Being a farmer whose work and success depends highly on the weather, I carefully monitor the local weather forecasts and work hard to understand how weather patterns and the computer models used by forecasters actually work. I appreciate that meteorologists use a number of different computer models to decide how the weather is likely to behave, however I also understand from my past professional experience that all great theories are great in theory, but not necessarily in reality. To explain why I make that statement, I need to explain to you what I have learned about complex computer models and how they operate. Please bear with me as I reveal the secrets of computer modeling as simply as I can. It is a concern I have touched upon in many prior posts on our farm website, but have not explained in detail. Depending on your political perspective, you may or may not like what I have to say, but you are free to confirm it by talking to any professional Transportation Planner for a Metropolitan Planning Organization (MPO), and it will be confirmed.
You see, I once used computer models designed to predict local traffic patterns in Macon, GA and Anniston, AL as a tool to decide when and where road improvements should be designed and scheduled. Those models were based on a number of critical assumptions about the factors that can influence travel patterns. They include the locations and size of trip attractors (employers, stores, institutions, recreational facilities) and trip production factors (concentrations or density of households, household incomes and the distance between trip generators and trip attractors). Some of those factors could not be accurately or objectively measured, especially those influenced by human choice and behavior. We may be the most intellectually advanced species on our planet, but that hardly makes us very rational or predictable. That fact is why I have often said that you cannot reduce human behavior to a formula. The computer program would then take all this data and apply a “gravity model” (which was the formula for Newton’s Law of Gravity) to calculate traffic volumes on the local street network of the community.
While a “Gravity Model” is a useful and scientifically rational way to understand how traffic patterns work, it never worked precisely correct. Whenever we applied the model to the community’s street network, the traffic volumes predicted by the computer never matched the actual traffic volumes we recorded on those streets. Why? Because the computer model was based on a number of critical assumptions about human behavior which can be influenced by a number of independent variables that cannot be reduced to an accurate and reliable formula. These assumptions also have varying degrees of influence on the patterns that they are used to model. In order to “fix” the inconsistency, we had to “adjust or manipulate” the data inputs through a series of iterative (updated and repeated) model runs until the traffic volumes it predicted more closely matched the actual traffic volumes we had counted on the major roads. However, when you manipulate the data (assumptions) you input into the computer to predict traffic volumes in that way, you potentially compromise the integrity and the margin of error in your results. That margin of error may, in turn, reduce in an unpredictable (and perhaps unmeasurable) way the reliability of future traffic projections by the computer. However, this is the way that most, if not all, of our most sophisticated and complex computer models are developed and actually work.
This professional work experience I gained working with computer models used to understand complex patterns affected by a wide array of independent variables is precisely why I want to understand and critically evaluate the basic or core assumptions for the climate and water quality models that fuel our divisive political debates over climate change and water quality. Those core assumptions are never discussed publicly, but as my example clearly shows, they are the critical issues to consider when assessing how reliable or accurate their forecasts are. The fact that meteorologists consider many different weather model predictions to determine where a storm will go and how it will develop shows that they understand what I am saying and that their models are subject to the same sources of error as the traffic models I once used in my planning career.
I wanted to explain all this to you because the public deserves to understand that no matter what a computer model predicts, the results are influenced greatly by the core assumptions made by the programmer and how they had to be “adjusted” to make the model reflect reality. At either point, the “scientists” who create the program can and do introduce their own biases into the model, whether deliberately or accidentally. It is at those levels that critical thinking must first be applied to decide how accurate the model is. Just look at how much divergence there can be between different storm track models in predicting the path of one storm over 24 or 36 hours, and you can see why I would ask a lot of questions about climate models designed to accurately and reliably predict global weather patterns over the next fifty years. As for opinions regarding how to fix the problems a computer model identifies, a much wider critical debate can be conducted. This may explain why many scientists are unwilling to explain their critical assumptions. It is hard to defend assumptions that may be open to controversy.
I hope you will excuse my philosophical sidestep into the world of computer modeling, but I always try to explain the reasoning behind my thinking so that I can’t be easily misunderstood or worse yet, misrepresented. That is far too common in the world today.
