solution The initial post processing task was to obtain National Spatial Reference System coordinates for the survey control. This was accomplished by providing the Online Positioning User Service (OPUS) the static GPS data for each of the control points. OPUS is maintained by the National Geodetic Survey and provides highly accurate coordinates that are highly consistent with other users (typically within centimeters). OPUS uses the static GPS file to compute three independent baselines to the point from three Continually Operating Reference Stations (CORS). More information on the OPUS process and accuracy can be obtained at http://www.ngs.noaa.gov/OPUS/about.html. 2.1.2 Adjustments to upland topography OPUS will provide new coordinates for each control point. The coordinates for the control points were updated with the OPUS derived coordinates so that the RTK base station positions were adjusted. By fixing the RTK baselines between the base station and the rover positions, the rover positions were then adjusted to reflect the control adjustment. The adjustment process was conducted in Trimble Geomatics Office software and verified by checking the inverse between the updated control points and the updated control check shots performed in the field. The horizontal and vertical adjustments were recorded for adjusting the bathymetry data. 2.1.3 Adjustments to bathymetry The GPS data associated with the ADCP was collected in a separate software package (Winriver II) requiring manual adjustments of the positions of the bathymetric data. The ellipsoid heights and horizontal coordinates were adjusted at the GPS antenna phase center. Additional vertical 30 adjustments to include the antenna height above the water’s surface and the ADCP draft were computed. The measured depth from the sonar face to the bottom was deducted to assign an ellipsoid height to the sounding. The corrected bathymetry was imported into Trimble Geomatics Office to be combined with the upland topography. The compiled 3D bathymetry and topography were projected onto a best fit line to provide 2D cross sections for hydraulic analysis and modeling. 2.1.4 Determine river slope The river slope in the direction of river flow was determined by extracting the vertical and horizontal GPS data collected during the down river float. Outliers were deleted and a best fit line was used to develop a linear relationship between the change in GPS ellipsoid height and distance. 2.1.5 Determine the Manning roughness coefficient (n) The average roughness (𝑛) for each transect was determined using the Manning’s equation (Equation 2). 𝑄=𝐴𝑛 𝑅ℎ2 3� 𝑆𝑜1 2� 2 where 𝑄 is the measured discharge (m3/s), 𝐴 is the cross sectional area (m2), 𝑅ℎ is the hydraulic radius (m), 𝑆𝑜 is the slope and 𝑛 is the roughness (s m-1/3). The discharge used in the Manning’s equation was the discharge measured in the field for the particular transect. The cross sectional area and hydraulic radius were determined using the measured and corrected water surface elevation (stage) with the projected 2D profile. 2.1.6 Stage adjustments The river stage was observed using the HOBO pressure gauges. The measured absolute pressure was translated to river stage referenced to the ellipsoid. This was accomplished by first assigning a time synced reference elevation which was measured at the water’s surface next to the gauge using RTK GPS. Barometric pressure was downloaded from Weather Underground historical achieves and deducted from the absolute pressure so that the gauge readings were true hydrostatic pressures. The end result was stage measurements at the gauge location referenced to the ellipsoid. The stage readings were projected to each transect using the slope of the water’s surface and the distance from the gauge to the transect. 