Something I’ve been interested in for the last several months is sea ice data. What makes it interesting is that as I understand it, models demonstrate the poles should be most sensitive to global warming leading the planet temp, especially in the Arctic. Recently I have been able to process the monthly and daily gridded arctic data as provided by NSIDC. The daily values allow a better analysis of trend than can be provided by the monthly data.
If you’re like me you recall the claims of fastest melt rate ever were made about 2007 , I fully believed them, because the graphs showed a much more negative value than in the previous 30 years as shown in Figure 1 below.
This effort was originally intended to investigate how bad the melt rate was in comparison to the naural variation, I didn’t get that far yet. Accessing and processing the gridded data was critical to the analysis, so I spent the time reading the literature and writing code. Having full access to the NSIDC data allows some interesting analysis, they do an excellent job on their site.
There are two primary algorithms used for processing ice data NasaTeam and Bootstrap. The descriptions of the data state the difference between the two is very small and the sets are interchangeable except that bootstrap is recommended for trend analysis in research publications. Bootstrap is only provided in monthly data format while nasateam is provided in both monthly and daily provided you’re willing to download over 1G of data, write code to process it, refit the land and missing data mask and sum the results. I am. Also, Nasateam provides a near real time version of the polar ice data which has a different land mask and hasn’t been processed for missing data. This data isn’t as clean but I wanted to use it. I applied the same land mask as the rest of the series to insure that there was a consistent baseline for trend analysis. The missing data from Jan 2008 onward created noise in the series which I simply filtered out using a 7 day sliding window filter.
The mask looks like this Figure 2
The brown is land, black edges on land are coastline and light blue is the satellite data not measured. This mask is applied consistently through the entire data series. There was some question about masking on one of my other posts at WUWT where visually the land area seemed to change size, in the case of the NSIDC data they apply masks consistently except for the satellite hole and the near real time data.
The Nasa Team version of the arctic ice data looks like the plot below for Feb 2009 (note the small size of the satellite data hole). This graph was created in R using the actual NasaTeam masks and data. I used the worst case land and polar masks to adjust the entire dataset to eliminate problems with consistency. Figure 3
Of course it’s an interesting picture, but what I wanted to know when I started this post was how bad was the worst melt rate in history and what is the actual melt area. In the plot below the arctic is loosing sea ice at a rate of only 56K km^2/year. Of course sea ice area went up in the antarctic during the same time frame though. Note the strong recovery in 08 of Figures 1 and 4, which actually exceeds values of most of the record, matching data back to 1980. Much of this is first year ice so the melt in 08 was expected to be a new record.
If you recall, in 2007 and 08 we were treated to headlines like this, which most of us accepted with a shrug.
Scientists warn Arctic sea ice is melting at its fastest rate since records began
NASA data show Arctic saw fastest sea ice melt in August 2008
Arctic Just Witnessed Fastest August Ice Retreat in History
I processed and analyzed the nasateam land area and missing data masks spending hours understanding different variances they list on their own website. After nearly everything I could find (except satellite transitions errors) was corrected (a different post) and corrections for variance in the measured pixel size, the final result in 30 day trends of arctic sea ice looks like the graph below (Figure 5). This graph is a derivative of the ice area plot. The maximum peaks and valleys represent the maximum rates of change in 30 day periods through the ice record.
Looking at this plot of the 30 day slopes of actual nasa gridded data, the maximum ice melt rate occurs in 1999 and in 2004 not in 2007. Surprisingly the maximum ice growth rates occur in 2007 and 2008, I don’t remember those headlines for some reason. Don’t forget when looking at the 2008 – 09 peak, the data is preliminary and hasn’t been through the same processing as the other data. From looking at the unprocessed data I doubt it will change much.
Certainly the 30 year arctic trend in ice area is downward, even the most committed global warming scientist has to admit this happens regularly in climate along with regular 30 year uptrends. The questions are, did we cause it or not, and was CO2 the instigating factor. The rapid recovery of ice levels has to have some meaning regarding the severity of the problem. This goes directly in the face of accellerated global warming and the doom and gloom scenarios promoted by our politicians and polyscienticians.
Why are my conclusions different from the news reported records? I think it’s likely due to the fact that the scientists used the monthly data which is processed using a weighted filter of the daily data that incorporates a longer time frame than a single month. This means their use of the monthly data to establish a monthly trend was in error and the real record down trends were actually set in 1999, 2003 and 1984. While the record uptrends were in 2007, 2008 and 1996.
22 thoughts on “Arctic Sea Ice Increases at Record Rate”
Thanks Jeff, it’s great to see an original analysis that starts from scratch with the raw data. Awesome work. I’d love to see a headline like “Record Arctic Sea Ice Increase” in the mainstream press, but alas I will have to settle for seeing such only in the blogosphere.
Any chance you could publish your data sources and R code so that others can play around with this?
One thing, the extrema of the rate of change are probably not all that important, as they are attained for only a short period of time. Certainly the absolute height of the peak correlates with the area under the peak, but as you can see, when comparing it to total sea ice, the correlation is not terribly strong.
Great work Jeff. Thanks for this. Your new freinds at NSIDC would be interested in this (perhaps they know already). BTW, on my screen, your graph is cut off at about 2002.
