Progress 09/01/18 to 12/31/21

Outputs
Target Audience:Anyone who utilizes or depends on snowpack water runoff can benefit from this research. This includes hydropower plants, boaters, fishermen, wildlife biologists, agricultural irrigators, municipal water supplies, and other industries. Over the course of this project NWB Sensors has worked with people from these groups and groups who collect the data and publish it for the broader public benefit. During the Phase II SBIR effort NWB Sensors engaged many potential customers and added upcoming product information to our website for their benefit. We reached out to or were contacted by researchers from four universities, and three non-profits. We also engaged Montana's local back country recreationalists and the avalanche forecasters that serve them who showed much enthusiasm over our technology, worked with a Montana ranch to measure snow water runoff, and we established working relations with other companies that serve groups like these. Some of these contacts were one-off, while a handful resulted in continuing exchange of information about our research progress. Most importantly we engaged the government agencies at municipal, state, and federal levels that operate large data collection networks to monitor available water for public use. Our two closest collaborators were the Natural Resource Conservation Service (NRCS) and the Montana Climate Office (MCO). The NRCS operates the federal Snow Survey Program in the western U.S. and allowed us to collect research data next to their operational sites with existing snow pillows, snow scales, and other sensors. We worked closely with staff in Montana, Oregon, Idaho, and Utah to coordinate data collection and validation efforts across a wide variety of geographical locations. We also kept the NRCS Snow Water Equivalency working group up to date on our efforts and disseminated data back to them and the states. The MCO also allowed us to co-locate prototype equipment at their operational sites, and regularly respond to inquiries about our research from them. Utilizing funding provided by the State of Montana to aid in commercialization and development of our snowpack sensing product we were able to send representatives to the 87th Annual Western Snow Conference in Reno Nevada, and to the 76th Annual Eastern Snow Conference Meeting in Fairlee, Vermont. NWB staff attended portions of the virtual 88th Western Snow Conference to share with attendees about our development and will continue to promote our system at these conferences. Changes/Problems:An objective of the original Phase II proposal was to attempt to retrieve snow depth directly from the algorithm. This would allow us to eliminate the need for an ultrasonic snow depth sensor to be part of our final product. Some of the meathods proposed were too complicated and expensive to empliment, and the derived snow depth product was very noisy and inacurate. Therefore, the final product of this research still utilizes the ultrasonic snow depth sensor. What opportunities for training and professional development has the project provided?This project has provided on the job training for three interns from Montana State University. This training has included knowledge of electrical design, software development, project management, and Snowpack measurement. In addition NWB was able to hire an aspiring snow scientist, and avid back country recreationalist who was trained to make manual snow samples, make snow liquid water content measuremetns, and collect snow pit data. Professional development has included learning and implementing coding practices and implementing project management software. In preparation for production of commercial projects NWB Sensors has been utilizing workshops from local industry groups on product production, product exports and Quality Management Systems. In addition to these workshops, we had bi-weekly training sessions on ISO-9001 quality system management. Financial support for this training has come from the State of Montana. How have the results been disseminated to communities of interest?Refer to the "Target Audience" section which already covers the potential customers and comunities we have reached. What do you plan to do during the next reporting period to accomplish the goals? Nothing Reported