2.2 Modeling with the Manning equation 2.2.1 Modeling of discharge for observed stages Once the average roughness was determined for each transect, the discharge corresponding to the measured stages were computed using the Manning equation. A MATLAB program was created 31 to determine the river cross-sectional area and wetted perimeter based on the river profile and the river stage. The MATLAB program also determined the discharge based on the Manning equation for a given river stage. An example product of the MATLAB program is the plot of stage (i.e., river surface elevation) versus discharge for Transect 2 at Red Devil (Figure 16). Figure 16. Plot of river surface elevation (i.e., river stage) versus discharge at Red Devil 2.2.2 Modeling of discharge versus cross-section average velocity The MATLAB model was also used to output the relationship between cross-section average velocity (m/s) and discharge (m3/s, Figure 17). Within Excel, a power law relationship between velocity and discharge was developed. This finding was reported in the site reports as some energy companies (e.g., Ocean Renewable Power Corporation) found this type of data useful for their planning. This relationship was also used to determine the velocity and power density for the open-water-average discharge at each site, which is displayed in the plot of the statewide distribution (Figure ES-2). Figure 17. Cross-section average velocity versus discharge using the Manning equation and MATLAB model for transect 2 at Red Devil. 57.5 58 58.5 59 59.5 60 60.5 61 61.5 0 1000 2000 3000Elevation (m) Discharge (m3) Transect 2 velocity = 0.0698 Q0.3698 R² = 0.9999 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 1000 2000 3000Velocity (m/s) Discharge, Q (m3) Transect 2 Power (Transect 2) 32 2.3 Validation and application of models 2.3.1 Validating Model Output To validate the Manning equation-based, calculated relationship between river stage (H) and discharge (Q) (e.g., Figure 16), the calculations were compared to H versus Q relationships based on data from three USGS gauge sites (Kaltag and Pilot Station on the Yukon River and Crooked Creek on the Kuskokwim River). The calculated and measured H versus Q relationships were in agreement at Kaltag and Crooked Creek. For example, at Kaltag, during a time period in which measured (USGS) discharge ranged from about 500 to 17,000 m3/s, our estimated flow rates (based on our MATLAB program and the Manning Equation) differed from the measured flow rates by only 6.5% (on average). However, the calculated and measured H versus Q relationships were not in agreement at Pilot Station. The difference at Pilot Station is likely caused by non- uniform flow associated with the 180 degree bend in the river just upstream of the measurement site. The Manning equation assumes that the flow is uniform. Appendix A displays the calculated and measured H versus Q relationships at Kaltag, Pilot Station, and Crooked Creek. 2.3.2 Time of travel and gauge factor estimation In order to obtain statistical data on flow rate at the various village sites, an effort was made to determine the relationship between flow rate at the village site and flow rate at a nearby reference USGS gauge site (with a long record of data). The relationship between flow rates at the two sites was assumed to be of the form: 𝑄𝑣𝑖𝑙𝑙𝑎𝑔𝑒(𝑡)=𝐺𝐹 𝑄𝑈𝑆𝐺𝑆 (𝑡+𝑇) 3 where Qvillage = flow rate at a given village site (m3/s), t = time (days), GF = the gauge factor, QUSGS = flow rate at a nearby USGS gauge (m3/s), T = time of travel between the USGS gauge and the village. The time of travel, T, would be negative (or positive) if the village is downstream (or upstream) of the USGS gauge site. The time of travel was calculated based on the average velocity in the river between the village site and the USGS gauge and on the travel distance. The average velocity was computed based on the open-water-period average velocity at the village site, at the USGS gauge site, and at any other study sites in between. Average velocity was computed based on the average discharge during the open water period using the flow rate versus velocity relationships derived at the village sites (e.g., Figure 17). Once the time of travel was determined, the gauge factor (Equation 3) was adjusted to maximize the agreement of the left hand side and right hand side of Equation 3 for the period of time when data was available at the village site. Since rivers tend to be gaining rivers in Alaska, the gauge factor tended to be greater 33 than one when the village site was downstream of the USGS gauge site. Table 2, below, provides the study sites, the corresponding USGS reference gauge, the gauge factor and time of travel. In some cases, the rivers were braided