New system technologies for snow water equivalent monitoring with improved temporal and spatial resolution PIs: Jeff Frolik (Engineering), Christian Skalka (Computer Science), and Beverley Wemple (Geography) The University of Vermont MOTIVATION Seasonal snow packs are an important element of the water and energy balance in high latitude and high elevation settings. In many settings, snow packs also support an active recreational economy during winter, and snowmelt supplies a considerable fraction of water used for agriculture, drinking water, industrial demands and ecosystem services. In addition, adequate understanding of the spatial distribution of seasonal snow water is critical in flood and water supply forecasting Since the density of snow varies both spatially and temporally, a common measure of seasonal snow pack is the aggregate snow water equivalent. The snow water equivalent (SWE) of a snow pack measures the amount of water stored therein; SWE increases due to mass accumulation (i.e., continued snowfall and wind loading), and decreases due to mass loss (i.e., sublimation and melt). SWE is most commonly determined by measuring the density profile of the snow pack from manually collected snow cores. More recent methods extrapolate such data spatially using ground-penetrating radar data collected from the snow surface1. To improve resolution, other research has employed LiDAR to interpolate between measurement points 2 . In spite of this advancement, measuring and quantitatively understanding the spatial distribution of SWE on scales larger than a few hundred square meters still requires extensive human resources. Prior research has demonstrated that SWE varies due to the interaction between meteorology (wind speed and direction, radiation) and both small- (vegetation or rocks) and large-scale topography (hills, valleys, slopes)3. . Developing a system to cost effectively collect data to improve the spatial and temporal resolution of SWE monitoring would have important impacts on both the scientific study of snow and social infrastructures dependent on snow. To date, using sensor technology to assess SWE has had some success but current approaches are costly and not scalable to the desired spatial resolution. The prevailing technology is based on snow pillows. This approach uses pressure sensors to detect the weight of a snow pack on flat rubber bags filled with antifreeze, buried level with or slightly below the ground. Snow pillows benefit from being as large as possible to maximize snow pack profile measurement, with sizes typically near 200 gallons. This makes snow pillows difficult and costly to deploy, as well as an environmental hazard. An alternative approach uses large plates sitting on load cells to weigh the accumulated snow. But both approaches have been found to be susceptible to snow bridging and other thermal 1 effects that reduce their accuracy. Finally, these types of systems can only be installed in large, easily accessed, flat areas, and are therefore not feasible for remote, mountainous terrain or environmentally sensitive areas. Another shortcoming of prevailing technology has to do with data transmission and storage. The communications and storage infrastructure for in situ systems using data loggers such as the CR1000 and radio modems for communications run several thousands of dollars and are not easily programmed to adapt to specific application needs, e.g. to implement in-network processing of data for reducing data storage requirements and controlling sensor power consumption.. PROJECT OVERVIEW Our interdisciplinary research project will reconsider in situ SWE monitoring systems from the ground up. This whole-systems approach distinguishes our work, and will integrate new sensor technologies with new computational platforms for data acquisition, in-network data processing, and data storage/retrieval. The emphasis throughout our project will be on leveraging cutting-edge technology to significantly reduce the size, weight, cost, and power requirements of SWE monitoring systems. New sensor technologies will work by measuring snow pack attenuation of electromagnetic radiation, in particular microwave and cosmic gamma radiation. New computational platforms will be based on wireless sensor networks composed of motes, which are low-cost, tiny, low-powered programmable devices that communicate via radio transmission. Motes are designed to easily interface with arbitrary sensors, so the computational and sensing aspects of our project can be developed orthogonally. Another significant component of our research will be the application of our system to benefit existing hydrological research projects at UVM. These projects rely on SWE data that is currently collected