Observations of a Skiing Actuary – Part 2

In the last post, Tony Crocker offered some insights into what he’s been seeing at a general level in his snow-tracking pursuits.   Here, I present some of the actual findings.

In e-mail correspondence, he clarified the calculations of the numbers and provided some additional thoughts.   While this isn’t a guest post, but a summary of our correspondence, all work is Tony’s and most of the pertinent observations come from him.

The Data

Tony has broken North America into 8 regions.   He used his own judgment in defining the regions, but in his words “My regional definitions are somewhat arbitrary (though most skiers would consider them reasonable), and the regions are big enough that within a region there can sometimes be some areas with good seasons and others with bad seasons in the same year.”  The regions are as follows:

1 – California

2 – Pacific Northwest

3 – Interior Canada

4 – U.S. Northern Rockies

5 – Utah

6 – Northern & Central Colorado

7 – Southern & Western Colorado

8 – Northeast

Within each of the regions, the data point is “Percentage of snowfall relative to normal.”  The final data point for the region is a straight average of all the observations in that region.   The Data is tracked as far back as 1970-71 season.   Some of the years 1970-75 had significant snowfall, but he excludes those years in some calculations.   The main reason for excluding the data is the low number of data points available.

Data Considerations:  (1) Spread of resorts within region could cause weighting issues which do not appear to have been considered; (2) number of resorts with good information can differ from year to year; (3) the expected snowfall will change from year to year as data is added, but Tony recalculates historical data to account for that.

Observations (1975-2009)

*Most variable region = California shows a standard deviation of 31%.   The most consistent, least variable, data is the Interior of Canada (15%).   Overall North America had a standard deviation of 13% over the time frame.

*The longest trend, including the less credible data of the early 1970s shows an overall trend from 1970 – 2009 of -2.9%.   However, Tony ran trends from 1972, 1975, 1982, 1987, and 1992 that were all positive.  The highest trend in overall North American snowfall, in fact, is since 1975, at +9.8%.

*Every single region has a positive trend since 1975.  Some regions show some negatives during the shorter trend time-frames.   The region with the largest trend is the Pacific Northwest, +21.1% since 1975.  But in a show of how volatile results can be, this same region actually has the largest negative trend back to 1970 because of high snowfall early in that decade.

*There is some significant correlation in snowfall between many regions, though not all.    It’s not necessarily surprising that the Utah and Colorado regions are very highly correllated.   Other areas of high correlation (over 50%)  are:  California/U.S. Northern Rockies; California/Utah; California/Southern & Western Colorado; Pacific Northwest/Interior Canada; Pacific Northwest/U.S. Northern Rockies; Northern Rockies/Utah; Northern Rockies/Both Colorado Regions.

*Negative correlation is indicated between: California/Canadian Rockies (most distinct negative correlation); Canadian Rockies/Utah; Canadian Rockies/South & West Colorado; California/Northeast.

Conclusions:

I asked Tony, based on his experience and observation, what are his main conclusions.  The following is his response:

1 – No trend in snowfall over the past 35-40 years in North American ski areas despite rising temperatures during most of that period.

2 – Rising temps don’t necessarily affect precipitation, but do affect the rain/snow line.  This is not yet an issue at almost any ski area in North America (Whistler base and Snoqualmie might be exceptions) but there are other ski regions of the world (Australia comes to mind) where it might be more serious.

3 – data is very volatile and it’s dangerous to claim trends based upon typical measuring periods like 5 or 10 years. 

Charts:

I’m providing three charts:  the overall North American chart, the California chart and the Pacific Northwest chart.   On the Pacific Northwest chart, keep in mind that from 1976 forward the trend is largely positive, but from 1971 forward it is negative.   California shows the volatility of the region, and the difficulty in being able to make grand assessments in the short term regarding snowfall and temperature.

North American Snowfall

California Snowfall

pacific_northwest_snowfall

Thanks for the information, Tony.

