Source: FLAT EARTH, INC. submitted to NRP
SWEDAR – AN AUTOMATED LOW-POWER SNOW WATER EQUIVALENT SENSOR
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
COMPLETE
Funding Source
SMALL BUSINESS GRANT
Reporting Frequency
Annual
Accession No.
1024271
Grant No.
2020-40000-33002
Cumulative Award Amt.
$649,710.00
Proposal No.
2020-06705
Multistate No.
(N/A)
Project Start Date
Sep 1, 2020
Project End Date
Aug 31, 2022
Grant Year
2020
Program Code
[8.4]- Air, Water and Soils
Recipient Organization
FLAT EARTH, INC.
985 TECHNOLOGY BLVD STE 102
BOZEMAN,MT 59718
Performing Department
(N/A)
Non Technical Summary
The goal of our Phase I USDA-NIFA research contract was to develop a prototype Snow Water Equivalent (SWE) sensor and companion techniques that provide high-resolution snow property estimates, to improve forecasting of water resources for agriculture and public consumption, mountain weather, avalanches, floods, and hydropower. Our measurement techniques involved the use of Ultra Wide Band (UWB) radar technology. While radar has been used by snow scientists for research for decades, due to the complexity of interpreting the radar signal, the cost of the instrumentation, and the power required, UWB technology is not currently used operationally. The largest barrier is the lack of autonomous algorithms for interpreting the radar signal; radar images are normally interpreted by scientists, requiring manual intervention.The goal of this Phase II research contract will be to complete the development of a SWEdar sensor, which processes the radar data on-board, in real-time, without any manual intervention, and transmits SWE estimates and other weather observables hourly to water and snow stakeholders.To achieve this Phase II goal our proposed work includes the following:Improving algorithms to solve the wet snow problem that we observed in late season of our Phase I results. This pertains to the transition between wet and dry snow. We believe that accurate detection of the snow surface and internal layers, along with prior knowledge of snow conditions from the dry snow measurements, will allow us to transmit accurate SWE estimates in all conditions. The new prototype algorithm uses data from multiple radar systems operating at different frequencies, and a new lidar sensor to accurately detect the snow surface after low density snowfall. We have determined optimal frequencies, but work remains to optimize and test the algorithm that combines data from all sensors, for operational use in a real-time configuration.Transitioning the hardware design from the prototype phase to a commercially available UWB based radar SWE sensor. This effort will involve combining X1 and X4 radar technology, and a low-cost lidar, and controlling all sensors from a single low power microprocessor. A satellite modem will be integrated for direct to sensor software upgrades and data retrieval. Temperature, pressure and humidity sensors will be added to round out the quality of data collected and expand our market base.Implementing our unique Remote Sensing as a Service (RSaaS) business model as part of the commercialization and marketing plan for the SWE sensor.We anticipate that by the end of the Phase II effort we will have resolved the wet snow problem we encountered in Phase I, by mixing radar frequencies and incorporating lidar depth measurements in the late season. We also anticipate completing the development of the next generation SWE sensor hardware. A unique remote sensing platform is created when you combine this next generation sensing technology with our RsaaS business model. By the end of this Phase II effort a new concept on remotely sensed SWE will be commercially available, marketed, and sold in several regions worldwide. ?
Animal Health Component
50%
Research Effort Categories
Basic
0%
Applied
50%
Developmental
50%
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
11202102020100%
Knowledge Area
112 - Watershed Protection and Management;

Subject Of Investigation
0210 - Water resources;

