Brandon Orr · Web viewThe Weis Markets store network lies in the circled area, noted as the most dangerous area east of the Mississippi River by the FMCSA because of the high number

A customized, user-friendly weather index predicting grocery store use, designed for Weis Markets.

Michael PageBrandon Orr

Anna SchneiderKarily Villanueva

Pe r i o d i c R e p o r t On F i n a n c i a l l y Im p a c tf u l Th r e a t s

Decision Dynamics

Decision Dynamics is a company that serves businesses around the United States and its territories by contributing in weather event decisions. It was established in 2012 by four Meteorology graduates from The Pennsylvania State University. Our vast, combined experience of over 16 years in the weather field makes Decision Dynamics the company that will guarantee your business efficiency, accuracy, and profit in weather related circumstances through our forecasting knowledge and communication skills.

Weis Markets

Weis Markets was founded in 1912 by Harry and Sigmund Weis in the state of Pennsylvania. Currently, it operates in five states in the Mid-Atlantic region of the United States: Pennsylvania, New York, New Jersey, Maryland, and West Virginia. Its goal is to provide customers with the best product quality and shopping experience through eco-friendly and low-cost products. Weis Markets believes in buying local products and it is one of the largest purchasers of local products within its five-state region.

Our Mission

Decision Dynamics serves to fill the void between meteorology and businesses. Weather is a critical component of many companies, and if it goes overlooked can cost them millions of dollars. In today’s economy, saving money is crucial to the survival of a company. One of Decision Dynamic’s many goals is to help companies, like Weis Markets, make wise business decisions to give them the extra edge over their competitors and save money. We work personally with our clients to develop a product that will prevent the weather from becoming a financial burden and use it as an opportunity to make money.

Impacts of Weather on Weis Markets

The impact of weather on businesses is immense, especially for the grocery industry. Many people base their shopping habits on the weather. Business skyrockets in advance of a forecasted weather event, whether it is a winter storm, tropical system, or flooding rains. The PROFIT index can predict the onset of these events, and the influx of paying customers that come along with it. If the store is not prepared, extra customers can cause staffing problems and leave managers with headaches trying to keep the store stocked. Impending weather systems can also wreak havoc on the transportation sector of grocery stores. Weis Markets employs numerous tractor trailers to distribute products in a timely manner to reduce spoilage. Weather not only alters the routes of the trucks, but can cause costly accidents. These are few of the many impacts weather has on the grocery industry, and the Decision Dynamic’s PROFIT Index, can help companies like Weis Markets better handle these events.

2 | P a g e

Weather Prediction

Virtually every aspect of life is impacted by the weather. Some impacts are trivial, such as the decision to carry an umbrella on a rainy day, while others are far reaching. Companies, for example, can make or lose millions of dollars in one weather event. As such, it is critical for weather forecasts to be as accurate as possible.

Weather prediction began thousands of years ago in early civilizations. Initially, farmers, sailors, and others intimately familiar with the outdoors used recurring astronomical and meteorological events to help predict future conditions.

During the nineteenth and twentieth centuries, regional and global meteorological networks became common, providing more reliable observations from which forecasts could be made.Today, those networks have been developed even further. The United States relies on its Automated Surface Observing System (ASOS) for most ground based observations. The surface data is augmented by upper-air observations that are taken twice daily at 92 sites across the country. During each observation, a weather balloon is released into the atmosphere with instruments that send back information about temperature, pressure, and humidity among other variables. Finally, satellites are also used to acquire data from parts of the globe where surface and upper-air data is sparse, such as over the oceans. The first United States weather satellite was launched into space in 1959.

With such a tremendous amount of data, it is impossible for a human to create a forecast without the assistance of technology. For this, sophisticated computer programs are run to model the atmosphere over time.

These models provide an objective forecast of the future state of the atmosphere by solving a set of equations that describe the evolution of variables important in meteorology such as temperature, pressure, and wind. The starting point for each model is the initial state of the atmosphere, as conveyed by observations. It is critical that the initial observations are accurate to ensure that the model can display future conditions with as few errors as possible. Still, errors tend to compound with each time step, making the model less accurate further out in the forecast period.