Perspectives and Observations From a Skiing Actuary

The following “guest post” is from Tony Crocker, an actuary and avid skier who has documented snowfall as a result of his hobby.   The post below is actually somewhat dated – he had sent it last October, so keep that in mind with regard to his references to sunspot activity.

I asked for his permission to pass on some of the results of his study as a guest post.  He granted it and then I subsequently dropped the ball on passing it along.   I am first sharing his initial e-mail to me as a guest post, as he has some general thoughts and ideas based on his observations, and his own study.   The post below is not to be taken as advocacy on my part with regard to the strategy on taxation of carbon versus payroll tax, but it’s another viewpoint in the debate.

Perspectives and Observations from a Skiing Actuary

Guest Post by Tony Crocker

We have a few things in common.  I’m an actuary in Los Angeles, only a BA in statistics, no formal training in climatology. But I’m also an addicted skier and began collecting snowfall data in 1991, going back as far as 1967 in a few cases.  This was published in Powder Magazine in 1995, and I’ve continued to collect and expand it, now covering 101 locations in North America, summarized on my website http://bestsnow.net.

Because of my collection of snow statistics I do get asked the global warming questions from time to time.  Putting cause and effect aside, there is little question that temperatures increased from ~1975-2000 or so.  My snow data covers that period, and there is no trend in snowfall.  The reason for this is that precipitation does not necessarily decrease with rising temps, and in some snowy locations you could make the argument that it might increase.  At any rate, from a skier’s perspective the danger is from a rise in the average rain/snow line.  Fortunately in western North America virtually all ski areas are well above the typical rain/snow line of even the past decade. There are other places in the world (Australia, small areas near the villages in the Alps) where a rising rain/snow line is becoming a problem. 

You might wonder about the Northeast, where altitudes are low and rain is the key issue in degrading snow conditions. The problem there is temperature volatility and varied sources of weather. If storms come from the Gulf of Mexico they will be rain, just as they were 30 years ago, and if from Canada or the Great Lakes it will be all snow.  Nor’easters (Atlantic-based storms) are more complex, depends on whether they run into a cold mass of air onshore.  Hurricane Wilma in late October 2005 ended up running into such an airmass and dumped 4 feet of snow in the NH and VT mountains.

As a statistician I am not impressed at all with the accuracy of IPCC models and suspect there is no more than a 20% chance that they are correct.  However, as an insurance actuary I would advocate precautions against low probability disaster scenarios. So I’m willing to tax carbon, providing the revenue is exactly offset by payroll and income tax cuts.  Some of the tax should be on OPEC oil imports, since downside of US dependence on OPEC is clear while the downside of too much CO2 might be a big problem but more likely is a less pressing issue.  I would take the position that even if taxing carbon is irrelevant environmentally it’s still better for the economy than the payroll/income taxes it would replace.

At any rate the flat temperatures of the past decade and possible decline over the last couple of years should be giving the IPCC etc. pause, and they should be trying to figure out what needs to be added to their climate models.  After all, CO2 output has been rising rapidly, and a decade ago they said temperatures would if anything increase faster than over the previous 25 years.

Presumably you’ve read the David Archibald paper predicting a replay of the early 19th century Dalton minimum and an ensuing sharp drop in temperatures. The key soft spot in these predictions is the feedback effect of water vapor from increased CO2.  The IPCC etc. says the water vapor feedback will at least triple the greenhouse effect of the CO2.  Archibald says the water vapor feedback is negative.  Neither of these assertions can be proven as far as I know. 

You say that significant temperature declines are likely if the next solar max is <100.  Archibald predicts it will be 40.  His climate assumptions are questionable, but he made the sunspot predictions nearly 2 years ago, estimating that the solar minimum might not occur until July 2009.  While August 2009 was the first spotless month in 95 years, the 2010 increase in sunspot activity confirms that the smoothed minimum between Cycles 23 and 24 bottomed in December 2008.

As a skier I will be pleased if temperatures continue to stay flat or decline over the next 20 years.

