LTER members participate in the Environmental Sensor Network Workshop at Hubbard Brook
Don Henshaw (AND), Corinna Gries (NTL) and Lindsey Rustad (HBR)
From October 25-27, 2011, 72 people participated in the Northeastern Ecosystem Research Cooperative (NERC) Environmental Sensor Network / LTER SensorNIS Workshop at Hubbard Brook Experimental Forest LTER in New Hampshire. The meeting agenda, participants and notes are posted online at http://im.lternet.edu/projects/SensorNIS. The workshop focused on the emerging roles of environmental sensor technology and management in providing high quality, near real-time data on the physical, chemical, and biological properties of ecosystems. The workshop featured plenary talks from NEON, CUAHSI, USFS, USGS, LTER, NCEAS, and DataTurbine participants. Talks also highlighted sensor network capabilities at several LTER and northeastern research sites. Twenty-eight LTER participants from 20 LTER sites joined over forty northeast research station representatives and Hubbard Brook researchers to create a diverse mix of researchers, graduate students, land managers, information managers, and field technicians.
A pre-workshop survey showed that there is no lack of important and pressing research questions that new and/or more sensors could answer at this point in time. However, cost, technical and software know-how were the top reasons limiting sensor deployment, especially in cases where managing the data deluge requires specialists and pricy hardware solutions at the home institution. Remote locations pose an especially difficult challenge in terms of data transmission. Although almost all participants are maintaining a standard weather station, only about 15 % of the respondents stated that they have sensor systems at their site that meets their needs. A wide variety of systems are employed at these research sites. These are used in response to the different research questions, environmental conditions, and options for data transmission. Campbell Scientific hardware and software are the most frequently used technologies.
Much of the workshop was devoted to breakout group discussions that considered software and technology for both managing the data and performing quality assurance and quality control (QA/QC) in a streaming data context. Plenary talks included topics on the NEON Information System, the DataTurbine initiative, extensions to the Kepler scientific workflow system to accommodate sensor networks, as well as the Georgia Coastal Ecosystem MATLAB-based Data Toolbox. Sensor site establishment, types of sensors and communication platforms, data collection issues, data processing middleware, data archiving tools, and data access options and trade-offs were also considered. The types of quality control procedures that can be used for streaming data were examined and the need for documenting data screening and flagging procedures was emphasized. We also explored the idea of Data Levels to differentiate raw data from quality assured data and gap-filled, modified or derived data. The results indicate the need for a community-developed knowledge-base or “best practices” document for sensor network establishment and management.
An outline for a best practices document was generated as a workshop product and published online (http://im.lternet.edu/resources/im_practices/sensor_data). The outline follows the sensor data life cycle and on the data collection side considers selecting sensors, building a sensor platform, and choosing a site. The cycle outline includes streaming data from field to lab and managing the data streams with middleware to apply QA/QC routines and archival storage. Data access is considered at several steps along the way with the different goals of monitoring sensor health versus use to answer scientific questions. The intent is that this best practices outline will be populated interactively with short one-page descriptions by community members, although the meeting organizers will likely target specific individuals to provide specific content as a means to assure completion of the document.
The best practices for sensor networks document will provide a quick resource for sites establishing a sensor network or implementing a sensor management system. A research paper describing QA/QC procedures for streaming sensor data was also outlined and the authors will pursue publishing this. Other “next steps” include preparation of an LTER training proposal such as “tools and training for sensor network establishment and management”, virtual water cooler discussions on sensor network topics, proposing a sensor workshop for LTER ASM 2012 at Estes Park in September, and planning a future DataBits edition devoted to sensor network experience articles from sites. Meeting organizers will take advantage of future workshop products as a means of continually updating the online best practices document.
The workshop was funded in part through National Science Foundation (NSF) LTER supplemental funding for information management. NSF encouraged LTER sites to use these funds for coordinated, cross-site projects, and funds were granted to LTER sites to participate in this workshop. Additional funding was provided by NERC to cover northeastern participants and Hubbard Brook meeting costs. Lindsey Rustad (HBR) of the USFS Northern Research Station, Don Henshaw (AND) of the USFS Pacific Northwest Research Station, Corinna Gries (NTL), Jamie Shanley (USGS), and Peter Murdock (USGS) comprised the planning and organizational team. Hubbard Brook research staff provided considerable meeting support including hosting a site field trip and preparing meeting and poster session space.