manually via snow cores. We will enhance and hope to eventually replace manual data collection, but in the meantime manually collected data will serve as ground truth for evaluation of our prototypes. STATE OF THE EXISTING WORK Our present work involves researchers from electrical engineering (Frolik), computer science (Skalka) and Geography (Wemple) along with undergraduate and graduate students from these varied disciplines. With seed funding from NASA/Vermont Space Grant Consortium, we have developed a system using compact snow pillows and load-cell scales as a simple proof-of-concept of our computational platform. This wireless sensing system has been deployed to monitor four locations at the base of Mount Mansfield (VT) during the 2007-2008 snow season. Manual snow core measurements will be performed bi-weekly to provide a point of comparison for our real time data. To complement this effort, we have visited with researchers at the Central Sierra Snow Laboratory (CA) and at field sites in the Sierra Nevada range managed by UC-Berkeley and UC-Merced to investigate state-of-the-art approaches and discuss research directions. 2 PROPOSAL DETAILS In this section we detail the direction of our future research. We will consider the three main components of our research program in turn: new SWE sensor technology, WSN computational platforms, and in situ application and evaluation of our integrated SWE monitoring system. New SWE sensor technology. In order to sidestep the problems associated with snow pillows and scales (i.e., size, cost, susceptibility to snow bridging), we propose to investigate measuring SWE via attenuation of electromagnetic radiation in snow packs. By measuring the difference between radiation above and below a snow pack and applying known attenuation rates and analytic tools, it is possible to infer the water content in a snow pack. This approach is not affected by snow bridging, can be used in all sorts of terrain, and the sensor hardware is significantly smaller, cheaper, and more easily deployable. We propose to investigate two frequencies of electromagnetic radiation: gamma rays and microwaves. Measurement of SWE via detection of cosmic gamma ray attenuation in snow packs has been studied previously by Osterhuber et al.4. Gamma ray detection has distinct benefits in this application, including a free and constant radiation source, and a known attenuation rate in snow that is appropriate in the sense that it is cost-effectively detectable in even the deepest snow packs. Indeed, these techniques were successfully tested in very extreme conditions. We believe that detecting gamma ray attenuation is a promising approach, but there remain significant obstacles to overcome. In particular, to maintain low cost of deployment, it is necessary to find a balance between detector sensitivity and cost. Osterhuber et al. focused on detection of energies between 3MeV and 5MeV, to filter out interference from geologic gamma radiation sources. A drawback of this approach is that detection of energies in the 3MeV to 5MeV range requires the use of scintillation detectors or Geiger-Muller tubes, which cost several thousand dollars on the current market. This is too expensive for our project vision. However, a far cheaper solution is on the horizon. The Domestic Nuclear Detection Office (DNDO) of the U.S. Department of Homeland Security has recently incentivized the development of low cost gamma ray detectors sensitive in the 50KeV to 3MeV range as a component of the war on terrorism. This is documented and evidenced by the recent research funding efforts of the DNDO Transformational & Applied Research Directorate5 and a recent Request for Information on Gamma Radiation Detection on FedBizOpps6. Therefore it is likely that this technology will be available soon. In particular, our discussions with industry representatives suggest that detectors in the 500KeV to 1MeV range will be available in the nearest future for $150-$300. Although the sensitivity range of these detectors will be lower than the scintillation detector used by Osterhuber et al., we speculate that passive shielding techniques (e.g., lead enclosures) will filter interference sufficiently for these devices to be usable in a near-term, proto-type SWE monitoring systems. In the meantime, we intend to purchase a Cadmium Zinc Telluride (CZT) gamma ray detector from eV products that is sensitive in the 500KeV to 1MeV range. Experiments with this device will provide proof-of-concept for SWE monitoring using detectors in its sensitivity range, and allow us to anticipate the future of lower-cost devices with similar sensitivity. 