2010 Spring Arctic Sea Ice Extent

The decline in the sea ice extent in May and June of 2010 appeared to be extremely fast. According to NSIDC,

Arctic sea ice extent averaged 13.10 million square kilometers (5.06 million square miles) for the month of May, 500,000 square kilometers (193,000 square miles) below the 1979 to 2000 average. The rate of ice extent decline for the month was -68,000 kilometers (-26,000 square miles) per day, almost 50% more than the average rate of -46,000 kilometers (18,000 square miles) per day. This rate of loss is the highest for the month of May during the satellite record.

However, later on the same page, they also state under Conditions in Context:

As we noted in our May post, several regions of the Arctic experienced a late-season spurt in ice growth. As a result, ice extent reached its seasonal maximum much later than average, and in turn the melt season began almost a month later than average. As ice began to decline in April, the rate was close to the average for that time of year.

In sharp contrast, ice extent declined rapidly during the month of May. Much of the ice loss occurred in the Bering Sea and the Sea of Okhotsk, indicating that the ice in these areas was thin and susceptible to melt. Many polynyas, areas of open water in the ice pack, opened up in the regions north of Alaska, in the Canadian Arctic Islands, and in the Kara and Barents and Laptev seas.

This latter observation that the seasonal maximum was reached later in the season and the melt season started later is important. Regardless of specific annual weather conditions, May and June are melt season months in the Arctic. Furthermore, if there is more ice available, then it stands to reason that more melting will take place. What might a better way to look at the data than simply plotting the total extent?

From the JAXA site:

Why not graph the rate of change, as well? In particular, because a wider extent will naturally imply a higher areal melt under the same melting conditions, it makes sense to look at the daily percentage change.

To do this, I downloaded the JAXA daily ice data into R (from 2002 to the present). For convenience purposes, December 31 was deleted from both 2004 and 2008 to reduce the number of days to 365. The percentage change was calculated for each day for which the corresponding data was available. No infilling was done for missing data. The data was plotted:

(Click graph for larger version)

Here, all of the years prior to 2010 are plotted in gray and the current year in red. The plot gives graphic insight into the patterns of thawing and freezing: the thaw season goes from roughly mid-March to mid-September. The very high variability in October is likely due to a reasonably similar annual speed of recovery which is expressed as a percentage of quite varied minima starting points in September.

How does 2010 compare in May and June? For May, it is somewhat toward the lower part the combined record, but I would not classify it as extreme in any way. June was definitely below the other recent years during three periods of several days each. What will July and August look like? I guess we will have to wait and see…

The R script follows:

#get latest JAXA extent data

iceurl = url("http://www.ijis.iarc.uaf.edu/seaice/extent/plot.csv")
latest = read.csv(iceurl,header=F,na.strings="-9999")
colnames(latest) = c("month","day","year","ext")

#remove Dec 31, 2004 and 2008 (extra leap year day) for convenience
#fill in with missing values for early part of 2002 (for convenience)

arc.ext = latest$ext
which((latest$month==12)&(latest$day==31)) # 214 579 945 1310 1675 2040 2406 2771 3136
arc.ext = arc.ext[-c(945,2406)]
arc.ext = c(rep(NA,365-214),arc.ext)

#length(arc.ext)/365 # 9

#calculate changes as % of current value
#form matrix with 9 columns (one for each year)

pct.change = matrix(100*c(diff(arc.ext),NA)/arc.ext,ncol=9)

#plot data
#years 2002 to 2009 as gray background
#year 2010 in red
#add month boundaries
modays = c(31,28,31,30,31,30,31,31,30,31,30,31)

matplot(pct.change[,1:9],type="l",main ="Arctic ice Extent Change Relative to Area",xlab="Day",
ylab="Daily % Change", col=c(rep("grey",8),"red"),lty=1)
abline(v=c(0,cumsum(modays)), col="green")
text(x =14+c(0,cumsum(modays)[-12]),y =c(rep(3,9),rep(-1,3)), labels=month.abb,col="blue")

About these ads


Filed under Uncategorized

5 responses to “2010 Spring Arctic Sea Ice Extent

  1. I read this post about a day after you put it up. It’s good that you’re back at it.

    I see your point that a daily percent change isn’t that far out of whack but of course the integral of a slight difference can add up. Imagine the difference in interpretation if the data was by second or by week.

    I wrote one on sea ice from the latest NSIDC post. It was supposed to be a boring update on sea ice, because globally it isn’t really changing. The NSIDC proclaimed record losses and loss rates in May and June and covered up the the record highs in the June Antarctic.

    • RomanM

      I read your post and agree with you on the blatant bias of the NSIDC site. It is a shame that these guys can’t present a more reasoned and balanced view in their discussions. But hey, it is a vehicle which tends to “convert” thinking believers toward skepticism.

      My point was that it was misleading to look purely at the change in the amount of ice extent.

      When there is a greater amount at the beginning of a melt season or if conditions are less conducive for early melting, then it stands to reason that the melt rate will be faster later simply because there is a broader area over which the melt can take place. To put this into context, the melt should be expressed in relation to the total available.

      If one expects the effect to take place uniformly throughout the ice region, then the proportion (or percentage) is a reasonable statistic to look at. However, if the change occurs mainly at the perimeter, then using the square root of the extent would be better.

      For example, if the ice was a solid mass and the reduction was due to conditions at the edge of the sheet, say due to water melting or breakoff, then the latter would be the case. On the other hand, if the ice is disconnected floes and the cause for melting is solar radiation and/or ambient temperatures then I would expect the percentage change to be more appropriate for the evaluation. I chose the latter in this case.

  2. Tony Hansen

    Ahh, ’tis a fine thing that you are back Roman.
    Your absence was rather more noticable than we expected….. and we had expected ‘twould be noticable.
    Did you enjoy your golfing interlude?

  3. Tony Hansen

    I was just wondering what effects precipitation may have in various regions at this time of the year.
    ie. when is rain liquid…and if/when it is so, is its effect different on the edge of the ice pack compared to where the ice is more solid?

  4. Ruhroh

    Dear Sir;
    Am curious about your ‘statistical opinion’ of the method described by Briffa in Tranche II email 3436;
    “we are having trouble to express the real message of the reconstructions – being
    scientifically sound in representing uncertainty , while still getting the crux of the
    information across clearly. It is not right to ignore uncertainty, but expressing this
    merely in an arbitrary way (and as a total range as before) allows the uncertainty to swamp
    the magnitude of the changes through time . We have settled on this version (attached) of
    the Figure which we hoe you will agree gets the message over but with the rigor required
    for such an important document.
    We have added a box to show the “probability surface” for the most likely estimate of past
    temperatures based on all published data. By overlapping all reconstructions and giving a
    score of 2 to all areas within the 1 standard error range of the estimates for each
    reconstruction , and a score of 1 for the area between 1 and 2 standard errors, you build
    up a composite picture of the most likely or “concensus” path that temperatures took over
    the last 1200 years (note – now with a linear time axis). This still shows the outlier
    ranges , preserving all the information, but you see the central most likely area well ,
    and the comparison of past and recent temperature levels is not as influenced by the
    outlier estimates. What do you think? We have experimented with different versions of the
    shading and this one shows up quite well – but we may have to use some all grey version as
    the background to the overlay of the model results.”
    Probably it is a better use of your life force to consider that entry into the ‘reconstruction derby’ of which you hinted elsewhere.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s