Extension of sea ice climate time series with historical satellite data

Sea ice plays a critical role in climate change with strong effects on the planet’s albedo, ocean and atmospheric circulations. It acts as an isolator between the cold air above and the relatively warmer water below the ice, preventing the air from further heating from below. The melting of sea ice gives rise to an ice-albedo feedback, where a decrease in ice cover with high albedo leads to an increase of open water surfaces with smaller albedo, resulting in more absorption of solar radiation, which amplifies the warming of the ocean and thereby also melting of more sea ice. Arctic temperatures have warmed more than four times as much as global temperatures through the past 40 years, a phenomenon known as arctic amplification. Sea ice is therefore a significant climate change indicator.

It is important to create long and consistent climate data records (CDRs), to assess the development of sea ice better through the past, present and future. Long CDRs are necessary to determine climate trends, variability and extremes, to put current changes into perspective and provide better predictions for the future. In this PhD study, several different satellite-borne instruments have been investigated and used for computing sea ice parameters, such as sea ice concentration (SIC), sea ice type and sea, snow and ice surface temperatures, as well as related uncertainties. Algorithms have been developed to create and extend CDRs, where modern methods have been applied to older satellite data. All data sets and instruments face individual but also common challenges, especially when trying to create a consistent data record, which are discussed in this project.

Three data sets are presented in this work, starting with the SIC data set covering 1972-1977 based on the Electrical Scanning Microwave Radiometer (ESMR) of the NIMBUS-5 satellite. This ESMR was a predecessor to the multi-frequency sensors of modern CDRs based on SMMR and its successors. Compared to newer instruments, the ESMR only measured brightness temperatures at a single frequency (19.35 GHz), for 39 different incidence angles. Modern processing procedures included an atmospheric correction with a radiative transfer model (RTM) with ERA-5 atmospheric parameters, the use of dynamical tie-points to reduce biases of the atmospheric data, seasonal variability and instrument drift.

The second data set includes both SIC and sea ice type for 1975/1976 and is based on the Scanning Microwave Spectrometer (SCAMS) of the NIMBUS-6 satellite. This atmospheric sounder’s main purpose was to measure tropospheric temperature profiles and has therefore not been used previously to map sea ice. With its measured brightness temperatures at 21 and 32 GHz it was possible to distinguish sea ice types and improve especially the SIC estimations of the multi-year ice, compared to single-frequency SIC. Additionally, SCAMS does close one of the largest data gaps in the ESMR data set.

The last data set presented is the C3S IST CDR/ICDR, covering 42 years of ice surface temperatures (IST) for 1982-2023, based on NOAA and MetOp satellite borne Advanced Very High Resolution Radiometer (AVHRR) GAC data. In addition to RTM atmospheric corrections and uncertainty estimations, these thermal infrared data also need to be filtered for clouds. Additionally, to create a coherent data set through time and space, the algorithms have been tuned for each instrument and surface type, i.e. the data set covers sea and land ice, but also open water surfaces including the marginal ice zone.