Columbia Glacier, Alaska, USA Timelapse Photography Collection version: 2011-10-14 // Primary Contact Ethan Welty Institute of Arctic and Alpine Research 1560 30th Street Campus Box 450 Boulder, CO 80309 (206) 853-3712 ethan.welty@colorado.edu // Time Calibration Methods The Columbia Timelapse Photography Collection is a result of dozens of camera deployments over several years. Until very recently, the time zone setting and accuracy of the internal camera clocks were not properly documented, and as a result, it was necessary to retroactively time-calibrate the original sequences (to UTC) using environmental information contained in the images themselves. The time adjustment applied to each sequence is recorded in the Timelapse_Inventory spreadsheet. Most older sequences were adjusted with a simple time offset, described either in decimal seconds or in HH:MM:SS format. or Since August 2009, camera clocks have been compared against handheld GPS clock displays (accuracy 1 second) at time of servicing. The most contemporary sequences benefit from time adjustments that correct not only for simple offset, but for linear clock drift as calculated from repeat GPS offset measurements. Such adjustments are described by an offset as well as a drift, in decimal seconds per day, and a reftime, the UTC time at which the specified offset was measured. Although our measurements in the lab confirm that camera clock drift is highly linear, measurements from the field indicate that the magnitude of the drift varies significantly with temperature, with clocks slowing in winter. Since we only visit the cameras in early and late summer, our measurements of summer clock drift are representative, but we cannot precisely determine the accuracy of the adjusted capture times for the rest of the year. For fall, winter and spring, the corrected drift is an average of a highly varying, temperature sensitive drift. No such information exists for the older sequences however, and less precise strategies had to be leveraged to fix the sequences to UTC. Most of these rely on identifying, in the images themselves, events of known time. Tides (tidesandcurrents.noaa.gov/get_predictions.shtml?year=2008&stn=0919+Valdez) The periodic motion of the tides is readily observed in the time lapse sequences, either as change in the water level against the ice cliff or, in the case of a floating terminus, as vertical motion of the glacier itself. The timing of low and high tides can be inferred to within 1/2-1 frame interval, then compared against tide tables for nearby Valdez, AK to calculate the UTC offset of the camera. Sunrise, Sunset, Solar Zenith, and Peak Darkness (www.sunrisesunset.com/custom_srss_calendar2.asp) Whether or not the sun is visible in the images, it's cycle is recorded in the sequences by the angle of the shadows it casts, the color and intensity of its light, and the exposure settings of the camera throughout the day and night. Depending on the situation, an estimate of sunrise, sunset, or solar zenith (noon) – or a combination thereof – could be obtained and compared against astronomical tables. “Peak darkness” was used only as a last resort: camera exposure settings were used to determine minimum light intensity, which was assumed to fall halfway between sunset and sunrise. In a few instances, an observed calving event was identified in the seismic record, providing a highly accurate time calibration. The “continuity assumption” indicates that, with insufficient frames to use one of the methods described above, the sequence was assumed to need the same time adjustment as the previous sequence for that camera. In all cases, we did our best to provide a realistic estimate of the accuracy of the adjusted times. For most older sequences, accuracy is limited by framerate and the time “softness” of the observed event, and usually on the order of one hour. For the contemporary sequences, accuracy is limited by the temperature sensitivity of camera clock drift, and on the order of seconds to a few minutes.