Field Of Science
2020 - Engineering;
Goals / Objectives
The goal of this Phase II research contract will be to complete the development of a SWEdar sensor, which processes the radar data on-board, in real-time, without any manual intervention, and transmits SWE estimates and other weather observables hourly to water and snow stakeholders.?To achieve this Phase II goal our proposed work includes the following:Improving algorithms to solve the wet snow problem that we observed in late season of our Phase I results. This pertains to the transition between wet and dry snow. We believe that accurate detection of the snow surface and internal layers, along with prior knowledge of snow conditions from the dry snow measurements, will allow us to transmit accurate SWE estimates in all conditions. The new prototype algorithm uses data from multiple radar systems operating at different frequencies, and a new lidar sensor to accurately detect the snow surface after low density snowfall. We have determined optimal frequencies, but work remains to optimize and test the algorithm that combines data from all sensors, for operational use in a real-time configuration.Transitioning the hardware design from the prototype phase to a commercially available UWB based radar SWE sensor. This effort will involve combining X1 and X4 radar technology, and a low-cost lidar, and controlling all sensors from a single low power microprocessor. A satellite modem will be integrated for direct to sensor software upgrades and data retrieval. Temperature, pressure and humidity sensors will be added to round out the quality of data collected and expand our market base.Implementing our unique Remote Sensing as a Service (RSaaS) business model as part of the commercialization and marketing plan for the SWE sensor.
Project Methods
The work plan includes nine engineering milestones to ensure the completion of our Phase II proposal herein, from a technical standpoint. They will not necessarily be completed in this order, or in a sequential fashion, but because of the expertise required for each milestone they can be worked toward in a staggered and parallel effort over the course of the 2 years, for optimal productivity.SWEdar Rev 3.5 Fabout for 2020-21 Winter DeploymentThe Generation 3, revision 3.5, SWEdar Fabout includes implementing any critical hardware errata found on the original Gen 3 version for sustained and reliable functionality in the field. After a thorough review of hardware functionality and performance from our field deployments in 2019-20 we will make necessary, yet low risk schematic and PCB changes. The electronics and mechanical enclosures will fabricated, tested, and validated in our lab. They will be assembled in house. We will have approximately 10 prototype units made for 6 field deployments and 4 lab units, which will be used for cold-lab testing, software debugging, and continued improvement, e.g. algorithm features. SWEdar Gen 4 Design and BuildThe SWEdar Gen 4 Design and Build includes all of the latter facets of the electronic design process and of printed circuit board assembly (PCBA) fabrication. At the end of this task the hardware design will freeze to guarantee the unit correlates with the project goals and functions as intended, without early failure. This is considered a final, production-level build that incorporates all errata and upgrades from past 3.0 and 3.5 revisions. Moreover, new additions are planned for enhanced snow surface detection. Electronic Stormboard (ESB) Design and BuildThe ESB Design and Build includes all of the early facets of the electronic design process, from the research and requirements flesh out to the PCBA and mechanical enclosure prototyping, to guarantee the unit correlates with the prototype requirements document (PRD). The unit needs to function adequately to prove the concept, even though it will not yet be at production-level function, performance, and reliability. SWEdar Beta Firmware Design and DevelopmentThe SWEdar Beta Firmware Design and Development task encapsulates the embedded software suite needed to run the SWEdar unit in all required states and given a planned input, e.g. radar frame, Wi-Fi request, temperature reading, etc. The firmware initiates and manages communication, upgrades, data collection, data processing, and ultimately the proper output on the SWEdar hardware platform. The Beta release stage of this firmware will have a complete feature set and will be suitable for release to outside customers, yet will contain non-catastrophic bugs and need ongoing testing. The firmware not only includes low-level device driver compilation and algorithm porting, yet it all has to be wrapped in a custom, upgradable Linux operating system (OS). Furthermore, every stage has to be thoroughly unit tested. SESB Alpha Firmware Design and DevelopmentThe ESB Alpha Firmware Design and Development task encapsulates the prototype-level firmware needed to prove the functionality and utility of the ESB electronics, including requirements analysis, design, testing, and recursion of code. This firmware will be under-tested for widespread or public deployment, will have bugs, yet it will be reliable enough to prove the concept and usefulness of the ESB. The electronics for the ESB should be pared down compared to the SWEdar, so a minimal set of device drivers will be needed; however, this will be developed on a microprocessor (MCU) platform, without the luxuries of Linux, yet streamlined, low power, and low cost. The BLE implementation and the Algorithm development are two paramount duties, as the intelligence to extract snow depth and SWE from the direct pulse will need to be devised. This information is then relayed via a BLE pair to the adjacent SWEdar unit. SWEdar Dashboard Design and DevelopmentThe SWEdar Dashboard Design and Development is a web-based application and database to store and view, public or proprietary, SWE data transferred to the cloud from deployed SWEdar units. The user interface (UI) will be developed using a free, powerful cross-platform framework called ASP.NET. Moreover, it is linked to a cloud solutions backend, consolidated and manged by Amazon Web Services (AWS). It will utilize features such as Amazon Simple Storage Service (S3), Amazon Simple Queue Service (SQS), AWS Lambda, Amazon Relational Database Service (RDS), and AWS Elastic Beanstalk, that all work together to receive, parse, store, recall, display, and secure customer SWE data. The ability to make quick comparisons to the nearest SNOTEL SWE estimates will also be a valued feature. Software Field App Design and DevelopmentThe Software App Design and Development is the local application package that resides on a smart-tablet/laptop and communicates with the hardware via a wireless (or wired) connection to exhibit the data in a human-readable and intuitive way. The application will have features required for data logging, streaming, and analysis, as well as built-in macros for specific site visits, e.g. installation, maintenance, and tuning. Three platforms will be targeted including Windows, Android, and iOS; however, a Windows-based laptop app will be the first priority. Algorithm Improvement and TuningThe Algorithm Improvement and Tuning task encompasses all algorithm directions required by the data, research, science, and engineering behind enhancing the SWE estimate. Specifically, all the conceived, and yet to be conceived, ways to better tease the ground reflection from the data in a more consistent manner when buried in dense, wet snow. The saturated snow with higher liquid water content (LWC) will be a specific focus. The high-level task break down includes field work to collect data to bring back to the lab for post-processing and two parallel algorithm approaches: (i) continuing with the traditional rule programming algorithm by adding more sensor data and observations to refine the SWE estimate and (ii) defining and implementing a machine learning (ML) framework to recognize liquid water in the snow pack and/or recognize the ground, despite wet snow. ?Preliminary Regulatory Approval The Preliminary Regulatory Approval will ultimately be accomplished by enlisting a third-party lab for testing. However, FEI will support these labs with requisite firmware and hardware prototypes to carry-out said testing. The test reports will be generated by the third-party labs and submitted to the appropriate regulatory body, e.g. FCC, for approval and ultimately certification. The goal will be to self-certify and have the labs perform the tests and prepare the results for submission. If the regulatory body chooses to audit, the documentation is in place to show the units are within specified levels. Forecasted testing procedures include the following (depending on desired regional markets):FCC UWB TX compliance testing support for North American / Canadian radar emissionsETSI UWB TX compliance testing support for European radar emissionsWi-Fi compliance testing support for Wi-Fi emissionsBLE compliance testing support for Bluetooth emissionsCE compliance testing support for European noise level and spurious emissionsEMC testing support for North American / Canadian noise level and spurious emissions

