Assessing Disagreement in Two Long Records of Melt Onset on Arctic Sea Ice Derived from Passive Microwave Satellite Observations

A comparison of two long records of annual melt onset (MO) on Arctic sea ice from passive microwave satellite observations since 1979 reveals significant disagreement in trends and regional timing of MO. Inconsistencies between the Advanced Horizontal Range Algorithm (AHRA) and the Passive Microwave (PMW) algorithm are examined to improve confidence in annual estimates of MO timing and trends over the satellite record. Both methods detect melting on Arctic sea ice using brightness temperature observations (Tbs) from the Scanning Multichannel Microwave Radiometer (SMMR) onboard NASA’s Nimbus-7 and the Defense Meteorological Satellite Program’s series of Special Sensor Microwave Imagers (SSM/I) and sounders (SSMIS). The AHRA and PMW algorithms apply thresholds to the daily time series of Tbs to detect characteristic changes associated with melting. The algorithms return the day of year that liquid water first develops in the snow or on the sea ice surface giving the earliest indication that the melting season has begun. MO of the snow and sea ice coincides with moderate surface albedo decreases prior to more advanced melting stages including melt pond formation that occur later in the melt season. Reduced albedos from MO early in the melt season increase absorption of solar radiation throughout the melt season, enhancing the ice-albedo feedback loop.

Statistically significant disagreement in the long term trends of annual mean MO dates from the AHRA and PMW exist for most of the Arctic Ocean. Differences in regional mean MO dates determined by the AHRA and PMW have high regional variability resulting in differences as large as 38 days in the Barents Sea and as low as 1 day in the central Arctic. Causes for these primary differences in the time series of MO dates from the AHRA and PMW are investigated. Sensitivity tests of the inter-sensor calibration applied to the Tb time series indicate that consistent application of calibration corrections improved agreement of the trends in MO date for most of the Arctic Ocean not apparent in the original MO data. Large offsets in the mean MO date still exist in the marginal ice regions where sea ice concentrations are lower and dynamic ice motion increases uncertainty in the detection of MO. In the marginal ice regions, the AHRA consistently biases toward an earlier MO date than reported by the PMW which often coincides with retreat of the sea ice edge. Spatial disagreement in the MO dates detected within the Arctic Ocean are examined in conjunction with atmospheric variables from NASA’s Modern-Era Retrospective analysis for Research and Applications, Version 2 (MERRA-2) reanalysis product to examine sensitivity of the algorithm thresholding methods on the Tb time series in the presence of atmospheric melt forcing conditions.