and the UAA team only surveyed and computed discharge at only a fraction of the river channels. At Anvik, for example, the gauge factor was unusually low because the river was braided and only one of the channels was surveyed for discharge. Table 2. Information about the measurement sites including the reference USGS gauge, the gauge factor, the time of travel to the village site from the USGS gauge, median flow and velocity, and power density. Site River Ref. gauge GF Time of travel (days) Q50 (m3/s) V50 (m/s) Power density (from V50) (W/m2) Transects used in power density estimatec 1 Bethel Kuskokwim N/A N/A N/A N/A N/A N/A N/A 2 Lower Kalskag Kuskokwim Crooked Creek 0.97 -2.0 1,571a (partial) 1.10 200 1 - 4 3 Upper Kalskag Kuskokwim Crooked Creek 0.97 -2.0 1,699 a (partial) 1.07 184 1 - 6 4 Aniak Kuskokwim Crooked Creek 1.2 -1.0 1,608 a (partial) 1.25 294 3 - 6 5 Chuathbal uk Kuskokwim Crooked Creek 1.4 -1.0 2,681 1.44 448 N/A 6 Stony River Kuskokwim Crooked Creek 0.9 1.5 1,160 a (partial) 1.39 400 3 - 5 7 Sleetmute Kuskokwim Crooked Creek 0.93 1.0 1,590 1.01 155 1-7 8 Red Devil Kuskokwim Crooked Creek 0.98 0.0 1,956 1.19 255 1-7 9 Crooked Creek Kuskokwim Crooked Creek 1.0 0.0 1,875 1.16 235 1-7 10 Napaimute Kuskokwim Crooked Creek 1.02 -0.5 2,025 1.84 446 1-4 11 Rampart Yukon Stevens Village bridge 1.04 -0.5 5,738 1.74 794 1-7 12 Stevens Village Yukon Stevens Village bridge 1.0 0.5 5,148 a (partial) 1.31 336 1-7 13 Beaver Yukon Stevens Village bridge 0.63 1.5 3,452 a (partial) 1.26 303 6 14 Tanana Yukon Stevens Village bridge 0.86 -2 4,705 a (partial) 1.43 441 1-7 15 Galena Yukon Pilot Station 0.75 7 8,529 1.15 228 1-7 16 Koyukuk Yukon Pilot Station 0.76 7.5 8,311 0.88 102 1-3 17 Nulato Yukon Pilot Station 0.73 7 8,223 0.90 110 1 18 Kaltag Yukon Pilot Station 0.75 6 8,438 1.20 260 1-7 19 Grayling Yukon Pilot Station 0.86 4 9,707 1.08 189 1-7 20 Anvik Yukon Pilot Station 0.34 3,862 a (west channel only) 0.88 101 1-6 34 21 Holy Cross Yukon Pilot Station 0.9 +3.5 9,976 a (partial) 1.03 165 3-6 22 Marshall Yukon Pilot Station 0.72 0.5 7,800 1.07 183 1,2,4,5,6,7 23 Pilot Station Yukon Pilot Station 1.0 0 10,894 0.87 97 1-7 24 St. Mary's Yukon Pilot Station 1.02 0 11,112 0.92 119 1-7 25 Mountain Village Yukon Pilot Station 1.04 0 12,000 0.72 56 3-6 26 Talkeetna Talkeetna/ Susitna Talkeetna 0.84 0 166 a 1.53 541 6,7 27 Whiteston e Tanana Fairbanks 0.81 1 587 1.82 898 4-8 28 Gakona Copper Million $ Bridge 0.07 1.5 170a, upstream of Copper Center 2.15 1493 22 29 Copper Center Copper Million $ Bridge 0.157 1.5 500a 2.23 1663 6-12 30 Chitina Copper Million $ Bridge 0.232 1 750 2.24 1677 1-5 31 Tellerb Bering Sea N/A N/A N/A N/A N/A 40 T1 aComplex system. Please see the “Site Investigation Report” for more detailed information. bTeller Alaska was the one site village site that was cited on a coastal inlet as opposed to a river. The power density provided is based on an averaged of the measured velocity data. cSee Site Investigation Reports for locations of the transects at the various sites. To confirm that this approach for determining gauge factor and time of travel gave accurate results, an alternative approach was chosen using a village site (Galena) which had a large amount (70 days) of field-gathered flow rate data over a period of time in which the reference USGS gauge station (Pilot Station) was also gathering data. In the second approach, we allowed the time of travel to vary from 0 to 10 days and computed the optimal gauge factor (with that time of travel assumed). Further, we determined the absolute value of the error (difference between right hand and left hand sides of Equation 3 above) for each assumed time of travel (Figure 18). The estimated time of travel from the first approach (7.4 days) was very similar to the optimal time of travel from approach 2 (7 days, Figure 18). 35 Figure 18. Percent Error versus Time of Travel for Galena to Pilot Station. 