Source: FLAT EARTH, INC. submitted to
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.
1015908
Grant No.
2018-33610-28275
Cumulative Award Amt.
$99,560.00
Proposal No.
2018-00634
Multistate No.
(N/A)
Project Start Date
Aug 1, 2018
Project End Date
Jun 30, 2019
Grant Year
2018
Program Code
[8.4]- Air, Water and Soils
Project Director
Roberts, D.
Recipient Organization
FLAT EARTH, INC.
985 TECHNOLOGY BLVD STE 102
BOZEMAN,MT 59718
Performing Department
(N/A)
Non Technical Summary
Intellectual Merit: This Small Business Innovation Research Phase I project will assist in producing the next-generation approach for monitoring critical water resources stored in the seasonal snowpack. Agriculture is a significant user of ground and surface water in the United States, accounting for approximately 80 percent of the Nation's consumptive water use and over 90 percent in many Western States. Seasonal snow provides 50-80% of the available water in the Western U.S., but is poorly resolved with a sparse network of sites that use technology that has not changed in over 40 years. Snow researchers have used microwave radar since 1979 and proven its great advantages for snow measurement, however commercially available radar systems are not suited to continuous monitoring of snow due to design and power requirements. Commercially available radar sensors used for snow are designed for short surveys, require an experienced user, and manual input. Post-processing is required, with limited information in near-realtime. Current system cost prohibit widespread use.Technical hurdles: The proposed R&D in this SBIR project will address what we feel is our final technical hurdle; resolving SWE in late season wet snow conditions. For the past several years, our research has focused on penetrating deep into the snow pack to detect the underlying ground and developing SWE related algorithms. These efforts resulted in reliable and accurate SWE measurements in the early to mid season dry snow conditions. A typical season can begin in October and end in late May to early June; we have produced accurate SWE values through the middle of March, or roughly 75% of the season. Through this past research we have concluded that an accurate measurement to the surface of the snow pack is critical in resolving the last inaccuracies related to late season wet snow conditions. This Phase I research will focus on accurate ranging to the snow surface by incorporating an additional higher frequency UWB low powered radar sensor. The new range data will be integrated into our existing SWE algorithms to improve the automated detection of the snow surface, and to allow analysis of frequency-dependent attenuation to allow SWE inversion in wet snow. Goals: Resolve the inability to accurately determine SWE in late season wet snow conditions. The result will be a sensing solution capable of estimating and monitoring the critical state variables of the snowpack (depth, SWE, density, liquid water content) autonomously in real-time. The goal is to be well positioned for a Phase II award where we can begin the prototyping and commercialization of our SWE sensor.Summary of plan: A new K/W-band radar will be integrated with the current snow radar design. Prototype multi-band radar systems will be installed at two locations that are coincident with standard snow observations. Automated processing will be improved and uncertainties in snow estimates defined. Broader/Commercial Impact: The commercialization of a Flat Earth SWE sensor would allow government and private organizations to expand their networks into these critical higher elevations, and cost effectively increase the number of sensors in their networks, improving the accuracy of forecasting for snow related water resources. Sustainable management of water in agriculture is critical to increase agricultural production, ensure water can be shared with other users and maintain the environmental and social benefits of water systems. ?
Animal Health Component
25%
Research Effort Categories
Basic
0%
Applied
25%
Developmental
75%
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
11202102050100%
Knowledge Area
112 - Watershed Protection and Management;

Subject Of Investigation
0210 - Water resources;