Impacts
What was accomplished under these goals? The goal of this project was to create a low-cost, low-power, low-maintenance snowpack monitoring sensor to increase public knowledge about available snowpack meltwater availability and improve water management in areas that depend on snowmelt as a water source. The technology developed utilizes already existing hardware for widely used Global Navigation Satellite Systems (GNSS) like the U.S. operated GPS. Building on GNSS reduced development and manufacturing cost. At the beginning of this work, we knew that a residual signal could be found from the GNSS data that related to snow water equivalency (SWE) and snow liquid water content (LWC), but we did not know how to accurately determine SWE and LWC. Snowpack is a complex ever evolving structure of ice, air, and water that makes it difficult to measure by electronic means, and early snowpack data products were noisy and showed unknown offsets. We had to theorize what could be causing the errors, and how to remove them. Using Phase II SBIR funding we tested our theories. We tested different antennas, receivers, locations, and data processing techniques. Researchers traveled to field test sites, observed snow conditions, and collected truth data to validate the tests results. All this data then was used to find patterns that impacted the final SWE and LWC measurements. The result was a generalized electromagnetic model of snowpack at GNSS frequencies, an advanced algorithm relating GNSS data to snowpack properties, commercially ready field tough hardware, and a new sensor that measures frozen and liquid snowpack water content (SWE & LWC). Snow liquid content measurements have not been easily made in the past and this sensor will make this new LWC data widely available which could help avalanche forecasters identify wet slab avalanche potentials, or help hydrologist see when water will start flushing out of a pack. Below are summaries of the work accomplished on each goal: GOAL 1: Mature the LWC and SWE derivation algorithm. Major algorithm development activities included the final selection of the non-linear solver used and porting it from MATLAB to C++ to serve development and operational needs. The software improvements include preprocessing of incoming GNSS data, a new integer ambiguity solver, improved antenna location solver, using all available satellite constellations and frequencies, and improved data filtering. Architecture changes increased processing speed, decreased development cycles, and saved power at remote sites. Experimental changes could be processed quickly at all stations for all years avoiding tuning to a particular snowpack or location. We keep improving the algorithm handling of fringe or edge cases that make the final snowpack data product deviate from actual conditions. The algorithm was expanded to use multifrequency GNSS for two reasons. First, operation results showed the SNR drops below the level required for phase measurements during melt in deep snowpack for single frequency systems. Utilizing multifrequency data we were able to track phase longer in these packs. Second, multipath signals at forested measurement sites make signal to noise ratio (SNR) readings troublesome as reflections can increase and decrease SNR. Modeling the multipath environment proved difficult but using a no-snow period we were able to remove persistent multipath sources. Also, including more satellite constellations and multifrequency data helped filters remove outlier SNR data. The snow model now incorporates antenna phase center shifts. A parametric model of the antennas in use in our systems was studied using simulations to understand shifts in antenna gain pattern and tuning as the dielectric changes around the antenna. This allowed us to understand antenna performance in different snow conditions. Comparisons between simulated results and field data show antenna variations that are consistent in shape but differ in magnitude. GOAL 2: Improve snow depth retrieval. We hoped GNSS derived depth could eliminate extra snow depth hardware, but derived solutions were noisy and overestimated depth thus adding errors to SWE measurements. So, we shifted efforts in favor of refining SWE and LWC retrievals first and will include an ultrasonic snow depth sensor as part of the system. GOAL 3: Build a prototype system. NWB designed and produced an embedded system capable of collecting, and processing GNSS data into snow parameters. Work focused on a low power, modular hardware system to fit different customer needs. A microcontroller collects receiver data and powers up a main processor when its memory is full. 2020 hardware changes included new depth sensor circuity, a cell modem, and other minor changes. 2021 added a multifrequency receiver to achieve customer desired SWE accuracy. The multifrequency systems give less noisy results and penetrate deeper snowpacks. More updates were designed but not manufactured that protect the power supply from incorrect voltages, reduced noise, add supply chain flexibility, and more. The data collection and algorithm code were ported to C++ for the embedded system and successfully ran on the prototype hardware. The application has many features. To name a couple it detects above/below snow antennas, detects an USB connection, eliminates corrupt data, auto-recovers from crashes or errors, works with single and multi-frequency receivers, and monitors many functional parameters. To track the evolving prototypes inventory, quality control and testing procedures were implemented. All these efforts make the prototype a commercially ready product that is fully functional, flexible, user friendly, hardened against user error, easy to setup as a standalone sensor or interface with existing equipment, ruggedized for tough field conditions, and ultimately a low maintenance sensor. GOAL 4: Validate the SWE, LWC, and depth measurements from the GNSS snow sensor. Every winter data collection efforts were made, and different hardware setups were tested at cooperator locations with adjacent snow pillows or scales. The first winter hardware from the Phase I project operated at two locations. We also tested an experimental multifrequency GNSS setup at NWB's test facility, but the data was unusable. For 2019-20 new prototype hardware was deployed at eight additional sites. Other receiver and LWC retrievals were tested at NWB's facility. Funding and software issues prevented full data sets that winter. The next winter saw ten updated prototype installs at nine sites. Finally, the 2021-22 season started testing multifrequency systems at three sites each with several setups. Manual snow sampling equipment was purchased and maintained for winter field trips. We established manual sampling procedures, and trained staff on them. Snow tubes provided comparable SWE data, and a Finnish Snow Fork provided comparable LWC data. A total of 58 manual samples were made at field test sites during the project. Every site was sampled once a winter, most twice, and four were sampled 6 plus times. The data informed how much spacial variability was at each site, and informed differences observed between the GNSS sensor and adjacent snow sensors. This analysis primarily focused on snowpack model and algorithm development therefore a final accuracy of the new snow system has yet to be developed. The large variety of snowpack conditions measured at locations across the western U.S. has helped us create an algorithm that is generalized for all snow and location conditions. GOAL 5: Report findings. See the "Target Audience" section for more details. Many photos and other documentation were made so that this research will be valuable to the larger snow sensing community We fielded several inquiries from interested researchers and shared limited preliminary data sets with them. We also wrote final technical reports for the USDA SBIR program.

Publications

• Type: Conference Papers and Presentations Status: Accepted Year Published: 2019 Citation: P.W. Nugent et al., "Witchcraft, Wizardry, and Water: The intersection of Physics, Electrical Engineering, and Snow Monitoring," Eastern Snow Conference Proceedings, 2019.

