Archive for October, 2011

The seasons

October 7, 2011 Leave a comment

…or, how not to analyze temperature and cloud cover.

A recent post on Watts Up With That by Erl Happ, titled “High clouds and surface temperature“, attempts to use the NCEP reanalysis data set to examine such a relationship.

First, as Erl is (presumably) not a scientist or researcher, I cannot fault him for sourcing his cloud data from a reanalysis product. Reanalysis products are quite appealing because they offer all sorts of variables – temperature, wind, moisture, and “value added” items like cloud fraction – in a gridded data set. The problem is that reanalysis data are not “real” data.

In general, most reanalysis products are assimilations of data from many sources – weather balloons, satellites for periods post-1979, surface stations, etc. – into the first time-step of a sophisticated, high-resolution weather system model.

Don’t get me wrong. Reanalysis products are extremely useful if they are used correctly. I have used them in a paper I co-authored and they are a major part of my current research. For “basic” variables, like temperature and the horizontal wind field, they are fairly good representations of the “real world”. The problem comes when you try to use these value-added things like cloud cover. I will not go into great detail, but suffice to say, you are better off using actual cloud observations from, for example, MODIS.

Okay, moving on to the real heart of the matter, Erl’s basic misunderstanding of the earth’s seasonality cripples his ability to perform any valuable analysis. I will first quote the relevant text:

“The minimum [in surface temperature] is experienced when the Earth is closest to the sun. The Earth is coolest at this time because the atmosphere is cloudier in January. January is characterized by a relative abundance of high ice cloud in the southern hemisphere. Relative humidity peaks in April (figure 6) when tropical waters are warmest. I suggest the variation in the minimum global temperature is due to change in high altitude cloud…”

A schematic of earth's orbit. Perihelion, the time when earth is closest to the sun, occurs in January - NH winter and SH summer.

Earth is closest to the sun in January and farthest in July – absolutely true. Now, the axial tilt of the earth determines the season each hemisphere is in at a particular time of year by directly altering the distribution of solar radiation. In northern hemisphere winter, say January, the northern hemisphere is tilted away from the sun while the southern hemisphere is tilted toward the sun, relative to earth’s orbital plane. This delivers more total radiation (and more radiation per unit area) to the southern hemisphere, and less to the northern hemisphere. However, the earth is closer to the sun and hence receiving more total solar radiation than it does in, say, northern hemisphere summer (July).

What we would expect then, all things being equal, is that the globally-averaged surface temperature would be highest in January when the earth is closest to the sun and receiving the most solar radiation.

Erl just asserts the temperature actually peaks in July without providing an illustrative figure. I hate assertions, so let’s look at some data. I’ve used the 2-meter temperature field from the Modern-Era Retrospective Reanalysis from NASA to generate the following plot of the global average surface temperature. As with all reanalysis data from US government agencies, it is free for anyone to download – atmospheric science enjoys a high degree of data sharing that many fields do not, and hey, your tax dollars paid for this so you should be able to use it! The time period is 1979-2010 and I’ve used area-weighting by the cosine of latitude.

Global average 2-meter temperature. MERRA data is courtesy of NASA's GMAO.

Whoah, it’s higher in July, when the earth is farthest from the sun! Erl was right, what gives?

Well, there is a crucial difference between the hemispheres – there is very little land in the southern hemisphere, and quite a lot of land in the northern hemisphere, especially in the midlatitudes. I’m going to claim that the northern hemisphere land masses are dominating this signal – despite the fact that the earth is 70% ocean! Why? Because they have such a high-amplitude seasonal cycle of temperature. This is due to the fact that the land surface has a much smaller heat  capacity per unit area than the oceans, so for a given amount of radiation absorbed, the land surface will increase its temperature more than the ocean.

Let’s apply a land/ocean mask to the same 2-meter temperature data to try to isolate the two surfaces. To make the comparison, I’ll compare their anomalies relative to their annual average.

2-meter temperature annual cycle for land and ocean. MERRA data courtesy of NASA's GMAO.

Bingo. As you can see, the amplitude of the seasonal cycle is dominated by the land surfaces. Indeed, the average temperature over the oceans barely changes at all, peaking in the late southern hemisphere summer/early fall as we would expect due to the thermal inertia of the oceans. This part of the signal is completely lost when you examine the global average temperature without separating land and ocean data – the global annual cycle has an amplitude of around 4C, while the ocean surfaces have an amplitude of less than half of a degree. Compare that to a nearly 12C amplitude for the land surfaces!

