Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | The Cryosphere Discuss., 6, 3503–3538, 2012 www.the-cryosphere-discuss.net/6/3503/2012/ The Cryosphere doi:10.5194/tcd-6-3503-2012 Discussions TCD © Author(s) 2012. CC Attribution 3.0 License. 6, 3503–3538, 2012 This discussion paper is/has been under review for the journal The Cryosphere (TC). Satellite-Derived Please refer to the corresponding final paper in TC if available. Volume Loss Rates A. K. Melkonian et al. Satellite-Derived Volume Loss Rates and Glacier Speeds for the Cordillera Darwin Title Page Icefield, Chile Abstract Introduction Conclusions References 1 1 1 2,3 2 A. K. Melkonian , M. J. Willis , M. E. Pritchard , A. Rivera , F. Bown , and Tables Figures S. A. Bernstein4 1Department of Earth and Atmospheric Sciences, Cornell University, Ithaca, New York, USA J I 2Centro de Estudios Cient´ıficos (CECs), Valdivia, Chile J I 3Departamento de Geograf´ıa, Universidad de Chile, Santiago, Chile 4St. Timothy’s School, 8400 Greenspring Ave, Stevenson, MD, 21153, USA Back Close Received: 12 July 2012 – Accepted: 31 July 2012 – Published: 31 August 2012 Full Screen / Esc Correspondence to: A. K. Melkonian ([email protected]) Printer-friendly Version Published by Copernicus Publications on behalf of the European Geosciences Union. Interactive Discussion 3503 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Abstract TCD We produce the first icefield-wide volume change rate and glacier velocity estimates for the Cordillera Darwin Icefield (CDI), a 2605 km2 temperate icefield in Southern Chile 6, 3503–3538, 2012 (69.6◦ W, 54.6◦ S). Velocities are measured from optical and radar imagery between 5 2001–2011. Thirty-seven digital elevation models (DEMs) from ASTER and the SRTM Satellite-Derived are stacked and a weighted linear regression is applied to elevations on a pixel-by-pixel Volume Loss Rates basis to estimate volume change rates. The CDI lost mass at an average rate of 3.9 ± 0.3 Gt yr−1 between 2000 and 2011, A. K. Melkonian et al. equivalent to a sea level rise (SLR) of 0.01 ± 0.001 mm yr−1. Thinning is widespread, 10 with concentrations near the front of two northern glaciers (Marinelli, Darwin) and one western (CDI-08) glacier. Thickening is apparent in the south, most notably over the Title Page advancing Garibaldi Glacier. We attribute this thinning pattern to warmer temperatures, Abstract Introduction particularly in the north, which triggered rapid retreat at Marinelli Glacier (∼4 km from 2001–2011). Conclusions References 15 Velocities are obtained over many of the swiftly flowing glaciers for the first time. We Tables Figures provide a repeat speed timeseries at the Marinelli Glacier. Maximum front speeds there accelerated from 7.5 m day−1 in 2001 to 9.5 m day−1 in 2003, to a peak of 10 m day−1 in 2011. J I J I 1 Introduction Back Close 20 The Cordillera Darwin Icefield (CDI) is a small icefield, located in the southernmost An- Full Screen / Esc des (Fig. 1) in Tierra del Fuego. The temperate icefield is coalesced around two main mountain peaks, Mount Darwin 2469 m a.s.l., (e.g., Koppes et al., 2009) and Mount Printer-friendly Version Sarmiento 2300 m a.s.l., (e.g., Strelin et al., 2008) and covers 2605 km2, measured from ice outlines derived from satellite imagery acquired from 2001 to 2004. It extends Interactive Discussion ◦ ◦ 25 roughly 200 km west-to-east from 71.8 W to 68.5 W and roughly 50 km south-to-north from 54.9◦ S to 54.2◦ S. The icefield is bounded to the north by the Almirantazgo Ford 3504 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | and the Beagle Channel in the south. Precipitation during the winter comes predomi- nantly from the south/southwest (Holmlund and Fuenzalida, 1995), and the E–W ori- TCD entation of the CDI leads to an orographic effect with greater snowfall on southern 6, 3503–3538, 2012 glaciers and drier, warmer conditions on northern glaciers (Holmlund and Fuenzalida, 5 1995; Strelin and Iturraspe, 2007; Koppes et al., 2009). There are few studies on the CDI compared to other temperate icefields (Masiokas Satellite-Derived et al., 2009; Lopez et al., 2010), such as the Alaskan icefields, the Northern Patagonian Volume Loss Rates Icefield (NPI) and the Southern Patagonian Icefield (SPI) (e.g., Arendt et al., 2002; Rig- not et al., 2003; Berthier et al., 2007, 2010; Glasser et al., 2011; Willis et al., 2012a). A. K. Melkonian et al. 10 Climate and mass balance studies are scarce for southern hemisphere ice bodies out- side of Antarctica (Holmlund and Fuenzalida, 1995; Lopez et al., 2010), due to the Title Page difficult access and weather. Here we focus on the third largest temperate icefield in the southern hemisphere (Bown et al., 2013), which along with the NPI and SPI, has Abstract Introduction experienced a rapid reduction in