Estimation of ice thickness on large northern lakes from AMSR-E brightness temperature measurements

• Published in: Remote sensing of environment 2014-07, Vol.150, p.1-19
• Main Authors: Kang, K.-K., Duguay, C.R., Lemmetyinen, J., Gel, Y.
• Format: Article
• Language: English
• Subjects: Algorithms; AMSR-E; Animal, plant and microbial ecology; Applied geophysics; Arctic; Biological and medical sciences; Brightness temperature; Canada; Earth sciences; Earth, ocean, space; Error analysis; Exact sciences and technology; Freshwater; Fundamental and applied biological sciences. Psychology; General aspects. Techniques; Great Bear Lake; Great Slave Lake; Ice cover; Ice thickness; Internal geophysics; Lakes; Mathematical analysis; Mathematical models; Retrieval; Seasons; Teledetection and vegetation maps
• ISSN: 0034-4257; 1879-0704
• DOI: 10.1016/j.rse.2014.04.016

An ice thickness retrieval algorithm utilizing data from the Advanced Microwave Scanning Radiometer-Earth Observing System (AMSR-E) was developed and applied to Great Bear Lake (GBL) and Great Slave Lake (GSL), Northwest Territories, Canada, for the period 2002–2009. The temporal evolution of vertically polarized AMSR-E brightness temperature (TB) at 18.7GHz was explored to estimate ice thickness between the late freeze-up (ice-on) and early break-up (melt-onset) periods. The sensitivity of AMSR-E TB at H and V polarizations to the seasonal evolution of ice thickness was examined statistically and with forward simulations of TB using the most recent version of the Helsinki University of Technology (HUT) model, which incorporates a freshwater ice layer. The strong relation found between TB at 18.7GHz V-pol and ice thickness was exploited for the development of a regression-based ice thickness retrieval algorithm. Simple linear regression equations allow for the estimation of ice thickness on a monthly basis from January to April; one, the Global equation, combines TB data for GBL and GSL, and two others, the Regional equations, use TB data for each lake alone. Estimated late-winter ice thicknesses on GBL were determined to be on average 5–10cm thicker than on GSL with the exception of ice season 2005–2006 when it was estimated to be 10cm thicker on GSL. For both lakes the 2004–2005 and 2008–2009 ice seasons experienced the thickest end-of-winter ice thicknesses, ranging from 130 to 134cm on GBL and GSL. The thinnest end-of-winter ice thicknesses were on average 120cm on GBL (2005–2006) and 123cm on GSL (2007–2008). Variability in air temperature, snowfall and subsequent redistribution by wind, and lake depth explain ice thickness variations within and between lakes over the seven winter seasons analyzed. Estimated ice thicknesses from AMSR-E compare well with coincident in situ measurements collected on GBL and GSL over a limited number of ice seasons and sites within the large passive microwave footprints (Mean Bias Error, MBE=6cm; Root Mean Square Error, RMSE=19cm). •Simulations show sensitivity of 18.7GHz brightness temperature to ice thickness.•AMSR-E 18.7GHz (V-pol) is used to estimate ice thickness on two large lakes.•The root-mean-square-error in estimated ice thickness is ~18cm.•Linear regression equations are used to generate monthly lake ice thickness maps.