1 Evaluation of radar measurements Hans-Peter Marshall, Boise State University and CRREL Snow Characterization Workshop, April 13-15, 2009
Locate instrumentation-related signals…
And get rid of them!
Locate causes of major reflections Metal reflectors placed at known depths, to determine cause of reflections in original signal
Metal reflector experiment
Accuracy of using mean dielectric properties to estimate velocity: < 2%
Comparing FMCW signal to in-situ electrical measurements radar => in-situ dielectric properties (Finish snowfork) [e.g. Harper and Bradford, 03] In-situ properties => physical properties (e.g. Sihvola et al, 1985; Schneebeli et al, 1998; Matzler, 1996)
In-situ Density and Wetness
In-situ Reflectivity
Radar Snow Water Equivalent Estimates
Comparison of radar with SMP at Swiss Federal Institute for Snow and Avalanche Research => Small diameter rod driven through snow at constant velocity, pressure measured at tip 250 measurements/mm Measures rupture force of grain bonds SnowMicroPenetrometer
Snowpit comparison, SLF, Feb 19, 2004
Multi-Layer Model (e.g. Ulaby et al, 1981)
3-layer model – complicated for thin layers
Depths of major reflections automatically picked
Comparison of FMCW radar and SnowMicroPen
Chuckchi Sea, Barrow March, meter profile on 1 st year sea ice 601 MagnaProbe measurements >3000 FMCW radar snow depths
Static comparison 1) Expected error = velocity uncertainty (1.5 cm) + radar resolution (1.5 cm) + difference in horizontal support (2cm) = 5cm 2) Mean values within 1.5 cm
Density/Velocity distribution from SWE cores +/- 5% uncertainty in depth estimate due to density variability
FMCW radar profile Mean measured density used to estimate depth from radar TWT
FMCW radar / Magnaprobe comparison 1)Similar variability, good agreement 2)Differences mainly due to different support and coregistering of measurements
Comparing point depths to radar measurements
1.7 km profile, x=10 cm, z=1.5 cm
Conclusions - limitations Signal attenuated in very wet snow Magnitude information from reflections difficult to interpret for thin layers No mechanical / microstructural information
Conclusions - advantages Rapid (50 Hz) estimates of snow depth, SWE, major stratigraphic boundaries Basin-scale areas can be covered Slab geometry can be measured Simulate active microwave remote sensors