Forecasting Lake Effect Snow in Lake Superior region
Summary
This exercise is designed to present the realistic problems of forecasting weather. Lake effect snows are hard to forecast because they depend on information that isn't part of the regular set of information and involve some pretty specific things that integrate the location of the site with surrounding environment. Even places close by can get totally different forecasts. When you have a regional forecast, it doesn't really address lake effect snows, unless the forecaster really focuses. So the exercise aims to show the value of broad critical thinking in meteorology, and it is very dramatic, because the difference between 36 inches and whiteout and clear blue sky is undeniable.
Collapse
Context
Audience
This is a beginning Atmospheric Science class, organized around the AMS effort and using their materials and exercises. It is aimed to complement those efforts.
Skills and concepts that students must have mastered
Students need to be on top of the first half of the class, which includes a lot about atmospheric data, real time collations, maps, fluid mechanics, latent heat, stability, remote sensing, upper air data, and meteorological models. These are not really mastered, but at least they are known issues.
How the activity is situated in the course
This is more a culminating exercise than anything else. It is aimed at personalizing our forecasting understanding, by focusing on something that profoundly affects living here (how much snow you will have to shovel to get to class). Doing this involves a creative commitment that isn't in the textbook or any of the very fine materials that AMS provides, and also involves critical thinking and very considerable transfer from environmental sciences and math. The duration of the exercise is several weeks, which allows for examination and evaluation during this process. When the first snowstorm occurs, hopefully early in the data collection, we can examine and dissect the conditions to see if students can become more accomplished at the process of forecasting. It may be possible to do this several times during the several weeks. If this first snow doesn't appear there will be a case study to look at, perhaps obtained from the local NWS office of a snowstorm event that we can practice with.
Goals
Content/concepts goals for this activity
Atmospheric motion, instability, boundary layer meteorology, microclimates, scientific uncertainty
Higher order thinking skills goals for this activity
Real time data integration and analysis, insight into challenging forecasting, use of inadequate data, self improvement through validation
Other skills goals for this activity
Students will work in groups to develop forecasts. This will enable them to learn from each other, because one cannot learn all the details of this process easily alone. During the last week the groups will compete in doing forecasting.
Description of the activity/assignment
This exercise is designed to present the realistic problems of forecasting weather. Lake effect snows are hard to forecast because they depend on information that isn't part of the regular set of information and involve some pretty specific things that integrate the location of the site with surrounding environment. Even places close by can get totally different forecasts. When you have a regional forecast, it doesn't really address lake effect snows, unless the forecaster really focuses. So the exercise aims to show the value of broad critical thinking in meteorology, and it is very dramatic, because the difference between 36 inches and whiteout and clear blue sky is undeniable. The exercise comes when students are 8 weeks into the class. The class is an AMS based class, which has already been described well in this workshop by Julie Snow from Slippery Rock. Our class is given in the fall semester and lake effect snow starts in October and is quite an issue in forecasts until April. The skills of a forecaster are tested, and you cannot use forecasts from nearby areas reliably. Finally, we live in a fantastic snow belt, so lake effect snow happens a lot. In a good year we get over 300 inches of snow, mostly at times that places nearby do not. You can drive to Houghton in the bright sun and be met by a wall of very active blizzard just a few miles out of town.
There are some excellent tutorials available from COMET, and outreach of the National Weather Service. I use one done by Greg Byrd, which is available online or in a power point format. There are a number of things that must be learned before forecasting. These include some fluid dynamics of plumes, latent heat, remote sensing, upper air mapping, and the use of models. We cannot cover all them completely. I try to introduce all these things and give people entry points into the juicy parts of these topics, but do not expect students to understand completely. One thing you can spend a long time on are the satellite images. Here is one, just to whet your interest: http://serc.carleton.edu/details/images/13586.html
I have the students make a list of the critical parameters they think might be needed for a successful lake effect forecast. This is a challenge to prepare, but the idea is to include things that are even marginally useful and to collect data to see what is most important. We get a list of parameters like this:
This list is pretty good, but deliberately not complete, and we encourage students to add other things they think might be important. The next step is to find where you can get this information. I have web data sources for most (see below), and some of them are interrelated. You can do this exercise for any site around Lake Superior or probably many other lakes as well. For specific sites, the fetch length, upstream lake and opposing bay information are obtainable directly from the wind direction if you have a good map (Google Earth). So a spreadsheet for parameters related to wind direction can be prepared in advance and these parameters can be immediately available from the wind direction. Nonetheless the issue of sources for all this stuff must be addressed in an effort that spans several hours. The use of models is needed to look into the future where possible.