Progress 09/01/21 to 12/31/21

Outputs
Target Audience:Surprisingly, many contacts were made during this short reporting period between September and December. Academic and non-profit researchers in need of snowpack data reached out from Alaska, Montana, and Canada's Northern Territory. Further state and federal agencies with operational data collection networks for the public good also contacted us in the fall looking for last minute wintertime data collection solutions. NWB Sensors shared information with these entities from our latest processing runs and will follow up with these groups in June. The timeframe to build and deploy snowpack sensing setups was unfortunately too short for the current winter season. Changes/Problems:NWB Sensors had hoped to manufacture and test new prcessing PCB's before the end of the project. Work was done to try and accomplish this, however so many components were still out of stock, that the work load was too much to acheive success. We began replaacing many components and documenting software changes that would be need for the new circuitry, but we still needed to find replacements for the power input, and efforts were focused on finalizing the multi-frequency hardware since existing processing boards functioned well enough to complete another winter of data collection. What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest? Nothing Reported What do you plan to do during the next reporting period to accomplish the goals? Nothing Reported

Impacts
What was accomplished under these goals? This project is working towards an economical fully automated electronic means to measure snowpack. The technology in development utilizes already existing signals and receivers for Global Navigation Satellite Systems (GNSS) like the U.S. operated GPS. Building on GNSS reduces development and manufacturing cost since GNSS antennas and receiver systems are widely used. Existing snowpack measurement methods have severe limitations. Manual measurements have high labor costs and expose personnel to exposure and avalanche dangers. Automated snow pillow measurements require expensive custom parts, also require much labor to install and maintain, have inherent measurement errors, and must be filled with antifreeze. Newer automated snow scales tend to have small measurements areas that lend to large measurement errors and give erroneous reading or are damaged when melting snow re-freezes. Therefore, it's important to find alternative means to improve availability and accuracy of snowpack measurements as ~36.5 million US citizens depend on snowpack meltwater to live. During this short reporting period NWB Sensors, Inc. moved its hardware and software development to a multifrequency GNSS platform. The multifrequency data, while more costly to collect than single frequency data, shows large improvements in snowpack data quality. These multifrequency hardware and software updates where tested and then deployed in the field for wintertime real-world data collection. Goal 1: During the previous reporting period experimental multi-frequency data processing had been added to the algorithm. This reporting period improvements were made to make the algorithm more operationally sound. Once again, the architecture was reworked to gain even more speed improvements. These speed improvements were dictated because the multi-frequency data about doubles the amount of data processing, and the improvements focused on multithreading efficiency. The result is a multi-threaded multifrequency application that can run efficiently on low power embedded hardware in the field, with quick processing times that save precious power. There were also continuing data filtering improvements made to the algorithm. Fringe or edge cases keep being discovered that make the final processed snowpack data product deviate from actual conditions. As we get closer to a final product, we can dig into the details of these edge cases and improve the algorithm to handle them, thus reducing error in the data product. Finally, the antenna characterization work closed out. More simulations of the antenna in snow were run, and we confirmed the antenna ground plane designs were not causing issues with data collection. Goal 2: Nothing new to report on this goal. We continue to collect data with a sonic snow depth sensor. Goal 3: A printed circuit board (PCB) was created for the new multifrequency GNSS receivers. Previous experimental work used evaluation boards for the receivers, but to move towards a final product we needed to create our own circuitry and user-friendly packaging. Components were selected and tested on a breadboard. After successful testing the designed circuits were laid out in an electronics CAD program for manufacturing. Portions of the design were run through an electronics modeling program called SPICE to verify operation. Improvements were made over previous receiver boards that will reduce production costs. We decreased potential GNSS signal power loss to help improve measurements in very deep snowpack. This resulted in a fully verified, functional, and hardened receiver PCB to deploy in the field at several sites for multi-frequency GNSS data collection. It was hoped that the receiver data processing PCB design changes could be manufactured as well for field deployment. The problem was many components were not available for manufacturing, so the design needed modified to use alternative components that were in stock. Work was done to identify many replacement components that were near drop-in replacements. Ultimately, we ran out of time to fully complete this task and the existing processing PCBs were sufficient for another winter of data collection. However, the design now has many alternative components identified that make it easier to overcome future electronic component shortages. Lastly, the embedded software continued to improve creating a more reliable product for future customers. Many updates were made to improve system auto-recovery from crashes or errors to ensure continued uninterrupted autonomous snowpack data collection. Bugs found in the code that communicates with the cell modem and code calculating snow depth were fixed. The database generator for both single and multi-frequency GNSS receivers was integrated adding flexibility in the product line. Goal 4: To verify the new multi-frequency GNSS receiver data the new hardware from goal 3 was deployed alongside existing single-frequency hardware at three local Montana sites with co-located snow pillows and/or snow scales. The three stations comprised a deep snowpack location, a moderately deep snowpack, and a valley low snowpack/intermittent snowpack site to test under different conditions. Two multi-frequency antennas were picked for testing and verification also. The first was in the same product line as our existing single-frequency antenna, and therefore should provide a similar response. The second was more than half the price less expensive and is being tested next to the other antennas to determine if we can reduce cost inputs to the product. So, at two of our test sites we have a total of three prototypes deployed. Generally, multi-frequency GNSS antennas cost 3-4 times more than single frequency antennas, and this was a primary deterrent from developing multi-frequency systems. Early season all systems provided reasonable results; however, the multi-frequency systems gave smoother less noisy results. Based upon these very early results it appears that multi-frequency systems will be necessary to achieve customer desired data accuracy, and product costs can be reduced by integrating the low-cost antennas. Unfortunately, final results will not be known before June 2022. Another notable achievement was that the liquid water content in the snowpack was reported live for the first time this winter season. Last reporting period only reported the frozen snow water content. Goal 5: NWB Sensors continued to document hardware and software improvement for reporting purposes. We fielded several inquiries from interested researchers and shared limited preliminary data sets with them. We also began writing the final technical report for the USDA SBIR program.