3 We also propose to leverage low-cost microwave sensing to ascertain snow pack SWE. The attenuation characteristics of microwaves been well studied7,8 and in general are known to be high and dependent on water content. We will utilize this knowledge to develop a dual band system (2.4 GHz and 5 GHz) in which signal strength is monitored at various heights. While attenuation rates are expected to be 20 dB/m to 100 dB/m for these bands depending on snow properties, we contend that measurements with large dynamic range, at various heights and at different frequencies, can be used to solve for SWE and potentially snow structure. This approach is especially attractive since the hardware technology, in contrast to gamma ray detection, is cheap and readily available. Fig. 1 (left) illustrates our working concept. Multiple modalities for snow characterization are considered. First, a gamma detector will be placed at ground level while another will be located above the potential snow pack. The differential measurements will be utilized to determine the attenuation due to the snowpack and subsequently water equivalency. Second, microwave transmitters will be placed both at ground level and above the foreseen snow pack. Microwave detectors will be located at stations between the transmitters. Due to the expected large attenuations rates, signals from one side of the pack may not reach all stations. Furthermore, conducting measurements from the top and bottom will enable a better understanding of snow evolution as a function of height. Finally, an ultrasound measurement system will be used to determine snow height at the site location. Other parameters to be monitored include temperature and solar radiation. This suite of sensors will be connected to a wireless mote. A distributed set of stations will self-organize in order to communicate the data back to the desired sink (e.g., an existing network or satellite phone). Based on our proof of concept deployment, we expect communication links up to 100m (local geography permitting). Figure 1 Proposed multi-model SWE monitoring station (left), distributed deployment (right) WSN computational platforms. In addition to sensors, sensor systems include a means to acquire and interpret data from sensors and to log or transmit data for subsequent permanent storage. WSNs are a new computational platform allowing wireless radio communication between sensor nodes. An Ethernet interface allows easy 4 integration with personal computers. WSN networks are self organizing, in the sense that they will automatically form a network upon deployment and adapt to environmental changes and node failure. They are also low-cost, tiny, and robust, and easily integrate with a wide variety of sensors. Furthermore, WSNs are programmable, and also easily reprogrammable, allowing them to be adapted to a variety of applications, and subsequently refined in those applications and reused for others. All of these features make them an extremely appealing platform for environmental monitoring, for applications as diverse as detecting seismic activity of volcanoes9, monitoring soil moisture content 10 , and measuring snow height 11 . The low cost of WSN motes (sub $100 presently) and ease-of-use makes this technology within the means of virtually any user. Significant related challenges for our project will be developing affordable data logging and remote transmission solutions, and evolving in-network processing for data interpretation, data compression, and data storage. On-mote data logging can be obtained at very low cost, but current motes have only 1MB of external flash memory so storage must be either distributed through the network, or compressed, or both. In-network processing will allow weighted moving averages instead of raw data to be stored, significantly reducing space requirements. Remote transmission of data via cell or radio modems will be possible in appropriate areas; indeed, a single radio modem could service multiple SWE monitoring sites with line-of-sight. We propose to investigate each of these alternatives, and note that the use of WSNs allows flexibility and cost-effectiveness of SWE data collection solutions, especially via their in-network processing capabilities. Successful, cost-effective integration with our proposed sensor technology will also require in-network actuation and control of sensors. For example, our current prototype system must turn the load sensor in its snow pillow on and off for hourly sensing bursts to conserve system energy. Control of electromagnetic radiation detectors and microwave sources will be far more subtle and consider a variety of parameters besides time. Control algorithms that optimally balance power usage and detection precision to integrate WSNs with electromagnetic radiation detectors will be an exciting research challenge. Application