While meteorologists run and interpret dozens of different computer models, each model uses similar methods to produce a forecast. For example, all numerical models are based on the same set of governing equations. The difference in each model is subtle, and may result in different approximations or assumptions made in each equation.

While computer models are an invaluable asset to meteorologists, it is still the meteorologists’ responsibility to review each prognostication and assimilate that data into one human approved forecast.

3 | P a g e

Forecast Verification

With so much confidence placed into computer models when forecasting the weather, one has to wonder just how accurate they are. Since most computer models are run by governmental organizations, verification statistics are often updated and reviewed.

In the Northern Hemisphere, the most accurate computer model is regularly the European Centre for Medium-Range Weather Forecasts (ECMWF). The European Centre, which runs the model, is an intergovernmental organization supported by 34 European countries. The second most accurate forecast model is the UKMET, run by the United Kingdom’s National Weather Service known as the Met Office. The third most reliable model is the Global Forecast System (GFS) run by the United States’ National Oceanic and Atmospheric Administration (NOAA).

The accuracy of these three models was compared by NOAA scientists. The scientists analyzed each model’s forecast of 500 millibar (mb) heights and compared it to the actual outcome. The 500mb height refers to a level approximately 16,000 feet above the surface, at which the pressure is 500mb. Meteorologists often look at this 500mb level to gauge where surface highs and lows may reside based on the jet stream, or primary storm track, position at 500 mb.

Figure 1 shows the verification plot, shown as a monthly mean anomaly correlation for the five-day forecasts of 500mb heights. This value essentially compares a forecast to actual observed weather conditions, giving a sense of how much skill is involved with forecasting. A value of 1 indicates a perfect forecast, while a value of 0.6 or higher indicates a reliable forecast. In this case, ‘reliable’ indicates that by using the forecast, there is some added value to the user. The graphs below generally feature a ‘saw tooth’ shape, on average indicating more accurate forecasts during the winter than during the summer. The more notable contrasts in things like temperature and pressure during the winter, in addition to the larger scale weather features, make the conditions easier to model.

Accuracy of Forecast Models as Compared to Observed Conditions at ~5,500 metersIn the Northern Hemisphere

Figure 1: Model verification at 500mb level

There is a clear upward trend for each of the three models from the mid-1980s through the early 2010s. The American GFS model has shown the most improvement, rising from 0.55 in 1985 to

4 | P a g e

about 0.89 in 2011. Still, the ECMWF model remains most reliable at about 0.93.

The improvement in forecasts makes sense, given that the National Weather Service spent $4.5 billion to modernize equipment and improve forecasting over this time span. To show how accuracy would suffer without that investment, the government included the Climate Data Assimilation System (CDAS) model in purple. The CDAS is essentially a model frozen in time, giving a sense of how a model with 1980s technology would perform today. It underperforms extremely when compared to models fitted with modern technology, as one might expect.

NOAA says that today’s 3-4 day forecasts are as accurate as a 2-day forecast was 15 years ago. It goes on to say that predictions of rain 3 days ahead of time are as accurate as a 1-day forecast of rain was in the mid-1980s.

Figure 2 proves that point by looking at ECMWF model data dating back to 1981. Today, a 7-day forecast at 500hPa (same as 500mb) boasts an anomaly correlation of 75. That’s roughly the same as a 5-day forecast was in 1986. Likewise, today’s 5-day forecast is about as good as a 3-day forecast was back in the 1980s.

Figure 2: ECMWF forecast improvement since 1981

How Decision Dynamics Forecasts

At Decision Dynamics, we take forecasting very seriously. We invest in the most up-to-date technology to ensure our clients are obtaining the best forecast possible.

Before creating a forecast, each meteorologist thoroughly reviews current weather conditions locally, nationally, and globally to gain a broad perspective of the weather picture. Then, a comprehensive review of both numerical and statistical models takes place. Our meteorologists collaborate and discuss any discrepancies in the models before locking in on a final prognostication.

Finally, our own computers ingest the necessary variables to produce the PROFIT Index, a proprietary blend of weather variables designed to save your company money.