Publications

Progress 09/01/20 to 08/31/21

Outputs
Target Audience:The target audience for this work could be anyone who utilizes or depends on snowpack water runoff. This includes hydropower plants, boaters, fishermen, wildlife biologists, agricultural irrigators, municipal water supplies, and other industries. During the previous year NWB Sensors has primarily worked with people who collect the data and publish it for use by the groups just listed. Most organizations that collect snowpack data are federal and state governments. NWB Sensors continues to engage with USDA Natural Resource Conservation Service offices to collect experimental data and keep them engaged in our development process. We worked directly with staff in Montana, Oregon, Idaho, and Utah to coordinate data collection and validation efforts across a wide variety of geographical locations. We also shared preliminary data results with the NRCS Snow Water Equivalency working group and the Utah state Snow Survey Office. NWB remains engage with the Montana Climate Office also. We operated experimental equipment with MCO on a local Montana ranch to help determine water availability on the ranch and teleconferenced with staff on our development work. Further, NWB staff attended portions of the virtual 2021 Western Snow Conference to engage other attendees about our development, attempted to find collaboration opportunities with Montana State University staff, and continued assistance to a Montana Technological University staff member in snow data collection efforts. Changes/Problems:Previously we reported that we would not pursue a multifrequency GNSS receiver systems. However as outlined in this report we have been forced to reconsider that position, and we are now testing and designing a multifrequency system. This change was driven by new evidence that single frequency systems can go blind in very deep snowpacks. What opportunities for training and professional development has the project provided?NWB was able to hire an aspiring snow scientist, and avid back country recreationalist to help with field work on the project. The individual was given on-the-job training on how to collect manual snow samples, and snow pit data. How have the results been disseminated to communities of interest?As reported in the target audience section, new updated data reports were shared with the NRCS Snow Water Equivalency working group and the Utah state Snow Survey Office. NWB also teleconferenced with staff from the Montana Climate Office on our development work. What do you plan to do during the next reporting period to accomplish the goals?At of the end of this reporting period the project is approaching its overall end date. We planned to use the limited time remaining to develop our own multi-frequency GNSS receiver boards, deploy multi-frequency systems in the field for evaluation, and continue software development to better utilize multi-frequency data. The multifrequency systems should enable better performance in deeper snowpack, but more evaluation needs done. We will also continue to improve on the model for antenna gain and phase center variations due to snow. While we made progress on this there is still room for improvement. Lastly, we hoped to complete manufacturing of the new processing board design for testing as well.

