The idea is to use modern dGPS units - one fixed nearby and one on the glacier (rover) computing its location using data from the base station. So rather than record large volumes of data we get fixes computed onsite.
L1 only system but reasonably priced and quite well documented
C94-M8P - (docs) tests in UK with short baseline and very static stock antennas show 0.02 to 0.01m precision after about 5min relock.
piksi multi from Swift
L1 and L2 so expect higher accuracy
Rockblock Iridium unit
These will be used to transmit the location data.
need to look into onboard battery - as unit will be in sleep more a lot (20uA using sleep pin)
if using pwr header it can draw 470mA - (rather high for LiSOCs)
Batteries and power management
Plan to turn gps units off between measrements of about 5min with a fixed base station nearby.
Try Lithium ion?
need to evaluate the systems once running.