Readers may recall that I brought an issue with NSIDC’s sea ice data to their attention back in February 09. The official response from Walt Meier then was “its not worth blogging about”. Shortly afterwards that tune changed when it was announced:
Not only was the glitch I pointed out part of the sensor failure real, there was also a long term drift in the sensor readings.
Cryosphere Today followed soon afterwards, pulling the plug on their data plot.
This image below is what prompted me to bug NSIDC about the issue, the downspike was highly atypical of the dataset.
The glitch seen above turned out to be a “catastrophic failure” of the SSMI sensor on the Defense Meteorological Program (DMSP) satellite that NSIDC (and many others, including Cryosphere Today) also use. Here’s a look at the problem in more detail that WUWT covered in June.
So today, I was advised in an email that I should have a look at the data from Danish Meteorological Institute (DMI), which also uses DMSP SSMI data to make their web graphic presentation. To quote Yogi Berra, it was “deja vu all over again”. Here’s the DMI image from today:
Note the downward glitch in the black 2009 line. Source: http://ocean.dmi.dk/arctic/icecover.uk.php
Now compare that to the AMSRE sensor data (on the AQUA satellite) from JAXA:
The sea ice extent value from JAXA is 6.5 million km2
The sea ice extent value from DMI is pf course much lower, around 4.4 million km2
That is due to DMI choosing to plot “where areas with ice concentration higher than 30% are classified as ice”, while NSIDC, CT, and JAXA plot an area with at least 15 % sea ice. So the absolute difference isn’t what concerns me, it’s that darn glitch again.
NSIDC and CT have taken steps to filter out such things in their own records. DMI obviously hasn’t. The fact that such abrupt downward glitches are still observable in the DMSP SSMI data makes me concerned for the integrity of the dataset. In May 2009, NANSEN also had problems with SSMI data.
So the biggest names in sea ice data on the web, NSIDC, Cryrosphere Today, NANSEN, and now DMI, all of which use SSMI data, have had problems with it. The IARC-JAXA data has held steady.
This is why I prefer using the JAXA AMSRE data from the AQUA satellite to look at current and recent history sea ice events. The newer AMSRE sensor has not shown any such errors, drifts or glitches during its history.
It would seem to me that the time has come for these agencies to consider switching to the more reliable AMSRE/AQUA platform and splicing the older SSMI data set with appropriate considerations to make an accurate longer term record of sea ice since 1979.