Impacts
What was accomplished under these goals? Snowpack is a huge reservoir of water that nearly 2 billion people in communities around the globe rely on for drinking water, crop irrigation, generation of electricity, and other industrial processes. Snowpack producing less water than expected lead to water shortages and economic loss for these communities. Also, economic loss may occur from flooding events caused by excessive snow melt. Therefore, it is important to measure snow water equivalent (SWE) content for stream flow run off predictions and minimize losses due to varying snowpack. As the global climate changes these measurements and predictions continue to grow in value. Through this project NWB Sensors, Inc. developed technology to measure SWE and snow liquid water content (LWC) utilizing the Global Navigation Satellite System (GNSS) signals. Snowpack creates changes to GNSS signals that relate to the amount of water in the pack. This new GNSS sensor is less costly to make, and more environmentally friendly than existing measurement technologies. Snowpack is a complex ever evolving structure of ice, air, and water that makes it difficult to measure by electronic means. NWB's research shows information can be derived from GNSS data that relates to SWE and LWC in snow, but there are many challenges in creating an accurate model that relates the two in a wide variety snow conditions. A large effort to collect validation data using established snowpack measurement techniques gave valuable insight into the model and its tuning. This more precise understanding of snowpack electromagnetic properties will enable future snowpack sensing development. Goal 1: Mature the LWC and SWE derivation algorithm For this reporting period we moved all our algorithm and software development to the same C++ code base that runs on the embedded sensor platform that will be the final product. This sped up development along with a dedicated server used for algorithm development. Batch processing was done to process data collected from 9 plus stations over the last 5 years. By looking at all stations and all years we avoided the pitfall of tuning the algorithm for a particular snowpack type or location. Further, this allowed us to find software bugs more quickly, and forced us to improve algorithm speeds. Improved speed creates a better product by saving processing power at remote snow sites. Many bug fixes and changes to the data processing have culminated in better SWE & LWC data products with less noise and error. Major improvements were to the location solver, and the cycle slip detection approach was changed for robustness. Multipath signals at forested snow measurement sites make the signal strength, or signal to noise ratio (SNR), readings troublesome as reflections can increase and decrease SNR. Many data filtering methods were tested to improve noisy GNSS data, and we found including more satellite constellations helped improve SNR data. A test was done to empirically measure GNSS antenna performance in snow with different LWC. These measurements confirmed some antenna behavior modeled previously, but also showed that some modeled effects were not as strong. This data improved our algorithm for shifts in antenna gain and phase center due to snow. The new antenna information and trouble measuring SNR revived the effort to use multifrequency (L1 & L2 signals) GNSS receivers. The software was changed to utilize multifrequency data. Early results indicate solutions using L1 & L2 data reduce error and improve the LWC measurement. Goal 2: Improve snow depth retrieval Snow depth data is needed to constrain density and refractive index used by the algorithm. We hoped depth could be derived from GNSS data eliminating extra snow depth measurement hardware, but derived solutions were noisy and overestimated depth thus adding large errors to SWE measurements. After the algorithm was ported to C++ and many bug in the GNSS residual calculations were fixed this option was briefly revisited, but the results were the same. So, we continue to use a sonic sensor to measure snow depth. For testing the software was updated to optionally ingest snow depth data from other sources. Experimentation shows that the snow depth sensor should be above the below snow antenna, and considerations made for erroneous sensor data. Goal 3: Build a prototype system In the fall of 2020, hardware additions were created for the prototype system including circuity to fix communication issues with the depth sensor, and a cell modem reporting station data. An existing web server and database interface was repurposed to display data reported via the cell modems. To track the evolving hardware, quality control and inventory procedures were implemented. Testing procedures were established for hardware and software changes to insure successful field operation. A bug appeared mid-winter that caused a process to hang due to the increasing database size. This prompted development of scripts to update the embedded application in the field. A design update to the snow processing and receiver PCBs was done after two winters of operation and learning. The updates decreased power supply noise, added communications options, protected the power supply from incorrect or reversed voltage connections, included the fall 2020 additions, and several other layout changes to increase reliability. The receiver PCB was resized to for assembly, and hardware that truncated long data messages was replaced. Staff built test circuits to test the updated designs. We keep improving our hardware and identified better connectors for data, USB, and sensor ports that are watertight. The algorithm was ported to C++ for the embedded system. Additionally, the embedded application now detects above/below snow antennas, detects an USB connection, has cleaner log file output, fully utilizes the on-board RAM, and more. Corrupted receiver data issues were eliminated using checksums. System health monitoring was added including boot counts, supply voltage, temperatures, database size, available memory, cell signal strength, and much more. This health data was made available to on-site technicians and transmitted via the cell modem enabling fast detection of system failures. These improvements to the prototype put us very near a commercially ready product that has been field tested and debugged with the end user in mind. Making the initial product offering more user friendly, hardened against user error, easy to setup as a standalone sensor or interface with existing equipment, and ultimately a low maintenance sensor. Goal 4: Validate the SWE, LWC, and depth measurements from the GNSS snow sensor We deployed 10 prototypes at 9 cooperator sites. A temporary tower with solar power was installed at each location to host the prototype. 7 locations had snow pillows and 3 had snow scales. Detailed manual snow sampling procedures were developed to measure SWE and LWC. Sampling equipment was maintained and calibrated. Staff was trained on the equipment and procedure. Manual and GNSS data was saved and backed up after every trip for post processing. After the winter we removed all equipment from cooperator sites. A total of 24 manual samples were made. Every site was sampled once, most twice, and two were sampled 6 times. We created automated scripts to ingest snow pillow and scale data from cooperators. This data informed how much snowpack spacial variability was at each site, and informed differences observed between the GNSS sensor and adjacent automated snow sensors. Slight location variations affected when the sensors melted out, and how much snow they measured. This data helped us improve the snowpack model used by the algorithm and establishes early accuracy data for the new product. Goal 5: Report findings Many photos and other documentation were made so that this research will be valuable to the larger snow sensing community. See section on how results have been disseminated.

Publications

Progress 09/01/19 to 08/31/20

Outputs
Target Audience:NWB Sensors has maitained communication about project development with two groups that could ultimately benefit from this project: USDA Natural Resource Conservation Service Snow Survey Offices University of Nevada Reno College of Agriculture Changes/Problems:Delays leading to a one-year extension of the project Due to funding delays, NWB was unable to collect the snowpack data needed to complete the project last winter. Therefore, the project is being extended for one full year. The COVID-19 pandemic and resulting economic shutdowns hindered our ability to access our sites in the spring. This delays site access had delayed development, and along with impacts to supply chains for materials and components have contributed to the need for a one-year extension to the project. Project change shifting away from multi-frequency excess-phase analysis Our methods currently utilize the GPS L1 signal but can use multiple GPS frequencies. However, current work has been inconclusive as to whether utilization of the GPS L2 signal is of benefit. Work has been done to develop the GLONASS phase and C/No processing. GLONASS is only slightly shifted in frequency from GPS L1, and GLONASS phase did not provide unique information. The GLONASS C/No has shown to provide unique information that may be of benefit. Therefore, for the sake of efficiency as the project moves forward, we plan to focus on the GLONASS C/No benefits, not on L1/L2 processing or GLONASS excess phase processing. What opportunities for training and professional development has the project provided?This project has provided on the job training for three interns from Montana State University. This training has included knowledge of electrical design, software development, project management, and Snowpack measurement. In addition to this on-the-job training, additional in the field snowpack measurement training has taken place for all NWB Sensors employees. This project has led to inform professional development for many of the employees at NWB Sensors. This professional development has included learning and implementing coding practices and implementing project management software. In preparation for the production of commercial projects, NWB Sensors has been utilizing workshops from local industry groups on product production, product exports, and Quality Management Systems. In addition to these workshops, we have been conducting bi-weekly training sessions on ISO-9001 quality system management. Financial support for this training has come from an SBIR matching grant from the State of Montana. How have the results been disseminated to communities of interest?This process is ongoing and will continue through the completion of our project. We continue to report progress to the USDA-NRCS Snow Water Equivalency working group. We have established new partnerships with the University of Nevada Reno and a California Company: Alpine Hydromet. What do you plan to do during the next reporting period to accomplish the goals?Due to the rapid development of the SnoProb in fall 2019, there were features left out of the system, and design issues to be corrected in the next revision. With the requested extended period of performance, we will build another improved version of the embedded system to address the necessary changes in the SnoProb. This board will then be re-deployed at our snow-measurement sites for the 2020/2021 winter. This data collection effort will provide the data needed for performance evaluation. Algorithm improvements will then be implemented using any/all available snow-covered data. These improvements will result in a commercial system that should be ready at the completion of this project.