Progress 09/01/20 to 12/10/21

Outputs
Target Audience:Target Customers We can break our target customers down into three generalized user types; Researchers, Government, and Private Industry. Researchers are dominated by academic and government users and the sensing technology is used in support of snow science research. They are typically not too concerned about sensor pricing because they usually purchase small quantities. Researchers are an important group because they typically are first at vetting the technology, which can influence the Practitioner and Private Entity user types into purchasing new technology. Government customers are made up of some form of federal or state entity such as the Department of Transportation, Water Resource and Hydrology departments, Forest Service, Bureau of Reclamation, Army Core of Engineers, National Weather Service, and State Municipalities and Departments that rely on snow pack estimates as a critical part of their water supply forecasting operations. They operate and manage existing SWE networks and may maintain the SNOtel sites in a state or region. The sales opportunity with the SNOtel and similar systems managed by government entities is network densifying and replacing existing systems that fail. Private Industry would be comprised of engineering companies involved in ski resorts, hydrology, snow removal, agricultural, and agronomy consulting for farmers and ranchers, energy management, reclamation, and engineering efforts that require knowledge of SWE and Snowfall. Changes/Problems:Some design and development deviation from our proposed approach occurred in this effort. Early test results from the LiDAR sensor we incorporated into the SWE hardware proved to be superior to the radar solution we had proposed. After deliberation we concluded that LiDAR was the better sensing choice and revised the design to employ LiDAR instead of UWB Radar. We also discovered there was a significant market for this sensor as a stand-alone, low power, snow depth sensor (SNOdar). Development of this sensor is well ahead of schedule. Due to COVID related issues we were able to doubke the number of engineers on thei project and managend to complete the Phase II effort several months early. What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest?Yes, through our sales and markeitng efforts we have been in contact and secured sales from some of the industries leadsing inflencers including government organizations, hyrdo power entities, and ski resort industry. 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 main accomplishments of Phase II research was the successful completion of the development of awater contentprototype sensor (SWEdar) and the completetion of a novel snow depoth sensor (SNOdar). To accomplish this goal we implemented real-time on sensor processing of the radar data, and now transmit these SWE estimates and other weather observables hourly to water and snow stakeholders. We also resolved the late season wet snow problem we discovered in Phase I. This was accompolished with the development of a lower frequecny radar/antenna combination. We also the development ofa new electronic storm board capability as a function ofthe SNOdar sensor. Commercialization accomplishments include sales and marketing efforts worldwide. Sales and shipping product worldwide. Establishing reseller distribution across Scandinavia. CReated two business modles 1) Hardware purchase plus annual suport fee, 2) RSaaS model where the user is billed anually an dhardware is provided at now cost.