Field Of Science
2050 - Hydrology;
Goals / Objectives
The main goal of this proposal is to resolve the inability to accurately determine SWE in late season wet snow conditions. The result will be a sensing solution capable of estimating and monitoring the critical state variables of the snowpack (depth, SWE, density, liquid water content) autonomously in real-time. The goal is to be well positioned for a Phase II award where we can begin the prototyping and commercialization of our SWE sensor.?optimizing antennas for wet snow conditions,combining a 7-9 GHz and 24 GHz radar systems for accurate automated snow surface detection during snowfall events,collecting radar observations with the multi-band radar system in a wide range of snow conditions at multiple sites for one winter season, andusing the data collected to fine-tune the calibration of the automatic real-time algorithm, in both wet and dry snow conditions.
Project Methods
A new processing method was developed for tower-mounted radar in which SWE can be calculated using only the ground surface reflection in the radar signal, by approximating that the snowpack is a two-component system comprised of air and ice. For the majority of the winter accumulation season in alpine snowpacks the mean daily temperature is below freezing, the snowpack is entirely frozen, and this is a reasonable approximation. The new method uses the time delay between when the ground reflection appears without snow, and when it appears with snow on the ground, which greatly simplifies the processing and allows SWE to be calculated automatically on-site

Progress 08/01/18 to 06/30/19

Outputs
Target Audience:The goal of this research contract is to develop techniques that provide high-resolution snow property estimates, to improve forecasting of water resources for agriculture and public consumption, mountain weather, avalanches, floods, and hydro power. Stakeholders directly invloved in this Phase I research included: Idaho Power Company - currently using our Gen 1 systems and we shared this past seasons data with them. USDA-NRCS (SnoTel) - collaborated with them on collocation of sensors with SNOTEL sites, and have shown them comparisons to their SNOTEL sites. The USDA Northwest Watershed Research Center - discussed last season results and methodology. NASA SnowEx Mission - Gen 1 system currently being used in this project and we discussed our measurement techniques and results European Hydro Energy Companies - Showed results and Gen 2 hardware discussedestimating hydro output based on a netwiork of SWEdar sensors. Changes/Problems:During the 2018 - 2019 winter data collection season we encountered a fewchanges/problems. None were considered major in nature. Lost One Week of Data -The only gap in data transmission occurred for a little over a week in February, when the solar panels were covered with snow for over 3 weeks while Highway 21 was closed due to avalanche danger. This will be avoided in the future with the next radar generation (Gen 3 SWEdar) requiring much less power. Excessive Power Consumption - The Gen 2 SWEdar sensor contained three different radars. For prototyping purposes each radar required its own CPU. This resulted in excessive power consumption and required600W solar arrays, and a large enough battery bank to run the system for 3 weeks without solar input. This power issue will be resolved in the next generation sensor (Gen 3 SWEdar) when all radars will operate off of one CPU, dramatically reducing power consumption. 24 GHz Radar Malfunction - The Gen 2 SWEdar systems containthree different radars at different frequencies. The highest frequency system, a 24 GHz FMCW radar, had limited bandwidth and proved to be the least useful of the three systems. It also caused the entire Gen 2 SWEdar to crash after approximately 6 hours of operation. The cause of this is unknown, and early in the season it was decided, based on a month of data, to focus on the other two radar systems, since operating the least useful system put the data collection of the other radars at risk. What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest?Preliminary results from the 2018 - 2019 season have been shared with several communities of interest including regional and government level power companies. Results have also been shared with several government organizations that focus on waterresouorces and snow water content. What do you plan to do during the next reporting period to accomplish the goals? Nothing Reported