Impacts
What was accomplished under these goals? Since NWB Sensors' last report, there has been a significant gap in development due to a large gap in funding. Nevertheless, the technology being developed is still crucial to society's water management issues today. While we compute "normal" snow water supply conditions and use it to compare current data to, over 100 years of snowpack monitoring history shows that every year is different. Last winter, NWB Sensors strived to catch this variability by deploying prototype snowpack sensors across western states. Locations were selected that would represent different snowpack water content, snow depth, density, and temperature. By studying how our technology functions in different environments/snowpack, NWB Sensors will create robust hardware and algorithms to produce accurate snow measurements. Antenna design and system performance have been revisited, and we have observed the limits of the direct Excess phase plus C/No algorithm in the deepest snowpack during the melt. In deep snowpack during the melt, the receiver loses the ability to track the phase but continues to provide C/No measurement. Therefore, we must rely on algorithm improvements to provide accurate snow data during these periods. Algorithm improvements have converged on a method of solving the complex non-linear space that leads to snow measurement from the GPS signals. New snow stations based on our low-power modular circuit board were deployed, and we learned a lot about the operation of these systems in the wild. We are currently revising these boards to address the issues we saw in the 2020 winter in preparation for a deployment in the coming 2021 winter. Below are summaries of the work accomplished on each goal: Goal #1. Mature the snow properties derivation algorithms used to retrieve snow LWC and SWE from GNSS observables to the level of commercial readiness The algorithm development has progressed during the past year toward the requirements of a fully functional end product. The incoming data are automatically processed to the degree that can determine snow parameters, except for depth. These parameters include station parameters such as a highly accurate position vector between the antennas. Refinement is still necessary to reliably achieve the target accuracy in retrieved snow parameters we have set as a goal in our commercial launch. Work focused on refining the snow model, finalizing the non-linear solver used to solve snow parameters, solving problems with integer ambiguity in the excess phase data, excess phase cycle-slip correction, and automatically filtering out data of poor quality. Algorithm optimization will be completed during the code porting to the embedded system and will be a significant focus of the work over the next period of this project. A parametric model of the antennas in use in our systems has been built. This model has allowed us to understand how our antenna performs in a variety of different snowpack conditions. Comparisons between simulated results and field data show antenna variations that are consistent in shape, but the field results and modeled data differ in magnitude. We continue to work to understand these differences, and these differences should lead to an improved snowpack model. Operation results have shown that we lose the ability to measure excess-phase during the melt in the deepest snowpack at sites. At these sites, the C/No drops below the level required for a phase lock to be maintained; however, C/No measurement continues. This signal drop is more significant than can be achieved with antenna redesign. We concluded that improved algorithms tracking changes in C/No differences are the most likely method of improving operation in deep snowpack during the snowmelt. Goal #2 Improve Snow Depth Retrieval: Four options have been identified for this problem. The option with the highest chance for success with the lowest cost will be tested first, and so on until a reliable snow depth is automatically determined. Snow depth is currently obtained from an ultra-sonic snow depth sensor installed at our test sites. Work has been done to retrieve depth from the GPS signals themselves; however, it has not produced sufficiently accurate results. While it may be possible to retrieve depth with our systems, we are shifting our efforts to other aspects of the project and will include an ultrasonic snow depth sensor as part of the initial system. Goal #3. Build a prototype system, including the implementation of an embedded algorithm with the most appropriate commercial hardware, optimized for low-power operation in remote snowpack monitoring networks. NWB designed and produced a new version of an embedded system capable of collecting data, processing, logging, and distributing measured snowpack data. This integrated system (named the SnoProB for Snow Sensor Processing Board) was quickly deployed before the first significant snowstorm in November 2019. Goal #4 Validate the SWE, LWC, and depth measurements from the GNSS snow sensor via deployment alongside snow pillows in multiple snow types and use intensive manual sampling for ground truth with snow tubes (SWE) and the Finnish snow fork (LWC). Stations were deployed in 2019/ 2020 winter with partial success, and in preparation for a future deployment, much of the hardware, including antennas, remain in place. A total of eight full stations were in operation during the 2019/2020 winter, with an additional two partial installations in Montana that did not get SnoProB's installed due to snowfall and programmatic issues that limited our ability to access the sites. We plan to continue operating most of these stations in the coming winter so data validation can be adequately completed. An additional partial station was installed in Dr. Nugent's backyard to validate operation in shallow snowpack and to test snowfork operation. This station operated through multiple accumulation and melt cycles. Nearly continuous measurement of snow liquid content was possible at the Nugent site using the Finnish Snow Fork. This device only logs 25 hours of data and requires daily data downloads and interaction. Goal #5 Report findings to cooperators in the Natural Resource Conservation Service (NRCS), United States Department of Agriculture (USDA), National Oceanic and Atmospheric Administration (NOAA), and the National Weather Service (NWS), and other key agencies. See the section on how results have been disseminated.

