Annals rifGlaciology 24 1997 © Int ern ational Glaciologi ca l Society Using ground-penetrating radar to image previous years' summer surfaces for mass-balance measurement s J ACK KOHLER/ JOH N MooRE,2 MIKE K ENNETT,I R UNE ENGESET/ HALLGEIR ELVEH 0yl I Glacie r and Snow Section, NVE (No rwegian Water Reso urces and Energy Administration ), PO. Box 5091, Maj., 0352 Oslo, Norway 2 Th e Arctic Centre, University rif Lapland, Box 122,96101 Rovaniemi, Fin land A BSTRACT. In tradition al mass -b alance meas ur ements one es timates winter snow acc umulation by identifying the depth to the previo ll s summer 's snow or ice surface using a snow pr obe. This is l abo r-intensive a nd unr eli able for inhomog eneous summe r surfaces. Anot her me thod is to imag e inte rnal refl ection horizons using a g round-p ene tratin g rad ar ( GPR ), which has ad va ntages in speed and ar ea l cove rag e over traditional probing. How- eve r, to obtain quantitative mass-balance meas ur ements from G PR images one needs to co nve rt th e time scale to a de pth scale, not a s tr aig htforward problem. ''''e co mpar e a GPR section with dielectric profiles and visual s tr a ti gra phy of three snow cores, ma nual prob- ings, a nd previous mass-balance meas urement s. We relate chang es in snow-core dielectric prop e rti es to c hang es in density and to the travcl times of re fl ec ting horizon s in the GPR section, and co rrelate some of these re fl ecting horizons with previous summer s urfaces. We conclude that GPR ca n be used as a co mplementa ry tool in mass -balance meas ur ement s, giving a wide area l surv ey of winter accum ul at ion and n et ba lance for prece ding years. H owever, prop er ca libr at ion is essential for identifying spec ifi c s urfaces in the radar dat a. INTRODUCTION Each spring, glaciologists engaged in "conventi ona l" mass - ba lance measurements can be found on the surface of their respective glaciers, snow prob es in ha nd , thru sting into the previous winter's snow. Th eir goa l is to es timat e wint er accumul at ion by identifying th e de pth to the pr evious sum- mer's snow or ice surface ( 0strem and Bru gman, 19 91). Among the disadvantages of manual probin g, however, are th at it is relatively time-consu min g to track an often in- h omogeneo us that it ca n be h ar d work if there are any int ervening ice layers, as ther e often are, and that the presence of other ice layers in the acc umul ation ar ea often makes it difficult to distinguish the su mmer layer. Tower 5 km / Summit Another me thod is to im age inte rnal reOection hor izons using a grou nd-p ene trating radar ( GPR ), as eith er an alt er- n at ive or a complement ary m et hod to conventional manual probin g. Rad ar has been used previously to determin e accu mulation rat es by finding datable refl ecting hor izons in Ant arct ic firn ( Fo rster and ot her s, 1991; Wee rtm an, 1993; C. Ri cha rd son and ot her s, unpublished informa ti on) and on temperate glaciers ( Holm l und a nd Ri ch ardso n, 1995). 1850 .,---------------.... However, to o bt ain qu a ntit at ive mass-balan ce measure- ments from GPR images on e needs to conv e rt the time - dep e ndent radar r et urn signal to a depth-dependent strati- grap hic profile, not a straightforward problem. Th e com- pl exity of the snow strati graph y, the se nsitivity of the depth - time relationship to changes in density, chemis try, grain-size and water co ntent , and th e possibi li ty of phas e chan ges on refl ection and inte rference from nea rby re fl ect- ing hor izons make such conversion difficult. GPR imag es of s nowpacks are a convolution of the input rad ar wave with the physi cal prop erti es of the snow that ca use echo es, such as density co ntrasts or chan ges in snow 1840 .. .. .s 1830 c g .. > <D W 1820 1810 0 200 400 600 800 Distance along section (m) Fig. 1. Location map rif Hardangeljokulen, with eleva ", m prqfi.le along GPR section. 355 https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0260305500012441 Downloaded from https://www.cambridge.org/core. IP address: 149.36.2.136, on 23 Sep 2017 at 19:35:47, subject to the Cambridge Core terms of use, available at