ice-covered area (Rivera et al., 2007; Masiokas et al., Conclusions References 15 2009; Lopez et al., 2010; Willis et al., 2012b). The loss of ice at the CDI has been attributed to climatic changes in the region, in- Tables Figures cluding warming during the 20th century (Holmlund and Fuenzalida, 1995; Lopez et al., 2010) (from climate station data) in conjunction with decreased precipitation (Quintana, J I 2004). Since the mid-20th century, and despite a reduction in regional precipitation, the 20 CDI in particular has experienced increased precipitation on its southern side (Strelin J I and Iturraspe, 2007), with decreased precipitation and warmer temperatures on its Back Close northern side inferred from NCEP-NCAR climate model results (Koppes et al., 2009). Temperate icefields are disproportionately large contributors to SLR (e.g., Arendt Full Screen / Esc et al., 2002; Rignot et al., 2003); Rignot et al., 2003 claim this is particularly true of 25 the Patagonian glaciers, which they say account for 9 % of the non-polar contribution Printer-friendly Version to SLR. The CDI, along with the NPI and SPI, provides an opportunity to examine the response of different glaciers (e.g. calving vs. non-calving) in different climates (mar- Interactive Discussion itime on the southern side versus more continental on the northern side) to regional changes in climate (Holmlund and Fuenzalida, 1995), and unlike the NPI and SPI the 3505 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | contribution of the CDI to SLR has not yet been estimated (Lopez et al., 2010). Finally, the CDI is the closest icefield to the Antarctic Peninsula, a region that has also experi- TCD enced significant warming. Thinning and acceleration have been observed on glaciers 6, 3503–3538, 2012 in the Antarctic Peninsula and the NPI (Pritchard and Vaughan, 2007; Willis et al., 5 2012a), we assess whether this is the case for any glaciers on the CDI. Mass loss at the CDI might be contaminating GRACE measurements of nearby icefields (e.g. the Satellite-Derived NPI, SPI and Antarctic Peninsula), so our constraints on the mass loss rate occurring Volume Loss Rates at the CDI will help isolate this signal. dh A. K. Melkonian et al. In this study we calculate both the elevation change rates ( dt ) over the entire CDI 10 and measure glacier velocities using pixel-tracking applied to pairs of optical and radar images. With dh and an assumed ice density we can estimate the mass change rate dt Title Page and give a measurement of the CDI’s contribution to SLR, allowing us to compare its SLR contribution to other icefields. We can also use the surface elevation change rates Abstract Introduction to identify which glaciers are providing the largest contribution to SLR and should be the Conclusions References 15 focus of further study. Additionally, measuring glacier velocities allows an estimate of mass flux out of the glacier if the thickness is known. Our results will provide a baseline Tables Figures measurement over many glaciers and areas of the icefield for which ice velocities have not been measured. J I J I 2 Methods Back Close 20 2.1 Data preparation Full Screen / Esc The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) has a stereo-imaging capability, enabling DEMs to be generated on-demand by NASA’s Printer-friendly Version Land Processes Distributed Active Archive Center (LP DAAC) (Fujisada et al., 2005). dh Interactive Discussion ASTER DEMs (product 14) are used to calculate dt , while band 3N images (product 25 1B) are used for pixel-tracking. NASA’s Automatic Registration and Orthorectification Package (AROP; Gao et al., 2009) is used to co-register ASTER images and DEMs to 3506 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | a Landsat GLS image (available from the Global Land Cover Facility) and orthorectify the ASTER L1B images using the Shuttle Radar Topography Mission (SRTM) DEM TCD (acquired in 2000). Landsat GLS images are orthorectified to the SRTM DEM (Tucker 6, 3503–3538, 2012 et al., 2004), so co-registering the ASTER imagery to the Landsat GLS image effec- 5 tively co-registers them to the SRTM DEM (see Melkonian, 2011 and Willis et al., 2012a for details). Satellite-Derived Volume Loss Rates 2.2 Elevation change rates A. K. Melkonian et al. Horizontally co-registered ASTER DEMs are vertically co-registered and a weighted dh linear regression is applied to calculate dt for each pixel. 36 ASTER DEMs (derived 10 from imagery acquired from 2001 to 2011) and the SRTM DEM (acquired in February Title Page 2000) are processed, with an average of 4–5 elevations per pixel incorporated into Abstract Introduction the regression. Each elevation is weighted by the inverse of the standard deviation of the bedrock elevation differences between its ASTER DEM and the SRTM DEM.