Publications

Progress 09/01/18 to 08/31/19

Outputs
Target Audience:During the Phase II SBIR effort NWB Sensors is continuing to work closely with the USDA Natural Resource Conservation Service (NRCS), Snow Survey Program to establish testing locations for our sensors. These sensors will be co-located with NRCS snowpack measurement stations. We hired two student interns to aid in development on this project. We were able to collaborate during 2019 with the Montana Climate office and co-locate three of our sensors at their Montana Mesonet stations. One of these sensors remains at a station that is a collaboration between the Montana Climate office, a private ranch, a meteorological sensors company, and NWB Sensors. Utilizing funding provided by the State of Montana to aid in commercialization and development of our snowpack sensing product we were able to report on the results of this program. We sent representatives to the 87th Annual Western Snow Conference in Reno Nevada on April 15-18, 2019, and to the 76th Annual Eastern Snow Conference Meeting in Fairlee, Vermont on June 4-6, 2019. In addition to these conferences we have been able to reach out to a variety of both public and private entities in the hydrological communities. Many of which have expressed interest in placing orders of our systems in spring 2020. We are currently working to convert this interest to pre-sales. We are continuing to work with international marketing experts within the State of Montana's Department of Commerce and the United States Department of Commerce to develop plans for reaching international markets for our sensors, including Canada, Japan, New Zealand, and others. During winter 2019 personal communications have taken place with individuals in the back-country ski community in Bozeman, Montana. This active community is a major consumer of snowpack data, in particular due to the avalanche risk experience by people in the community. The idea of increasing snowpack sensing has been welcomed by this community and some individuals have offered to volunteer to aid in wintertime manual sampling at our research sites. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?Training activities This project has provided on the job training for three interns from Montana State University. This training has included knowledge of electrical design, software development, project management, and Snowpack measurement. In addition to this on-the-job training, additional in the field snowpack measurement training has taken place for all NWB Sensors employees to better prepare them for the 2019-2020 winter deployments. Professional Development This project has led to inform professional development for many of the employees at NWB Sensors. This professional development has included learning and implementing coding practices and implementing project management software. In preparation for production of commercial projects NWB Sensors has been utilizing workshops from local industry groups on product production, product exports and Quality Management Systems. We are currently discussing with the Montana Photonics Industry Alliance about becoming one of the first companies to participate in their Quality Management Systems mentorship program. How have the results been disseminated to communities of interest?NWB Sensors' attended the 87th Western Snow Conference in Reno, NV and shared limited information with members who were interested about our snow sensor research. Preliminary results were presented in a poster at the 76th Annual Eastern Snow Conference Meeting in Fairlee, Vermont on June 4-6, 2019. Results have also been shared with the Montana NRCS Snow Survey Office, and a product brochure has been added to our website at: https://www.nwbsensors.com/snowpack-monitoring. What do you plan to do during the next reporting period to accomplish the goals?Early in this next period we will finish the hardware development and deploy the hardware in our in-field experiments during the 2019-2020 winter. Data from these deployments will be processed using the finalized algorithms, and final measurements of system accuracy will be determined and communicated to interested parties. During the deployments and intensive ground truthing using manual field measurements will take place to provide truth data independent of both our sensors and the co-located snow pillows and/or snow scales. Also, during this 2019-2020 winter algorithm, data analysis, and system hardware will be refined using experiences from the field deployments, and feedback from collaborating partners. This work will lead to a final hardware revision that will be completed in spring and summer 2020 ready for commercials sales and commercial deployments by fall 2020.

