You may recall the guest post from Jeff Id of the Air Vent I carried about a week ago called Global Sea Ice Trend Since 1979 – surprising
In that post, a note of correction was issued because that we were led to believe (by Tamino) that the entire post was “invalidated” due to an error in accounting for ice area very near the pole. Both Jeff and I were roundly criticized for “not reading the documentation”, which was one of the more civil criticisms over there at Tamino’s site.
After further investigation It turns out that the error was in NSIDC’s public documentation, and they have issued a correction to it. Even more importantly the correction now affects NSIDC’s own trend graph, and they are considering how to handle it.
This episode illustrates how citizen science can be useful. Sometimes people too close to the science they publish can make mistakes, (we’ve all been there) which is why peer review of papers is important. But “web review” in this day and age of instant publication is equally important. It also illustrates how mistakes, however embarrassing initially, can be useful if you learn from them and study the cause. There is no shame in mistakes if they are corrected and you learn from them. But, the blogospheric noise of angry and sometimes juvenile criticism (on both sides) really isn’t useful as it often masks the real issue. The key is to put that aside and find the truth behind the error. Jeff has done that. His update follows below.
Merry Christmas to everyone! – Anthony
Based on The Air Vent post carried by Watts Up With That, the National Snow Ice Data Center has issued several corrections to the documentation of their sea ice area time series.
Guest post by Jeff ID
Most will remember my earlier post which plotted global sea ice trends. After initially concluding that the global ice level wasn’t decreasing measurably Tamino pointed out a problem in my analysis. After issuing my corrections, thanks and apologies to Tamino and the um…..thousands of readers of Watts Up With That, I went back to work investigating what was really happening to the ice area time series.
It was actually quite lucky that Tamino mentioned the step in the data and criticized me for not reading carefully (something which was mentioned in several comments on the various threads). When I first learned of it, I found the criticism was based on an entirely different set of ice area data with different source documentation. Still, I checked closely and found the tiny step in the time series and was convinced that I had missed something. I had spent a huge amount of time learning the data before I made my post so it was frustrating to say the least. Understand, I used several resources to check my work; not the least of which was the National Snow Ice Data Center (NSIDC) anomaly graph which has the same shape as the one I generated.
The first graph below is from the NSIDC website, the second is my calc. Differences in the noise between the two are explained by the daily resolution used in my graph compared to what my eyes tell me must be monthly data for their plot. They also seem to have an additional year (2007) in their data plot which is not available in the bootstrap time series I used.
After reading everything with great care, this graph and a similar one for the SH were used to verify my results before my original post. This paragraph below used to be on the NSIDC website describing the data of these time series.
In computing the total ice-covered area and ice extent, pixels must have an ice concentration of 15 percent or greater to be included; thus, total ice-covered area is defined as the area of each pixel with at least 15 percent ice concentration multiplied by the ice fraction in the pixel (0.15-1.00). Total ice extent is computed by summing the number of pixels with at least 15 percent ice concentration multiplied by the area per pixel. Sea ice concentrations are assumed to be 100 percent around a circular sector centered over the Northern Hemisphere pole (known as the pole hole) which is never measured due to orbit inclination. The Southern Hemisphere also has a pole hole; however, it does not affect this sea ice data set; since only land is under this hole. For SMMR, the hole is 611 km in radius and is located poleward of 84.5 degrees north. For SSM/I, the hole is 310 km in radius and is located poleward of 87 degrees north.
After checking this for about the hundredth time along with the rest of the extensive documentation, I wrote to the NSIDC and asked them to confirm that the area anomaly for the NH wasn’t corrected another way. Several emails back and forth later they confirmed that the area anomaly wasn’t accounted for. I then pointed out that the graph above and the paragraph in the data description were in error. After a short time, the NSIDC replied that they had their sea ice team was reviewing the data and planning an immediate update to their site.
That same day the site was corrected to read:
In computing the total ice-covered area and ice extent with both the NASA Team and Bootstrap Algorithms, pixels must have an ice concentration of 15 percent or greater to be included. Total ice extent is computed by summing the number of pixels with at least 15 percent ice concentration multiplied by the area per pixel, thus the entire area of any pixel with at least 15 percent ice concentration is considered to contribute to the total ice extent. Total ice-covered area is defined as the area of each pixel with at least 15 percent ice concentration multiplied by the ice fraction in the pixel (0.15-1.00). There is a circular section over the Northern Hemisphere pole (known as the pole hole) which is never measured due to orbit inclination. For the purposes of ice extent, pixels under the pole hole are always considered to be at least 15 percent. For total ice-covered area, the pixels under the pole hole are not used. The Southern Hemisphere also has a pole hole. However, it does not affect this sea ice data set because there is only land under this hole. For SMMR, the hole is 611 km in radius and is located poleward of 84.5 degrees north. For SSM/I, the hole is 310 km in radius and is located poleward of 87 degrees north. Note: The difference in pole hole areas between SMMR and SSM/I results in a discontinuity in the Northern Hemisphere ice-covered area time series across the instrument transition.
Link HERE. They are still considering how to handle the area anomaly graph.
Since this changes how you interpret area data substantially, there is no easy method for updating the trend graph. Still, the step in the data is quite small as shown below.
It occurs right after 1987.5 which corresponds to the 87 July/August boundary which is different from tamino’s reference. If we assume worst case that the NH hole in the data was 100% filled with ice (it wasn’t), the calculation from before produces a slight downslope in comparison to the flat trendless line in my original post. The result is only a trend equaling a 4% reduction in global sea ice over a nearly 30 year period. Not exactly disastrous either way. I am going to continue my work on this by matching (regressing) the last two years from other sites on the end of the data. With the recent global cooling, it should be interesting to see where global sea ice is today.
I need to offer thanks to Anthony Watts for putting the original post on his blog. His professionalism was commendable in handling this matter quickly and transparently. IMO this openness to correction is lacking on several AGW blogs. I also need to thank the NSIDC (particularly, Dave, Molly and the Sea Ice Team) who really blew me away with their responsiveness and professional demeanor in making these corrections.