2.3.3 Estimation of discharge, velocity, and power An analysis was conducted to estimate the cumulative density function (CDF) of discharge, velocity, and power density within the open water period at the various study sites throughout Alaska. The goal was to obtain an overview of the power density in the various rivers of Alaska (e.g., Figure ES-2). The following procedure was followed. First, the temporal extent of the open water period was obtained for each of the USGS gauge sites. Second, the daily average discharge data for each reference USGS gauge site was obtained for multiple years for the open water period. Third, the data set of discharge was analysed to obtain the probability density function (PDF) and cumulative density function (CDF) of discharge at the reference site. Fourth, the CDF of discharge at the study site was obtained by rescaling the x-axis (discharge) of the CDF of the reference site using the gauge factor. Figure 19 shows the CDF of discharge for an individual transect at Red Devil on the Kuskokwim. The vertical red line indicates the flow rate at which the extent of the measured bank topography was exceeded. Fifth, the CDF of cross-section average velocity at each measured transect in the study was obtained by using the Q versus V 0.0% 1.0% 2.0% 3.0% 4.0% 5.0% 6.0% 7.0% 8.0% 0 1 2 3 4 5 6 7 8 9 10 11Relative error (%) Time of Travel (days) Galena 36 Figure 19. CDF of discharge at a transect at Red Devil, Kuskokwim River. relationship for that transect (e.g., Figure 17). Figure 20 below shows some example CDF’s of cross-section average velocity at Red Devil on the Kuskokwim. The black line indicates the average CDF of the 7 river cross-sections studied. The red line indicates the velocity range at which the water level had risen above the level of the topographic data collected and estimated topographic data was assumed. The grey line indicates the CDF of cross-section average velocity of the 7th transect which had higher velocities than the average transect. Figure 20. CDF of the cross-section average velocity at Red Devil on the Kuskokwim. 0.0% 20.0% 40.0% 60.0% 80.0% 100.0% 0 500 1000 1500 2000 2500 3000P(Q) Q (m3/s) cdf Average Q, Open Water Season Transect Max Q 0.0% 20.0% 40.0% 60.0% 80.0% 100.0% 1 1.1 1.2 1.3 1.4 1.5P(V) V (m/s ) Average Velocity, Open water season Extrapolated range Max velocity; T7, open water season 37 The potential hydrokinetic power density (PD, W/m2) at a river location is defined by equation 4. 𝑃𝐷=0.5 𝜀 𝜌 𝑉3 4 where 𝜀 = device efficiency which is assumed to be 0.3, ρ = water density (1000 kg/m3), and V is the velocity (m/s). The power density is a measure of available power per square meter of turbine cross section. Figure 21, below, plots the CDF of hydrokinetic power density for Red Devil on the Kuskokwim River. The calculation is based on the cross-section average velocity in the various transects studied. The black line indicates the average CDF of power density of the 7 river cross-sections studied. The red line indicates the power density range at which the water level had risen above the level of the topographic data collected and estimated topographic data was assumed. The grey line indicates the CDF of power density from the 7th transect which had higher velocities than the average transect. Figure 21. CDF of the power density at Red Devil on the Kuskokwim. 0% 20% 40% 60% 80% 100% 100 200 300 400 500P(PD) PD (W/m2) Average power density, Open water season Extrapolated range Max power density; T7, open water season 38 Data on the representative median flow rate, velocity, and power density are provided for each site in Table 2 (above). 2.4 Numerical modeling of river hydraulics Numerical modeling of river hydraulics was conducted in order to provide the spatial distribution of depth-averaged velocity and power density at the 25, 50, and 75-percentile flows of the open river period, for each river site. 2.4.1 Data preparation for modeling The modeling process required the bathymetry and bank topography data, the average Manning roughness coefficient, the river surface slope, discharge at the 25, 50, and 75-percentile flows (i.e., Q25, Q50, and Q75), and corresponding water levels. To assist the modeler, all of the necessary data was pre-assembled in a spreadsheet for each site. In order to estimate model accuracy, measured velocities on the sites were compared with model results calculated for the flow rate that prevailed at the time of the site visit. However, due to noise and turbulence in the collected data, measured velocities from the site were processed in MATLAB prior to comparison to the model with the use of a Savitzky–Golay filter. The Savitzky–Golay method essentially performs a local polynomial regression on a series to determine the smoothed value for each point. The main advantage of this approach is that it tends to preserve features of the distribution such as relative maxima, minima and width, which are usually flattened by other adjacent averaging techniques, like moving averages (Figure 22). 