Impacts
What was accomplished under these goals? NWB has been busy improving both hardware and software for its new Snowpack Monitoring System. 2019 has shown us once again why we need more snowpack measurements. Heavy snows on the plains and in the mountains created large damaging floods along the Missouri river again this year. The US Army Core of Engineers reports a single levy repair will cost more than a billion dollars, and they are embarking upon a study to manage the Missouri river basin more intelligently. This new management plan will need snowpack data. NWB has also taken phone calls from people at the University of Nevada, USGS, and Montana Climate Office who want/need to collect more snowpack data. With these potential customers in mind design changes to our Snowpack Monitoring Sensor have been made to ensure it meets a variety of customer's needs including operational hydrology networks, researchers, and avalanche forecasters. Antenna design has been revisited to improve their ruggedness and robust operation in many different environments. Algorithm speed improvements have been made to process the data faster with less power. Algorithm accuracy improvements were made with antenna modeling. New low-power modular circuit board assemblies were designed and are being tested. It is an exciting time here at NWB Sensors as we prepare for the rapidly approaching winter season. Below are summaries of the work accomplished on each goal: GOAL #1: The snow model has been significantly improved by adding in antenna characteristics and methods to deal with the surrounding environment. The model now incorporates antenna effects such as phase center shifts. The antenna parameters are still being studied using antenna simulations. These simulations also give antenna gain pattern changes that will be incorporated into the model. Our antenna modeling has allowed us to understand how the gain pattern and tuning of the antenna changes as the dielectric around the antenna changes. This led to developing and deploying experimentally modified antennas last winter. These modified antennas have shown promise in removing problems where the antenna tunes away from the GNSS frequencies. Statistical analysis of these results is still underway and will be incorporated into a new antenna design for deployment in the 2019-20 winter. With improved antenna mounting designs the problems associated with ground reflections and snow-soil propagating EM waves is expected to be minimal so focus has not been put on adding these effects into the model. However, soil reflections are incorporated if needed, especially when determining snow thickness based off reflected signals. The snow model itself will still need to incorporate surface roughness and physical snow effects such as drifting, but these problems are minor compared to the antenna effects. There is still an open question as to how to map the derived values of index to the desired parameters of SWE and LWC. The largest effects on the index to SWE model seems to be related to the physical hardware used. Receiver hardware differences manifest in the quality of the data and therefore the derived indices. Cycle slip correction that must occur when the Phase Lock Loop loses signal lock for a satellite have been implemented in our preprocessed data. Corrections for true cycle slips between two adjacent datapoints are fixed by using a triple difference between data points to identify jumps then correcting all the data after the jump by adding a fixed value of cycles. A second regime of interest is those losses of signal over several seconds which we have also implemented corrections for. Modeling the multipath environment has proven difficult. However, a receiver that NWB is using has built-in multipath detection and suppression that seems to work well in the limited testing completed to date. Further tests are necessary to compare receivers in identical environments that are subject to strong multipath effects. It has been found that multipath is a minor effect on the phase and is primarily concentrated in C/No problems. Initial sky mapping has been done to look at the spatial distribution of C/No during no snow periods. With greater understanding of antenna effects these sky maps may be able to remove persistent multipath effects. During the 2018-19 winter a multi-frequency GPS L1/L2 antenna, GLONASS G1/G2, Galileo, and Bedu. During the winter damage occurred to the under-snow antenna that has limited the amount of data collected. These systems are planned to be re-deployed during the 2019-20 winter with antennas built by the same manufacturer as the antennas used in our single frequency systems. During snowpack melt out/ablation cycles NWB has noticed shorter periods of GNSS data, about 1 hour, worked best in the SWE and LWC retrieval algorithm primarily due to the rapid changes in the snowpack during the day. This is opposed to previous work where long durations on the order of 12 hours was used. Further work will be done to determine if these shorter periods can deliver reasonable results long term. Data can also be decimated from the 1 second per data point as well so long as each pass still contains a few hundred datapoints, meaning that for a 1-hour period decimating the data by a factor of 10 is possible and has been tested successfully. GOAL #2: Experiments during the 2018-19 winter implemented an acoustic snow-depth sensor alongside the GNSS snow sensors. This is planned to continue in the 2019-20 winter while we work to development snow depth retrievals. Work on this goal has been delayed in favor of refining SWE and LWC retrievals first. GOAL #3: Building a new prototype system optimized for low-power operation has been a major part of the workload for the project. Much of the work on this goal has focused on the implementation of a low-power hardware system. Low power PCBA's have been designed, built, and are being tested for two different receivers we are considering in the final product. Much work has been done designing low power communication busses, and highly efficient power regulators. The system has been designed to be very modular to fit a variety of customer needs. NWB is currently working to finalize the design for a low power processing board that will create the final data products from the receiver's raw observables. While the hardware is not yet in house, the software is being developed and tested with development boards/kits from the manufacturers of the processors we are using. A software design plan has been written, and initial embedded system applications are being programmed. The software will monitor temperature and power conditions and respond to outside requests from other devices. Code to collect data from the receivers and database it (packet parsing, and data preprocessing) has been ported entirely to C++ and is being tested. The data management and processing code will run on an embedded Linux processor. A microcontroller will collect the data from the receivers and will power up the main processor once enough data has been collected. GOAL #4: All manual snow sampling equipment has been purchased, tested, and is ready for use during the upcoming winter season. Initial manual sampling procedures were established the last 2018-19 winter, and staff members received training on proper sampling techniques. Three sites had intensive manual snow sampling to provide truth measurement of SWE. Initial tests were conducted to provide comparable LWC measurements in preparation for intense LWC measurements this coming winter. Preliminary analysis of data has been compared to co-located snow sensors. This analysis has primarily focused on algorithm development and algorithm improvement therefore a final accuracy of the new snow system has yet to be developed. GOAL #5: See the section on how results have been disseminated.

Publications

• Type: Conference Papers and Presentations Status: Accepted Year Published: 2019 Citation: P.W. Nugent et al., "Witchcraft, Wizardry, and Water: The intersection of Physics, Electrical Engineering, and Snow Monitoring," Eastern Snow Conference Proceedings, 2019.