Impacts
What was accomplished under these goals? 1. Optimizing antennas for wet snow conditions For this Phase I effort we designed and fabricated a multi-band radar system (Gen 2 SWEdar), that includes 3 separate radar systems, spanning 1-24 GHz. This required the design and development of two separate antenna types to best match the radar frequencies, required field of view, and gain required to penetrate deep wet snow. For the X1 frequency radar Flat Earth developed an AVA 0.8 - 4 GHzantenna. This is an Antipodal Vivaldi styleantenna designed for UWB radar applications. This antenna makes it a good choice for applications such as Ground Penetrating Radar (GPR). For the X4 frequency radar Flat Earth developed a compact Linear Tapered Slot style antenna. This antenna was designed to be able to detect the snow/air surface interface. 2. Combining a 7-9 GHz and 24 GHz radar systems for accurate automated snow surface detection during snowfall events. During the 2018-2019 winter, the focus of our research was to expand the frequency range of radar operation (from 2-6 GHz, to 1-24 GHz), to explore possibilities for producing real-time SWE estimates in all snow conditions. In addition, we tested amplifiers that increase the transmitted signal amplitude, so that the radar systems could be mounted over 3 meters above the surface, to allow measurement of deep snowpack conditions. To achieve the goal of acquiring radar data over a bandwidth greater than 20 GHz, we designed and fabricated a multi-band radar system (Gen 2 SWEdar), that includes 3 separate radar systems, spanning 1-24 GHz. Radar observations in wet snow are very frequency dependent, and the goal of this past season's research was to determine the optimal combination of frequencies to use for estimating SWE in all snow conditions. We built three Gen 2 prototype versions, tested them in a cold lab to temperatures down to -30 deg C, and installed 2 of these at both Banner Summitand Bogus Ridge (same sites as previous year). Our Gen 1 SWEdar systemswere also used to perform on board processing and transmit hourly SWE estimates all season. 3. Collecting radar observations with the multi-band radar system in a wide range of snow conditions at multiple sites for one winter season. Two Gen 2 multi-band radar systems were installed at both Banner Summit and Bogus Ridge. Our Gen 1 SWEdar systemswere also used to perform on board processing and transmit hourly SWE estimates all season. The only gap in data transmission occurred for a little over a week in February, when the solar panels were covered with snow for over 3 weeks while Highway 21 was closed due to avalanche danger. This will be avoided in the future with the next radar generation (Gen 3 SWEdar) requiring much less power. At both the Banner Summit and Bogus Ridge sites, SWE estimates were accurate to within a few centimeters, with values highly correlated with nearby SNOTEL sites.At the Bogus Ridge site in dry snow conditions, transmitted SWE values agreed with snow pit observations and the nearby Bogus SNOTEL site, to within a few cm of SWE.RMSE between the transmitted radar SWE and SNOTEL SWE was 2.9 cm in dry snow, with r2 = 0:96). At the Banner Summit site, transmitted SWE values (Gen 1 SWEdar) agreed with SNOTEL SWE values to within a few centimeters in dry snow conditions as well. RMSE was 2.4 cm, and r2 = 0:98. At this site the SNOTEL sensor is within 10 meters of the radar systems. Higher frequency radar is more sensitive to the snow surface and internal snow layers, but is attenuated more in wet snow. At the lower X1 frequencies, detection of the snow surface in dry cold conditions is very challenging, but the X1 can more easily penetrate wet snow. We can show forthe Winter 2018-19 season at Bogus Ridge, for the X1 radar, where the snow-ground interface is throughout the season. We can also show forthe same time period for the X4 radar system, a clear snow surface reflection and internal snow layering, but the initial layering and ground reflection disappear when the snow becomes wet at the end of February. 4. Using the data collected to fine-tune the calibration of the automatic real-time algorithm, in both wet and dry snow conditions. The X1 radar component of the Gen 2 SWEdar is clearly superior for penetrating wet snow conditions to detect the ground,while the X4 radar canaccurately detect the snow surface reflection. We are currently experimenting with algorithms that combine the information from the X1 and X4 radar frequencies into one retrieval algorithm. Preliminary results, using winter 2018-19 data, show the new algorithm implemetation improvement is clear, but this approach needs more testing. Our goal is to test this multi-band radar approach with real-time processing this coming winter.

Publications

  • Type: Journal Articles Status: Awaiting Publication Year Published: 2019 Citation: Robertson, M., H.P. Marshall, Roberts, D., and Kunkel, M. (in review). Continuous snow water equivalent measurements with ulra-wideband impulse radar: A new sensor and time- delay method of processing downward-looking radar. Hydrological Processes