39 Figure 22. Comparison of measured data and filtered data from one of the Red Devil transects. 2.4.2 Assessment of available hydraulic models The project considered three different software packages in the search of accurate calculations of velocity distribution. They were River 2D, CCHE 2D, and HEC-RAS. The models were compared based on ease of use, accuracy, and resolution of data, and the functionality of the software. 2.4.2.1 River 2D River2D is a two dimensional, depth-averaged model of river hydraulics developed at the University of Alberta by Dr. Peter Steffler. It is a finite element model which is based on the Galerkin weighted residual method. River2D can model both subcritical and supercritical flow, and it is capable of modeling wetting and drying. Ice covers with variable thickness and discontinuous ice covers can be modeled as well. River2D has been customized for fish habitat evaluation studies. 2.4.2.2 CCHE 2D The CCHE2D model is a two-dimensional, depth-averaged flow and sediment transport model. The flow model is based on depth-averaged Navier-Stokes equations. The turbulent shear 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 587750 587800 587850 587900 587950 588000Velocity, m/s Distance across river, m Measured Data Savitzky–Golay filter 40 stresses are modeled using Boussineq’s approximation, and three different turbulence closure schemes are available for the calculation of the turbulent eddy viscosity. The resulting set of equations is solved implicitly using the control volume approach and efficient element method. The numerical technique employed ensures oscillation free, stable solutions. 2.4.2.3 HEC-RAS The Hydrologic Engineering Center’s River Analysis System (HEC-RAS) was developed by the US Army Corps of Engineers. This software is designed to perform one-dimensional hydraulic calculations for a full network of natural and constructed channels. The basic computational procedure for steady flow is based on the solution of the one- dimensional energy equation. Energy losses due to friction and contraction/expansion can be accounted for. The momentum equation is used to estimate water surface profiles. HEC-RAS is capable of modeling subcritical, supercritical, and mixed flow regime flow along with the effects of bridges, culverts, weirs, and structures. 2.4.3 Assessment of available models The site data from Red Devil was used for the performance evaluation of the models (Table 3, below). A comparison of the model results and the measured data revealed that HEC-RAS has the most accurate estimation of cross-section average velocity. However, this program failed to accurately calculate velocity distribution across the channel. River 2D had an average error of 15.1% when estimating velocity distribution, compared with 13.5% for CCHE2D. The difference in error was considered insignificant. However, River2D models took a long time to run, and software mesh package required much more topography data in order to be able to generate a bathymetry file when compared to CCHE2D. Additionally CCHE2D has the lowest error in velocity distribution. The CCHE2D software also includes a sediment transport mode and mesh functions which were specifically developed for solutions when topography data for the site is limited to a few sections across river channel. Table 3. Summary of average computer modeling error relative to measured velocity data at Red Devil for three software packages and considering cross-section average velocity, and spatially distributed velocity. Error type HEC- RAS River2D CCHE2D Section average velocity 4.1% 7.3% 6.9% Velocity distribution across the section 19.0% 15.1% 13.5% 41 The comparison found that HEC-RAS significantly underestimates velocity magnitudes and that it took an excessive amount of time to convert data for visualization. River 