The Ultimate Peer Review

The following is a guest post by Bob Heiderstadt.   I, firstly, must apologize to him for not having posted it months ago when he first submitted it to me as a potential guest post.   Bob, I did not pass it along for any issues that I had with it, it is just one of those things where I put it in a folder to look at later, and just never did.

Anyway, it is not about Climate Change, per se.   But so much of the debate about climate science these days – and highlighted by the Climategate scandal – is about peer review.   Many times when a very eloquent and insightful argument is made on a blog somewhere, one of the immediate tools of defense of those dismissing the argument is because it isn’t peer reviewed.  While it seems an agreeable practice to have peer review, Climategate demonstrates a darker side to it:  when the review group is a smaller circle unable to review a work in an unbiased fashion, and instead looks for ways to discredit a paper not because of its science, but because of its conclusion, we have a problem.

Bob put together some thoughts on what an open and honest process may look like.   Such a process seems more open.   A paper should be judged by its content, scientific application, accuracy, and presentation.   Not by titles or preconceived conclusions.

The Ultimate Peer Review

Guest Post by Bob Heiderstadt

Wikipedia says “Peer review (also known as refereeing) is the process of subjecting an author’s scholarly work, research, or ideas to the scrutiny of others who are experts in the same field. Peer review requires a community of experts in a given (and often narrowly defined) field, who are qualified and able to perform impartial review. Impartial review, especially of work in less narrowly defined or inter-disciplinary fields, may be difficult to accomplish; and the significance (good or bad) of an idea may never be widely appreciated among its contemporaries. Although generally considered essential to academic quality, peer review has been criticized as ineffective, slow, and misunderstood.”

A PhD can be jokingly described as “someone who knows almost everything about next to nothing.”  Seriously though, someone who has earned a PhD does know significantly more about a particular subject than the vast majority of the population.  In fact, there may only be a handful of individuals who may as knowledgeable about a particular subject, regardless of whether it is in the social sciences or the hard sciences.

No PhD, regardless of how knowledgeable he/she is in a particular subject, can be considered an expert in all of the related areas that support his/her area of expertise.  A scientist, for example, may be an expert in his/her area of science, but may have only adequate training in statistical analysis or language to analyze or express the analysis.  This is not to say that their skills are inadequate in the support areas, but it’s impossible to be an expert in everything.  This is where the assistance of others who are more knowledgeable in those support areas provides a great benefit, or added value, to the expert in preparing a study for publishing.

The internet has become the ultimate peer review, not because a paper is reviewed by a select few experts, but because it can be reviewed by anyone, from a high school dropout with an interest in the subject, but with maybe a unique insight, to the PhD who is an expert in the area of study. Granted, some comments will not be helpful, but as the analysis of Stieg, et al has demonstrated, a team of people who are not necessarily experts in that particular area of science, but may be engineers, statisticians, economists, and yes, scientists can make a valuable contribution to the review.  It is that varied experience and knowledge that can bring a different point of view to a study and point out deficiencies or irregularities in the analysis or presentation that ultimately make the study better.  This is the true essence of peer review, to try to obtain the best product possible, while winnowing out the junk.

With this in mind, I propose the following:

1. Establish an internet web site to behave as a repository for papers submitted for review and give it a name similar to “SciencePeerReview.com or .org”
2. Establish different categories of science to be reviewed (hard sciences or soft sciences).
3. Provide sufficient storage to accept large papers and supporting data
4. Develop a way to pay for the web site (advertising or a fee paid by the submitting individual(s).
5. Accept only papers submitted with supporting data and procedures (data and software)
6. Submitted papers should be in a pdf, Microsoft Word^®, or Open Office format.
7. Allow any individual to review and comment on any paper with the following restrictions:
   • The individual must identify himself/herself to the web site and provide minimal information regarding education and/or credentials and experience.
   • Comments must be pertinent to the paper being reviewed and polite.
8. Optionally, a reviewer could receive an automated ID for the author, permitting anonymity.
9. Allow two different styles of comments:
   • Blog style – visible to other reviewers or optionally for the author only.
   • Document style – for detailed analysis of submitted paper (see 6 above), again with the option to post for other reviewers or directed to the author only.
10. Allow the author to request a specific review period based on the size and complexity of the paper so that there would be adequate time to complete a review
11. The author could respond to comments to clarify stated concerns (blog style).