You have uncovered an important frame of reference for understanding arctic sea ice in the context of climate trends. The rate of gain/loss would seem to be a more accurate indicator of underlying trends. It shows that the process of ice recovery began earlier than portrayed. Time will tell whether this recovery will be sustained. I think that this needs to be communicated to a wide audience. Have you talked to Anthony about posting there?
I find it hard to believe that those who study these numbers did not understand this already. Perhaps they did not. The recovery of ice extent in 08 goes a little deeper than just a “blip”, yet the story is still spun of tipping points, etc.
I consider myself to be open minded on AGW. I still do. I am thankful for your analysis but frustrated that this hasn’t been pointed out by those people we should be able to count on for accurate perspective.
Shoot me an email if you can. I’m getting an error when attempting to access RC and want to compare it to your experience to see if I too am banned from the site.
I saw an even worse cutoff of the graphs, then I hit CTRL- a few times to reduce the font size (Firefox) and the entire graph magically appears. My guess is that there is a way to format the page to prevent the graph from changing size when the font size is changed but I know only the barest minimum about HTML coding so I don’t know for sure.
Jeff, if you could add your learning on the center hole, i think i read that here,
and add a trend line to both grafts. maybe approach Anthony and post it on his blog also. I will e-mail this link to him also.
TX for you time.
BTW, on my screen, your graph is cut off at about 2002.
Firefox, full frame and it is hole. lol
DeWitt, are you saying I would need the Firefox browser otherwise I’m hooped?
Good work. Might be interesting to plot the change rate with the x-axis total duration of one year and each yearly period plotted as an individual, overlayed trace for the last ten years or so. It is hard to tell from the plot above the duration of the high change rate periods.
You alluded to this above, but from eye-balling the plot it appears that high rates of negative change are followed by high rates of positive change. That would seem to point to a localized winter phenomena as opposed to generally higher temperatures throughout the year.
Sorry about the graphs, I finally figured out how to get wordpress to thumbnail thanks to Digital Diatribes. It used to work for me but there was a missing link that I didn’t notice on the image page. See if it works better now, my future posts will take advantage of it.
#8 It’s more of a commentary on the polyscienticians take on fastest melt rate ever. Looking at the raw data, august 07 didn’t quite make the cut. With various levels of filtering (such as in monthly data) you might get it to make the cut but in reality the fastest melt rate seems to have occurred in 04 or 99. What we did get for certain is the fastest ice growth rate in 07 or 08.
There is a major problem here. Big Al has stated that the Arctic will be ice free by 2013. Your graph doesn’t appear to demonstrate this.
#10 – Turn it upside down. 😉
Firefox just makes it convenient to change font size. The problem is that either you weren’t running your browser at full screen size or your monitor has insufficient resolution to display the full page.
Have you tried calculating the annual change on a daily basis? It removes the seasonal variation and might reveal interesting features.
This is the code I used for generating trends in sea ice. You’ll have to fix all the quotes ’cause wordpress messes them up. In order to use the code you need to download the gridded daily nasateam data from NSIDC. To get the near real time data copy it into a directory after the other annual directory data which consists of a pile of directories/folders which say 1979, 1980 ,1981, ….etc. Directories should be in numerical order.
filenames=list.files(path=”C:/agw/sea ice/north sea ice/nasateam daily/”, pattern = NULL, all.files = TRUE, full.names = FALSE, recursive = TRUE)
for(i in 1:(length(filenames)-1))
fn=paste(“C:/agw/sea ice/north sea ice/nasateam daily/”,filenames[i],sep=””)
datamask=data225 & data < 251)
plot(date,masktrend,type=”l”,main=”Mask Area Daily Arctic Sea Ice NasaTeam\nLand Area + Measurement Hole – Mask Noise”,xlab=”Year”,ylab=”Kilomters^2″,xlim=c(1995,1997))
#savePlot(paste(“c:/agw/sea ice/Nasateam mask trend noise 1995 1997.jpg”),type=”jpg”)
I assume the frowny face is a special R code 😉
The rapid recovery of ice levels has to have some meaning regarding the severity of the problem. This is totally consistent with accellerated global warming and the doom and gloom scenarios promoted by our politicians and polyscienticians. Rapid recovery of ice levels corresponds to larger amounts of 1st year ice in the region, which grows and melts faster than older ice.
I’m reminded of the hysteria over the ozone hole a few years back. It only took my wife changing hair spray brands to stop this phenomenon. I suspect now that Barrack O Bama has been elected that global warming will take the same path
FoolFighters If what you say were true then we should see an increase in the melting side of things as well. But the Max melt rate was back in 1999. 30 years of data just can not tell us very much, but I am betting in the next 30 years we will see that there is a natural cycle and polar bears will still be with us.
I think FoolFighters is right. You might want to have a look at this picture . Time will point out whether man is causing this decline of sea ice. For now we only know the total amount of human emission of greenhouse gasses in comparison to natures ‘own emissions’. But there is still a lot unknown about the effects of the large amounts of greenhouse gas emissions of man.
Foolfighters was right as we all can see by now.
Don’t be silly! The graph clearly shows that the arctic ice is shrinking. Greenpeace said so. Oh, wait…
Terima Kasih atas infonya..