2D has a well- developed mesh generation mode but it tends to overestimate velocity magnitudes. CCHE_2D offers a variety of data processing functions and a high level of performance. The decision was made to use CCHE_2D as the modeling software because of its running speed, accurate high resolution velocity output, and its ability to also model sediment transport from the same model. Examples of depth average velocity plots at Red Devil from River-2D, CCHE_2D, and HEC_RAS are available as Appendix B. 2.4.4 Model Verification at specific sites For specific sites, depth-averaged velocity from the ADCP was used to calibrate the numerical model. The CCHE_2D model required data on the water level on the upstream and downstream boundary as well as discharge data. Measured data from the site visit was used as input for the model. Following CCHE_2D model development, modeled velocity was compared to measured velocity for the flow conditions that prevailed at the time of the site visit. For example, Figure 23 below compares modeled and measured depth-averaged velocity at Transect 3 from the Red Devil site. Figure 23. Comparison of measured and calculated depth-averaged velocity at transect 3 at Red Devil. 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 586650 586700 586750 586800 586850Velocity Magnitude (m/s) Easting (UTM) ADCP Data CCHE Model Filter Error: 10.08% 42 Section 3 Results 3.1 Map of statewide power density distribution. In order to generate a map of the in-river, statewide hydrokinetic power density distribution, a number of representative transects from each site were identified. The median discharge (i.e., Q50) and the median velocity (i.e., V50) were determined for each transect following procedures described in section 2. Then, the average of transect-averaged velocities was computed and the power density was determined (using Eq. 4). Figure 24 is a qualitative depiction of the results. Quantitative data is provided in Table 2. Figure 24. Map showing distribution of power density in Alaskan Rivers. Dot color indicates power density based on the median cross-section-averaged velocity at the site in the open water period. See Table 2 for quantitative data on the distribution. 43 3.2 Power and velocity contour plots For individual sites, power and velocity contour plots were generated for the flow rate at the time of the site visit and for the 25, 50, and 75-percentile flow rates. For the modeling of velocity and power density at the 25, 50, and 75-percentile flow rates, the Manning equation was used to estimate the water surface elevation at upstream and downstream boundary of the model domain. Figure 25 provides example power and velocity contour plots from the Red Devil site. Figure 25. Plot of the spatial distribution of depth-averaged velocity at Red Devil, AK 44 Section 4: Error Analysis The principal products of this report are the statewide distribution of power density (Figure ES - 2) and the spatial distribution velocity, depth, and power density at individual sites (e.g., Figure ES-1). 4.1 Uncertainty in the calculation of statewide distribution of power density The calculation of the statewide distribution of power density (Figure ES-2, Table 2) was based on the cross-section-averaged velocity at selected transects at each site for the open water median flow rate. At each site the standard deviation (and the coefficient of variation) of the transect velocity was determined, finding an overall coefficient of variation of about 6.8%. Given that the power density was proportional to the cube of velocity, this translated to a power density coefficient of variation of about 26%. 4.2 Uncertainty in the calculation of spatial distribution of power density for the 25, 50, and 75 percentile flows of the open water period. Comparison of measured velocity distribution and modeled velocity distribution (with the CCHE2D model, Appendix B) indicates that the typical modeled velocity at a given location had an uncertainty (error) of 13.5% (Table 3). Given that power density is proportional to the cube of velocity, we estimate the power density uncertainty to be about 46%. Site specific velocity uncertainty (error) is available in the Site Investigation Reports. The Site Investigation Reports for the individual sites provide an estimate of error in the modeled velocity at the various transects. This velocity error estimate can be used to estimate the power density uncertainty for the various sites. 