Publications


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

    Outputs
    Target Audience:The following snow depth and water content stakeholders were contacted during this reporting period: National Resources Conservation Services (Sold 6 SNOdar units) National Ski Areas Association (Will be attending the Jan '22 trade show with booth) Installed 5 SNOdars at a ski area in Montana. TeleNor a large Scandinavian telecommunications company. We have sold/shipped units to them and are now in reseller agreement discussions for distribution across all of Scandinavia. Hydropower entities. We have been in contact with and sold units to several hydropower entities in Idaho, Montana and Norway. Changes/Problems:Currently we have not experienced any schedule slippage, cost over-runs, or unexpected high costs. In February we committed over 725 hours of development towards this project. Some design and development deviation from our proposed approach occurred in this effort. Early test results from the LiDAR sensor we incorporated into the SWE hardware proved to be superior to the radar solution we had proposed. After deliberation we concluded that LiDAR was the better sensing choice and revised the design to employ LiDAR instead of UWB Radar. What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest?For us communities are our target markets.Initial commercialization and marketing efforts have begun. We enteredthe snow science and ski industryand hydropower markets this spring with the introduction of our stand alone Snow Depth Sensor (SDS) and Generation 3 of our SWE sensor. What do you plan to do during the next reporting period to accomplish the goals?We anticipate that by the end of this Phase II effort we will have resolved the wet snow problem we encountered in Phase I, and incorporate LiDAR ranging for late season depth measurements. We also anticipate completing the development of the next generation SWE sensor hardware. By the end of this Phase II effort a new concept in remotely sensed SWE and snow depth will be commercially available, marketed, and sold in several regions worldwide. We will continue with commercialization of both the snow depth sensor and water content sensor. Efforts will include trade shows, direct sales, establish one or two distributors and expand into Europe.

    Impacts
    What was accomplished under these goals? SWE Sensor Hardware By the time of this interim report we were able to make low risk schematic and PCB changes to the SWEdar hardware. The electronics and mechanical enclosures were fabricated, tested, and validated in our lab. We were able to build 10 prototype units, 6 for field deployments and 4 for use in our cold-lab testing (Fig 5). This hardware effort matched our proposed development schedule. We had hoped to deploy 6 SWE prototype sensors during the 2020-2021 winter season but were only able to deploy 3, Banner Summit, Bogus Basin, and Bridger Bowl.Complications related to COVID-19 prevented us from working with state and federal agencies to get 3 sensors deployed on state and federal lands. We anticipate we will accomplish this during the 2021 - 2022 winter season. Electronic Storm Board (ESB)/ Snow Depth Sensor (SDS) Prototype The ESB is a fresh approach to the classic manual storm board, where a human has to read the snow depth from a ruler, then weigh a precise volume of snow to calculate density. These measurements yield a SWE estimate of the most recent storm snow. We proposed to employ a high frequency radar as the sensing technology. Some design and development deviation from our proposed approach occurred in this effort. Early test results from the LiDAR sensor we incorporated into the SWE hardware proved to be superior to the radar solution we had proposed. After deliberation we concluded that LiDAR was the better sensing choice and revised the design to employ LiDAR instead of UWB Radar. We also discovered there was a significant market for this sensor as a stand-alone, low power, snow depth sensor (SDS). Development of this sensor is well ahead of schedule. We completed our first production run of 500 in Sept 2021.We beganselling SDS worldwide in early summer 2021 and began delivery in Fall 2021. Dashboard The SWEdar Dashboard (Fig. 4) is a web-based application and database to store and view, public or proprietary, SWE and weather data. A prototype of the dashboard has been developed and deployed at our 3 test sites. RSaaS Business Model As part of our Phase II commercialization effort we proposed implementing the "as a service" sales model that is widely used in many markets today. We developed an annual subscription option with included software upgrades, data and analysis options, and free hardware replacement in case of failure. Data is transferred to safe, accessible cloud storage via the sensor on-board sat modem. Progress on this Remote Sensing As As Service (RSAAS) business model has focused on the development of the user dashboard and back office functionality such as communications between our cloud service and a sensor, software updates over cellular network, and customer database implementation.

    Publications