Erl could have saved himself a lot of work and the rest of us a lot of trouble if he would have just used a land/ocean mask, as suggested in the comments at WUWT, to prove to himself that the “continentality” he so derides is in fact the case. Continents heat up a lot in summer and cool down a lot in winter because of their low heat capacity, relative to the oceans. Because there is much more land in the northern hemisphere, the seasonal cycle of global average temperature is dominated by the northern hemisphere seasonality over land. That is, global average temperature tends to peak in July and reach a minimum in January because the northern hemisphere land surface reaches its maximum and minimum temperatures in those same months. This is not a surprise to anyone who has taken a basic course on earth’s climate.

Yes, it’s counter-intuitive that the earth’s global average temperature is highest when it is farthest from the sun, but the explanation is very simple.

“Variation in cloud cover should be the first hypothesis to explore when the Earth warms or cools over time. You would have to be very naive to think that the inter-annual change in temperature that is most obvious between November and March could be due to something other than a change in cloud cover.”

Naive, indeed. This is a striking example of the Dunning-Kruger effect.

Categories: Uncategorized

The Arctic death spiral

October 1, 2011 Leave a comment

Those following the observations of Arctic sea ice extent and volume were probably not surprised when the summer minimum numbers rolled in and 2011 had the lowest or second-lowest sea ice extent since monitoring began in the 1970’s. The downward trend shows no sign of stopping, and the distinction between lowest or second-lowest is unimportant, as year-to-year extent is influenced by surface wind patterns. No, what is important is this graph from the National Snow and Ice Data Center:


Sea ice extent from the NSIDC


The greyed-region is +/-2 standard deviations, with the central line the 21-year average for 1979-2000. The current extent values, especially at the summer minimum, are strikingly low. Just eye-balling the chart suggests that 2007 and 2011 are approach a deviation of more than 3-sigma from the average. The other years in the past decade are almost as low. What about the volume of sea ice in the Arctic?


Sea ice volume from PIOMASS

While the summer minimum sea ice extent has approximately halved, volume has decreased by almost 75%. This is especially troubling – thin ice responds more rapidly to variations in temperature and weather patterns, and the volume of multi-year ice is in rapid decline.

But sea ice is not the only victim in the Arctic. Last week The Conversation posted a stunning article, “Canadian ice shelves halve in six years”. From the article:

Half of Canada’s ancient ice shelves have disappeared in the last six years, researchers have said, with new data showing significant portions melted in the last year alone.

“Since the end of July, pieces equaling one and a half times the size of Manhattan Island have broken off,” said Luke Copland, researcher in the Department of Geography at the University of Ottawa.

These are shelves that have existed since long-before Europeans arrived. Let that sink in.

From a related article from ABC News:

Between 1906 and 1982, there has been a 90 percent reduction in the areal extent of ice shelves along the entire coastline, according to data published by W.F. Vincent at Quebec’s Laval University. The former extensive “Ellesmere Island Ice Sheet” was reduced to six smaller, separate ice shelves: Serson, Petersen, Milne, Ayles, Ward Hunt and Markham. In 2005, the Ayles Ice Shelf whittled almost completely away, as did the Markham Ice Shelf in 2008 and the Serson this year.

Ice shelves are massive, floating platforms of ice, often at the terminus of marine glaciers. Unlike sea ice, which thins and thickens with the seasons and is constantly jostled around by winds, these shelves are more permanent, though still dynamic, features. They are only native to the Arctic regions, as the ice would otherwise melt long before it reached the sea if annual-average temperatures were not sub-freezing. The primary mechanism for ice loss from these shelves is calving – when the ice reaches a certain distance beyond the grounding line, where it is anchored to the seabed, chunks mechanically shear off to form icebergs.

However, ice shelves across the world have been losing mass over the past decades, many at an ever-accelerating pace, including the dramatic collapse of the Larsen Ice Shelf in Antarctica. Glaciologists have pinpointed two major sources for this acceleration – warming ocean waters that undermine the shelf from below, and surface melt-water pools that chisel vertical fractures into the shelf, greatly reducing its structural integrity.

The accelerating decline of sea ice, ice shelves, and glaciers is but one line of evidence that demonstrates the world is warming. Unfortunately, the loss of ice contributes to the ice-albedo feedback and is set to not only disrupt ecosystems, but threaten water supplies for the millions that rely on glacial meltwater. Perhaps, though, the visibility of this phenomenon will finally start to resonate in people.

The great retreat of Jakobshaevn Isbrae in Greenland, via GRID-Arendal

We live in a changing world – we clearly have the power to disrupt it for the worse, but that also means we have the ability to shape it for the better.

Categories: Arctic, Ice and snow