 

Whether the final paper is presented on this site, or through some other venue, should be at the option of the author, although if published somewhere else, the appropriate reference should be made.

Since this is about peer review, if you read it, please comment on it to make it better, or comment even if you think it is a lousy idea.  That’s the whole point.

The Unifying Theory of Earth’s Climate

Stephen Wilde contacted me regarding an article that he has presented exclusively on CO2Skeptics.com. It is definitely worth the read. I have posted a couple excerpts here, but I encourage you to CLICK HERE to read the entire paper. Many thanks to Stephen for giving me a heads up on this.

The Unifying Theory of Earth’s Climate
Guest post by Stephen Wilde – excerpts from his paper of the same name.

The claims of those who worry about human damage to the climate become ever more strident despite, or perhaps because of, the real world data rapidly diverging from that which they anticipated.

(Figure 1) The failure of alarmist predictions

It is now ten years since the 1998 culmination of a period of thirty years of unusual ocean warmth that resulted in the atmospheric temperature peak of that year. Additionally during that period the sun was more active than ever previously recorded. ( Figures 2 and 4)

Figure 2 The high solar activity from 1940 to 2000.

AGW proponents accept that the relative coolness of the past 10 years (Figure 3) is a result of cooler oceans but refuse to accept the corollary that the primary cause of the warmer period was warmer oceans. Warmer oceans also expand. ( Figure 5) and release natural CO2. The apparent levelling off in the sea level rise is coincident with recent cooler ocean surfaces.

See CO2Skeptics.com for the full presentation.

My own comments:
Mr. Wilde hits on a couple points I have made myself, which shows an understanding of cyclical changes in trends and the ability to apply simple logic. Most of us – even the skeptics – acknowledge that warming occurred from the 1970s to the end of the century. So, when there has been no warming for the last number of years, yet we hear the arguments that the latest year is still in the top X of our records, it’s kind of silly. I have used the analogy of climbing a mountain, getting to the top, and coming down the other side. Even though your are heading down once over the top, your first steps are still close to the top. In fact, if you trended elevation by time, you would continue to see a positive trend line for quite a while as you headed down, because those elevation points would be near the top of the mountain, even if descending. Ignoring the recent negative changes in elevation and only looking at the overall trend line would lead one to suggest that you’re still climbing up the mountain. While we cannot prove definitively that this is happening with temperature, to completely ignore the possibility is to put blinders on. If this truly is a cycle, then Mr. Wilde is absolutely correct in his assertion that points will cluster for a time at peaks and troughs.

In addition, Mr. Wilde criticizes climate models that are “built upwards from innumerable details rather than downwards from a verifiable overarching concept.” I have also addressed this issue here, and completely agree. I work in a profession that relies on modeling. I can’t tell you how often I scrap a more complex model that tries to capture all the details through numerous inputs in favor of a simple model that looks at things from a broad perspective. The issue is not that complex, comprehensive models are poor in concept. The issue is that if you are building a model in that fashion and you are missing anything, you end up with cross biases where things get inappropriately attributed to certain factors, and the results are nonsensical. More often than we’d like to admit, the better approach is to simply admit that we don’t fully understand all the impacts of all the components, and we need to accept that high-level, general, and simpler models are actually better. Mutliple times a year I read a story about something dealing with climate where there is some unanticipated effect of something-or-another. This tells me, then, that models built on a need for comprehensive analysis are erroneous.

Mr. Wilde’s summary conclusion is that it’s all about the oceans. This encourages me to get back to me more comprehensive correlation analysis on all the ocean indices. (Still waiting on that December PDO reading…)