to Hydrology Research and System Evaluation. In order to perform a field test of our sensor, we will deploy our prototype as a data-collection node in an established snow monitoring network associated with the Mt. Mansfield paired-watershed study (MMPWS). The MMPWS is a collaborative research project between the University of Vermont (PI B. Wemple) and the U.S. Geological Survey (PI J. Shanley). Its objective is to develop baseline hydrologic and water quality data for high-elevation settings in northern New England, and to examine the effects of development on water, solute and sediment dynamics 12. This research site is one of a number of long-term forest ecosystem monitoring and research sites supported by the Vermont Monitoring Cooperative (http://sal.snr.uvm.edu/vmc/). Snow studies associated with the MMPWS commenced in 2003 with spatially-distributed field sampling using manual snow coring methods collected at peak accumulation and approximately bi-weekly during the snowmelt period. These data have been used to develop precipitation lapse rates to spatially distribute snow water over the 5 watersheds. Current research efforts are aimed at evaluating the effects of the forest canopy and forest openings on snow accumulation and melt and will provide a field validation dataset for simulation modeling of snowmelt dynamics. Deployment of our prototype SWE system within the MMPWS snow monitoring network is an ideal test case. The field sites are remote, situated in complex terrain, and require considerable manual labor to access and maintain. We propose installation of a single, continuously monitoring prototype sensor system near the USGS gaging station at 44°30'14"N, 72°46'56" W. This site is regularly accessed by our survey team, who will conduct roughly bi-weekly manual surveys of snow pack depth and SWE as a ground-truth for our sensor measurements. Our system design includes (2) nodes for sampling within a forested site and (2) nodes for sampling in an adjacent clearing. For our current deployment, the individual SWE stations costs are ~$1,200 per site (primarily due to load cell and pressure sensor costs). We expect the use of microwave and gamma ray detection to provide better data (due to the phenomenon being impervious to snow bridging effects) at a potentially lower cost. Data collected from the prototype system will be used in association with other manually surveyed data at our research site to examine the influence of topography and land cover on the spatial distribution of snow water and the dynamics of snowmelt, as part of a graduate student research project under the direction of PI B. Wemple. We will make the dataset from this installation publicly available through the online data archives managed by the Vermont Monitoring Cooperative. 1 Serreze, M., M. Clark, R. Armstrong, D. McGinnis, and R. Pulwarty. Characteristics of the western United States snowpack from snowpack telemetry (SNOTEL) data. Water Resources Research 35(7), p.2145-2160, 1999. 2 Deems, J., S. Fassnacht, and K. Elder. Fractal distribution of snow depth from LiDAR data. Journal of Hydrometeorology 7(2), p. 285 – 297, 2006. 3 Dingman, S., Physical Hydrology, 2nd Edition. Prentice Hall, Inc., 656p., 2002. 4 Randall Osterhuber, Frank Gehrke, and Ken Condreva. Snowpack Snow Water Equivalent Measurement Using the Attenuation of Cosmic Gamma Radiation. In Proceeedings of the Western Snow Conference, April 1998. 5 DHS DNDO Transformational & Applied Research Directorate, Ongoing Research Projects: http://www.dhs.gov/xlibrary/assets/DNDO_List.pdf. 6 DHS DNDO Request for Information on Mobile COTS and Gamma Radiation Detection Equipment: http://www.fbo.gov/spg/DHS/OCPO/DHS-OCPO/Reference%2DNumber%2DRFI%2DRadNucCOTS/Synopsis.html 7 Ulaby, F., R. Moore and A. Fung, Microwave Remote Sensing: Passive and Active, Vol. III, Artech House, Inc., 1986. 8 Abe, T., Y. Yamaguchi, M. Sengoku, Experimental study of microwave transmission in snowpack, IEEE Trans. Geosc. and Rem. Sens., Vol. 28, No. 5, September 1990. 9 Werner-Allen, G., K. Lorincz, M. Ruiz, O. Marcillo, J. Johnson, J. Lees, and M. Welsh, Deploying a Wireless Sensor Network on an Active Volcano. IEEE Internet Computing, Special issue on Data-Driven Applications in Sensor Networks, March/April 2006. 10 R. Musaloiu-E., A. Terzis , K. Szlavecz , A. Szalay, J. Cogan , J. Gray, Life Under your Feet: A Wireless Soil Ecology Sensor Network. EmNets 2006 11 Mountain Hydrology Research Group, UC-Merced 12 Wemple, B. C., J. Shanley, J. Denner, D. Ross, and K. Mills. 2007. Hydrology and water quality in two mountain basins of the northeastern US: assessing baseline conditions and effects of ski area development. Hydrological Processes, 21: 1639-1650. DOI 10.1002/hyp6700 6