The debate over tree ring data, the Yamal Larch, and whether or not Briffa’s studies show that there was no Medieval Warming Period or Little Ice Age has faded from the news and the blogosphere of late. Himalayan glaciers and missing data have the front page. But I would like to go back to tree rings.
One of my beefs with long term temperature reconstructions based on tree rings is that they don’t record any information when the tree is not growing. For southern latitudes where the growing season is most of the year, that’s less of a problem (but still a problem). In higher latitudes, the growing season is shorter, sometimes only a few months. Various tree ring chronologies dismiss this, claiming that they’ve got enough surface station data to show that the winter temperature averages and the summer temperature averages are not statistically different from their respective means. I’m not so certain since the studies I’ve looked at don’t actually produce the data to make that claim, and anecdotally I am the veteran survivor of many unusually cold winters that were followed by blazing hot summers and mild winters that were followed by cool summers.
Much of the criticism of tree ring studies is that they can be influenced by so many factors, and that a lot of studies have shown that the driving factor is precipitation, not temperature. Other factors also mess with tree ring data, such as nearby trees falling down and exposing the data tree to additional sunlight and ground moisture. 400 years later, who’s going to know? In my backyard I have a stand of red oak trees that are over 100 years old. One of them stands apart from the others. We had a late frost one year and that one tree, being in the open compared to the others took the brunt of it. The leaf buds on all the trees were severely damaged, but that one got clobbered. I thought it would die. It clung to life and finally recovered, but 15 years later it still looks stunted compared to the rest of them. Again, 400 years later, who would know?
But the only way to refute the claim is to show some data that conflicts with the tree ring chronology. Siberian surface station data not being easily handy to start an argument with a larch tree, I went looking for alternative data sets. The Red River flood plain and local Red Oak trees more or less confirm my hypothesis.
The Red River flows through Winnipeg, Manitoba. It’s catchment area is 500 miles long, and it is also fed by another major river, the Assiniboia. The combined catchment area is almost 30,000 square kilometres. More importantly, the catchment area is a north temperate climate. Snow that falls in November doesn’t melt until March, sometimes April. When it does melt, it melts in a VERY short time period. As a consequence, the entire precipitation of the winter has to flow through the Red River in a couple of weeks. Since we have peak flow rate data from the Red River going back to 1892, we can see what the relative size of that winter’s snowfall was:
The “Flood of the Century” for example, in 1997, sticks out like a sore thumb. We had a massive amount of snow that year, and the flooding created a lake larger than one of the Great Lakes. 1950 also sticks out. Before 1892 we have historical and other evidence showing that there was a flood in 1826 that dwarfed 1997 with peak flow rates another 30% to 40% higher. So back to the tree rings.
The Geologic Survey of Canada published this report in 2001 http://www.gov.mb.ca/stem/mrd/geo/pflood/p_pdfs/climaticextremesinsmb.pdf
They used tree ring data to reconstruct climate for the last several hundred years in the same general area. Unlike Birffa et al, they did not claim that this was representative of temperature trends. They claimed that it was representative of precipitation. They even went so far as to correlate to temperature, and concluded that increased temperatures had a NEGATIVE effect on tree ring growth, though so small as to be negligible, and that the main driver was in fact precipitation. So now we have a credible proxy for growing season precipitation in the same general area:
So, do the peak flow rates in the Red River reflective of the previous winter’s total snowfall show up in the tree ring data? It seems like they don’t. The “Flood of the Century” in 1997 doesn’t stand out from the data, though annual precipitation is rising for that time period with a large spike in 2000. The much larger flood in 1826, which should be even more dramatic being a peak flow rate much higher than that of 1997, is also not visible. In fact, according to a tree ring study focused on identifying large floods (tree partly submerged in flood water in spring produce a distinctive tree ring), the 7 largest floods of the last 350 years or so were in 1997, 1979, 1950, 1852, 1826 (the largest) 1762 and 1747. In brief, not a good match. Now that is NOT conclusive evidence. There is nothing to say that the following year’s rainfall wasn’t lower than normal, and the average for the year about what the tree rings show. Unlikely, but possible. The more logical conclusions to draw are that:
1) Tree rings are a lousy proxy for temperature in the first place as they are dominated by precipitation, not temperature, and potentially contaminated by events for which there is no record.
2) Tree rings are reflective of the growing season only, and exclude significant climactic trends that are most pronounced in winter.
Any long term temperature reconstruction based on short growing season trees (like Larch trees from Siberia) is not only suspect as a temperature proxy, but is also a data set that excludes significant off season data.