Once students know what they are looking for and how to find it, the exercise starts its data collection. Every day or every 6 or 12 hours beginning when conditions get close to "LES favorable" students collect information on these LES predictors. They also make LES forecasts for each period and include that information in the spreadsheet. The next day the real snowfall data is added to the spreadsheet, and this can be used as validation data for the forecast. This data collection needs to be done for several weeks (November and December in my case, usually a good time for LES).
The data analysis is the most challenging part. Spreadsheet plots which test the sensitivity of various parameters singly and together are possible. There is a lot of sophistication possible if there is enough LES to analyze. Overall, results should be a good experience with imperfect data addressed to a real-time problem. Models and real data, remote sensing, and balloons are all integrated and there are quite obvious weaknesses.
On the final day of class student groups will compete by doing forecasting which employs the LES techniques. This might reflect the most recent snow event. A more important element of this submission will be their evaluation of LES prediction parameters. Not only do we consider the actual forecast, but we discuss which parameters were successful? Which are inconclusive? What suggestions for improved forecasts are possible from the experience? The format of this will be short presentations with time for discussion.
There are some excellent tutorials available from COMET, and outreach of the National Weather Service. I use one done by Greg Byrd, which is available online or in a power point format. There are a number of things that must be learned before forecasting. These include some fluid dynamics of plumes, latent heat, remote sensing, upper air mapping, and the use of models. We cannot cover all them completely. I try to introduce all these things and give people entry points into the juicy parts of these topics, but do not expect students to understand completely. One thing you can spend a long time on are the satellite images. Here is one, just to whet your interest: http://serc.carleton.edu/details/images/13586.html
I have the students make a list of the critical parameters they think might be needed for a successful lake effect forecast. This is a challenge to prepare, but the idea is to include things that are even marginally useful and to collect data to see what is most important. We get a list of parameters like this:
850 mb wind direction
850 mb temperature
Lake Superior surface temperature
fetch length
opposing bay?
Inversion layer height
topographic lift factor
wind shear evidence
upstream lake
upstream moisture factor
snow/ice cover issues
This list is pretty good, but deliberately not complete, and we encourage students to add other things they think might be important. The next step is to find where you can get this information. I have web data sources for most (see below), and some of them are interrelated. You can do this exercise for any site around Lake Superior or probably many other lakes as well. For specific sites, the fetch length, upstream lake and opposing bay information are obtainable directly from the wind direction if you have a good map (Google Earth). So a spreadsheet for parameters related to wind direction can be prepared in advance and these parameters can be immediately available from the wind direction. Nonetheless the issue of sources for all this stuff must be addressed in an effort that spans several hours. The use of models is needed to look into the future where possible.
Once students know what they are looking for and how to find it, the exercise starts its data collection. Every day or every 6 or 12 hours beginning when conditions get close to "LES favorable" students collect information on these LES predictors. They also make LES forecasts for each period and include that information in the spreadsheet. The next day the real snowfall data is added to the spreadsheet, and this can be used as validation data for the forecast. This data collection needs to be done for several weeks (November and December in my case, usually a good time for LES).
The data analysis is the most challenging part. Spreadsheet plots which test the sensitivity of various parameters singly and together are possible. There is a lot of sophistication possible if there is enough LES to analyze. Overall, results should be a good experience with imperfect data addressed to a real-time problem. Models and real data, remote sensing, and balloons are all integrated and there are quite obvious weaknesses.
On the final day of class student groups will compete by doing forecasting which employs the LES techniques. This might reflect the most recent snow event. A more important element of this submission will be their evaluation of LES prediction parameters. Not only do we consider the actual forecast, but we discuss which parameters were successful? Which are inconclusive? What suggestions for improved forecasts are possible from the experience? The format of this will be short presentations with time for discussion.
Determining whether students have met the goals
After the student groups present their data I will ask students to submit a forecast discussion, which will be patterned after NWS discussions, which can be observed online every day of the year and which should be familiar to all students. In this case students will describe the information they found to be critical in making their forecasts. This should help us to evaluate what they learned about the process.
More information about assessment tools and techniques.Teaching materials and tips
Other Materials
- Power point about LES, Greg Byrd NWS (PowerPoint 2.3MB Oct22 08)