45 References EPRI (Electric Power Research Institute). 2008. System Level Design, Performance, Cost and Economic Assessment -- Alaska River In-Stream Power Plants. Palo Alto, CA. EPRI-RP-006-Alaska. October 31, 2008. EPRI (Electric Power Research Institute). 2012. Assessment and Mapping of the Riverine Hydrokinetic Resource in the Continental United States. Palo Alto, CA. 1026880. December 2012. 46 Appendix A. Assessment of Manning Equation. Figure A-1. Plot of discharge (Q) versus river stage at Kaltag based on the Manning Equation ( ) and based on USGS measurements ( ). Figure A-2. Plot of discharge (Q) versus river stage at Crooked Creek based on the Manning Equation ( ) and based on USGS measurements ( ). 0 5 10 15 20 25 30 35 40 45 50 0 10000 20000 30000 40000River Stage Height (m) Discharge (m3/s) Kaltag Transect 1 Q vs H USGS gage at Kaltag, AK 49 50 51 52 53 54 55 56 57 0 2000 4000 6000 8000River Stage Height (m) Discharge (m3/s) USGS gage at Crooked Creek, AK crooked creek transect 2 47 Figure A-3. Plot of discharge (Q) versus river stage at Pilot Station based on the Manning Equation ( ) and based on USGS measurements ( ). 0.00 5.00 10.00 15.00 20.00 25.00 0 5000 10000 15000 20000 25000 30000River stage Height (m) Discharge (m3/s) USGS gage at Pilot Station, AK Pilot Station transect 7 48 Appendix B: Assessment of hydraulic models. Figure B-1. HEC-RAS -calculated distribution of depth-averaged velocity at Red Devil. Figure B-2. River 2D -calculated distribution of depth-averaged velocity at Red Devil. 49 Figure B-3. CCHE2D -calculated distribution of depth-averaged velocity at Red Devil. Figure B-4. Interpolated measured depth average velocity data at Red Devil. 50 Appendix C. Average Dates of Open Water Period at Selected Alaskan River Communities Data compiled from National Weather Service Alaska-Pacific River Forecast Center. Original data available from http://aprfc.arh.noaa.gov/data/maps/brkup_map.php and http://aprfc.arh.noaa.gov/php/frzup/getavgfrzup.php Location River Average date of breakup Average date of first ice Number of Open water Days Percentage of year Noatak Noatak 20-May 4-Oct 137 38% Colville Colville 31-May 27-Sep 119 33% Buckland Buckland 18-May 27-Aug 101 28% Ambler Kobuk 18-May 5-Oct 140 38% Kobuk Kobuk 16-May 6-Oct 143 39% Shungnak Kobuk 18-May 30-Sep 135 37% Koyukuk Koyukuk 13-May No data Hughes Koyukuk 11-May 7-Oct 149 41% Allakaket Koyukuk 11-May 6-Oct 148 41% Bettles Koyukuk 10-May 30-Sep 143 39% Alakanuk Yukon 22-May No data Mountain Village Yukon 18-May 19-Oct 154 42% Pilot Station Yukon 17-May 20-Oct 156 43% Marshall Yukon 15-May No data Russian Mission Yukon 15-May 17-Oct 155 42% Holy Cross Yukon 15-May 20-Oct 158 43% Anvik Yukon 16-May 14-Oct 151 41% Grayling Yukon 13-May 13-Oct 153 42% Kaltag Yukon 14-May 15-Oct 154 42% Nulato Yukon 12-May 14-Oct 155 42% Galena Yukon 12-May 12-Oct 153 42% Ruby Yukon 12-May 12-Oct 153 42% Tanana Yukon 10-May 10-Oct 153 42% Rampart Yukon 12-May 13-Oct 154 42% Stevens Village Yukon 12-May 12-Oct 153 42% Beaver Yukon 11-May 12-Oct 154 42% Fort Yukon Yukon 10-May 12-Oct 155 42% Circle Yukon 9-May 17-Oct 161 44% 51 Eagle Yukon 5-May 21-Oct 169 46% Manley Tanana 3-May 10-Oct 160 44% Nenana Tanana 2-May 20-Oct 171 47% Fairbanks Tanana 29-Apr 16-Oct 170 47% Big Delta Tanana 3-Apr 11-Oct 191 52% Tanacross Tanana 14-Apr 9-Oct 178 49% Northway Tanana 23-Apr no data Gakona Copper 30-Apr 18-Oct 171 47% Gulkana Copper 29-Apr no data Napakiak Kuskokwim 12-May 21-Oct 162 44% Bethel Kuskokwim 12-May 16-Oct 157 43% Kwethluk Kuskokwim 11-May no data Akiak Kuskokwim 10-May 18-Oct 161 44% Tuluksak Kuskokwim 9-May no data Kalskag Kuskokwim 7-May 14-Oct 160 44% Aniak Kuskokwim 7-May 19-Oct 165 45% Crooked Creek Kuskokwim 7-May 7-Oct 153 42% Red Devil Kuskokwim 6-May no data Sleetmute Kuskokwim 5-May no data Stony River Kuskokwim 6-May no data McGrath Kuskokwim 7-May 9-Oct 155 42% Nikolai Kuskokwim 23-Apr no data Skwetna Skwetna 30-Apr 29-Oct 182 50% Sunshine Susitna 1-May 14-Oct 166 45% Talkeetna Susitna No data 19-Oct Talkeetna Talkeetna No data 14-Oct