The point I wanted to make in this post that only adds to the computer modeling issues I have discussed is the influence of “microclimates.” That is something that most farmers clearly understand, especially those in our Potomac Highlands region of West Virginia. Microclimates are variations in local weather patterns typically caused by or resulting from sudden and abrupt changes in topography. They can have a significant impact on the growing season as well as daily outdoor activities. They might affect one very isolated valley or hollow or larger areas where the surface elevation changes significantly over short distances. We have both of these microclimate influences throughout our region. In fact, changing microclimates over geologic periods of time may be one factor that could affect the relative integrity of long-term historic temperature data assumptions used in climate models.
Our farm is located in the North Mill Creek Valley at the base of Cave Mountain. However, only three ridgelines (twelve miles by air) to our west is the massive Allegheny Front escarpment, where the base elevation dramatically increases by 2,500 feet or more in a span of one mile or so. Elevations in the North Fork Valley at the base of the Front are less than 2,000 feet above sea level, while the land west of the summit ranges between 3,200 and 4,500 feet, causing dramatic weather changes over a very short distance—especially in the winter months. On the plateau to the west of the Front, annual snowfall can easily (and often does) exceed 150 inches, while our average annual snowfall in the Mill Creek Valley is roughly 25 inches. Major snowstorms that move northeast along the Atlantic coastline will drop more snow (or rain, as the case may be) on us and less beyond the Front, while the opposite is true for snowstorms that approach us from the west. No matter what weather conditions we experience at our farm, those on the Front are typically more severe, unless an upper-level mass of warm air keeps the higher elevations warmer, while trapping colder air at the lower elevations. This is an effect that does occur occasionally throughout the year and that I first introduced to our readers in my February 9, 2017 post entitled, “Angel’s Breath.”
We experienced the dramatic effects of our microclimates two days ago, when we traveled over the Allegheny Front and across the plateau to Elkins. Where the ground has been bare at our farm since December 22, we found a complete blanket of snow as we crossed the plateau and drove down through Canaan Valley, where Canaan and Timberline Ski Areas flank the slopes of Cabin Mountain. We drove through woodland forests of snow-draped evergreen trees and unbroken fields of white throughout our journey. It was a distinct change from our own golden-brown hayfield and our fallow vegetable garden, which is now covered with a contrasting green carpet of winter rye, which we planted to infuse nitrogen in the soil.
Our snowfall total for the season to date is about 11.5 inches, ten of which fell in our big December 16-17 snowstorm. Our winter temperatures have not been exceedingly warm or cold, relative to what we would typically expect to see, but our long-range forecast for late January and early February suggests that a polar vortex will dive into our area, pushing our low temperatures down to the low single digits. Our average coldest temperatures are supposed to fall between -5 and +5 degrees, and if the current forecast holds, it appears we will bottom out in that range for the fourth time in the five winters we have retired at our farm. We also managed to get our second white Christmas in those five winter seasons, where the long-term average probability of white Christmases in our area is about thirty percent. I don’t know how much snow they have received above the Allegheny Front, but I do know it is a lot more.
The low temperature at our farm this morning was 18°, which was about five degrees below the average low for the date. It is one of the coolest mornings we’ve had since late December, and Calli was unprepared for it. When I opened the door to our front porch at 5:00 AM this morning, Calli approached the doorway, but froze when the cold air hit her face. After some reassuring coaxing, I managed to get her to go out, but she returned after only thirty minutes outside and dashed into the house and onto the bed in the guest bedroom, bypassing her typical visit to her food bowls for a few crunchies and cat treats. She promptly curled up on Barb’s quilting fabrics and slept for another hour. She had completed her morning bathroom break somewhere in the front yard and returned to the warmth and security of her current bed-down site. As our official Peeper Pond Meteorologist, she had confirmed for me just how relatively cold it was outside this morning. Her weather reporting acumen remains far more accurate and reliable than anything I could get from a computer model, despite the peculiarities of our own little microclimate.
Our lingering question is whether or not our goats became pregnant this past fall when our guest buck, Little Bit, visited our farm. I don’t believe that Snowball is pregnant because she has exhibited some signs of repeated heat cycles since his visit. I am much more hopeful that Essie may be pregnant. She seems to have grown bigger during the past two months, but I’m not sure if it’s due to her pregnancy or her more lethargic winter behavior. Unfortunately, it’s difficult to know if a goat is pregnant, especially now that her breeding season has lapsed. We’ll just have to remain hopeful (as farmers typically are) for the next two months when she should be due to give birth.
Until then, I wish all our loyal readers a happy winter.