|| Checking for direct PDF access through Ovid
During the 2006–2007 winter season, 17 sites across the US including Alaska tested an automated snow measurement system. This article aims to describe successes and failures of this system and provide insight into data collected this season. The system was designed in collaboration with both Environment Canada and Snow Sensor Study participants during the summer of 2006. This system included three Campbell Scientific SR-50 sensors oriented 120° from one another and a temperature probe centred in the plot. Data collection efforts were successful with minimal amounts of data missing because of system or sensor failures. The system integrated automated retrieval of data from dataloggers, as well as automated file transfer protocol (FTP) to the study website for data archival and graphical display.Overall, the sensors and installation worked well with only a few problems noted. The sensors compared well with both manual observations taken adjacent to each sensor as well as traditional total snow depth (TSD) on ground measurements. The comparison to depths, taken adjacent to the sensors, allows for investigation of frost heave and indicates periods where the sensors were not functioning properly. The comparison to TSD on ground reveal problems with siting at some locations that are recommended to be remedied by re-installation or re-location of those sites prior to the 2007–2008 snow season. These results are preliminary and research will be ongoing for signal processing, snowfall algorithm development and optimal installation in preparation for the 2007–2008 snow season. This research has potential to return important snow observations to national weather service(NWS) observing networks that were discontinued when automation began as well as provide continuous snowpack monitoring to data users. Copyright © 2008 John Wiley & Sons, Ltd.