NSIDC s Passive Microwave Sensor Transition for Polar Data

NSIDC’s Passive Microwave Sensor Transition for Polar Data Donna J. Scott NASA NSIDC DAAC Passive Microwave Team with Walt Meier Goddard Space Flight Center

Background – PM sensor record Sensors: Passive microwave sensors provide continuous and near-complete record from 1978 -present Timeline of passive microwave sensors 1970 1980 NIMBUS-5 SMMR 1990 2000 2020 2010 NIMBUS-7 • • Passive Microwave Era F 8 F 11 • • F 13 • F 15 F 16 DMSP Chinese sensors are in orbit, but access to data is not realistic F 17 F 18 F 19 F 20 NASA EOS AQUA No other passive microwave sensors are on the horizon until after 2020 → the nearly 40 -year record of continuous, consistent monitoring will end if current capabilities do not last significantly longer than planned AMSR-E JAXA GCOM-W AMSR 3, 4 • • • 2 F 15 - launched 1999 (>16 years old) F 16 - launched 2003 (>12 years old) F 18 - launched 2009 (>6 years old) AMSR 2 - JAXA sensor launched 2012 (Only AMSR 2 is operating within its nominal 5 year mission) F 19 launched 2014 (failed February 2016) F 17 launched 2006 (capability for sea ice ended April 2016) F 20 launch was cancelled (in storage) Additional Impacts include: Sea ice concentration/extent Snow extent Timing of melt onset over snow & ice Rainfall, total precipitable water, cloud liquid water Soil moisture Ocean wind speed

Background – PM Product Team Non-EOS passive microwave data (SMMR-SSM/I-SSMIS) In-house data production LEAD USO Tech Writer OPS Scientist Sci Developer 3

Timeline of recent satellite transition work • F-19 NRT parallel processing stream set up Nov 2015 • F-17 begins showing bad data • F-18, F-16 parallel processing streams set up • Provisional F-18 data released Apr 2016 Feb 2016 • F-19 fails prior to calibration • F-18, F-16 discussed as options 4 • F-16 calibration work complete, data running internally Jun 2016 • NISE F-16 calibrations in work Aug 2016 Sept 2016 Jul 2016 May 2016 • Wrap up NRT • NISE F-18 • F-18 calibration satellite transition calibrations complete, completed, data running internally • Look to RSS for F-18 officially released data release for time • F-17 fixes made, data series data transition running internally

Considerations in priority setting • • Brightness temperature data vs. sea ice data Internal NSIDC sea ice data needs PM § NOAA@NSIDC SII § NASA funded ASINA o 2. 5 million views in 2015 • NASA missions needs ASINA SII § Missions use NISE as input. These missions not reporting the impacts in F-17 Remaining considerations • Near-real-time and time-series data no longer consistent with platform sensor § F 18 not ready for long time series – RSS does not have F 18 online yet. • 5 New approach to provenance and stewardship practices related to NSIDC near-real-time data sets

Prioritizing transition of DAAC PM data sets 6

F-17: What happened? • • 4/5/16: Solar panel position change compromised integrity of primary sea ice algorithm channel 16 (37 Ghz vertical polarization) 4/13/16: Solar panel repositioned improving channel 16 problems. Sea ice data at NSIDC still showing problems 5/25/16: Reports of optimal sensor function after fix to spike detection algorithm. Long term quality unknown and data still at risk 5/26/16: NSIDC decision to move to F-18 Bad data Bad Data Images April 28, 2016 TBs and Ice Extent 7

Science • Scott Stewart – NSIDC PM science contractor • Julienne Stroeve – NSIDC • Walt Meier - GSFC 8 NSIDC: An overview

Sea ice product transition • Sea ice concentration derived from brightness temperature using empirical relationships • Coefficients (tiepoints) assigned for pure surface types (water, two ice types) • Intercalibration approach is to adjust tiepoints to match extent/area between sensors during an overlap period • Weather filters, thresholds of TB ratios, used to reduce false ice retrievals over open water § Effectively help define the ice edge 9 NSIDC: An overview

Calibration Effort for F-18 § During the calibration efforts, a provisional F-18 data stream using F -17 tie points was used to enable ASINA to continue reporting sea ice trends § NSIDC investigated the calibration of algorithm tie point values to best match the sea ice extent from F-17 over a 12 -month period from 03/01/2015 through 02/29/2016. § Current F-17 tie points provided the best match in sea ice extent, so no adjustment to the tie points were made for F-18. The average difference between F-17 - and F 18 -derived sea ice extents were approximately 20, 000 sq km. 10

F 17 to F 18 SSMIS sea ice calibration • NSIDC investigators found that brightness temperature values for F 18 and F 17 were very similar for channels related to sea ice concentration. Brightness temperatures are highly correlated (R = 0. 996) 11 Histograms are indistinguishable

F 17 to F 18 SSMIS sea ice calibration • • Attempts to adjust the NASA Team tie points led to worse correlation of extent values, so the values used for F 17 continue to be used. This is likely due to the sensitivity of ice edge to small TB changes, including effects of weather filter thresholds Differences in extent are small (Average daily difference < 10, 000 km 2) – is this true; looking at the next plot and from the numbers I have the mean difference is ~20 k

F 16 and F 18 extent differences 13 NSIDC: An overview

Science Behind Satellite Transition Issues matching sea ice area and extent § Simply regressing TBs does produce as close of a match is possible Daily average of swaths o Difference in observation times and number of observations at a location o § Cannot optimize both area and extent – i. e. , best area match possible will produce an extent difference that is not minimal § Extent is sensitive to weather filters – can potentially adjust thresholds to better match extent without denigrating area differences too much § Seasonal affects also – summer melt season is problematic 14

Future activities • Goddard will produce a final F 18 intercalibration using RSS TBs • There is much potential to improve intercalibration of previous sensor products if/when resources become available § § 15 New version of RSS TBs Longer sensor overlap periods Use of AMSR-E or AMSR 2 as a baseline Use swath data instead of daily averages? NSIDC: An overview