5 | P a g e

Extensive research was done by Decision Dynamics into what types of weather events people are likely to prepare for, and when they are likely to stock up for a weather event. Based on this research, as well as our extensive knowledge of the weather in the Mid-Atlantic region of the United States, we have designed an index that will save you money. We have designed this index to react to specific weather events that are likely to impact your business economically.

The PROFIT (Periodic Report On Financially Impactful Threats) Index will serve as a way to assess how the upcoming weather may affect business at Weis Markets. The idea would be to have a program that would pick an event type and then run through the variables of the weather event to get an assigned sales impact value of 0 through 10. This factor, along with other variables for each weather event, will be used to get a final value between 0 and 10, with 0 being no impact and 10 being an extreme impact to sales.

A color coded map overlaid with the locations of Weis Markets around the Mid-Atlantic region will be provided. The user would then choose a store location, and an individual index value will show up for the selected store location. Next to the assigned value, there will be an explanation of what type of weather event is likely to occur and how severe it may be. Higher values indicate more severe events, and therefore they are more likely to increase profit. The more extreme the event, the more people will need to prepare for it. For this, the PROFIT index will extend out to seven days and will be updated daily as the weather models update.

Weather Variables

The PROFIT index will warn you of significant Lake Effect Snow, Nor’easters, Excessive Heat, Flooding, and Tropical Storms and Hurricanes. Our research has shown that these events are the most likely to affect Weis store locations.

6 | P a g e

Pe r i o d i c R e p o r t On F i n a n c i a l l y Im p a c tf u l Th r e a t s

        Figure 3 shows some of the natural disasters that impact various regions in the United States

Lake Effect Snow

Given the locations of Weis Markets, business is directly affected by Lake Effect Snow events. Lake Effect Snow forms when cold air flows over a relatively warm body of water. In this case, The Great Lakes would be the water source to help produce this snow event. There are many factors that go into forming lake effect snow, and Decision Dynamics has taken those into account. The variables we have considered are: the temperature difference between the lake and the air, fetch, directional shear, wind speed, forecasted snowfall, probability of precipitation, and the day the storm will hit.

The temperature difference between one of The Great Lakes and the 850mb layer (approximately 5000 feet above the ground) of the atmosphere is very important. With a temperature difference of less than 13 degrees Celsius, no snow event involving the lake will occur. This number is based on extensive research done by fellow meteorologists. The higher the temperature difference, the stronger the event is likely to be, having 13 as a minimum threshold.

Fetch, which is how long the stream of cool air spends over water, is also a very important variable that we have taken into consideration. This value is generally measured in kilometers, and has to do with air flowing over the lake. The more time the wind flows over the water, the more moisture it will gather. Generally, if the stream of cool air passes over the length of the lake rather than its width, it will generate a stronger storm.

How much the wind direction changes with height, known as directional shear, is important for determining the organization and strength of the snow bands. Too much shear causes the snow bands to become disorganized and will produce little if any snow. Directional shear is measured in degrees; shear of less than 30 degrees is ideal, while greater than 60 degrees is detrimental to the development of Lake Effect Snow.

Wind speed at about 5000 feet is another important variable that needs to be considered when forecasting lake effect snow. Stronger winds, usually between seven to fifteen meters per second,

7 | P a g e

produce more intense bands of Lake Effect Snow.

All of the variables above are extremely important when it comes to knowing how intense a Lake Effect Snow event may be. However, the possible intensity of the storm is not the only variable we have considered when making the index. Decision Dynamics has also taken into account how much snow is forecasted for each location. The average amount of snowfall each region receives is also taken into account and factored into the forecasted snowfall amount. This is important because certain areas are more accustomed to snow than others. For example, the average snowfall amount for December in central Maryland is 3.1 inches, while the average snowfall amount in Binghamton, New York is 18 inches, according to the National Climatic Data Center. This difference will be reflected in how people react to snow storms and is factored into the PROFIT index for your benefit.

The probability of precipitation and the day the storm hits are also considered. Our research has shown that profits will either increase or decrease depending on the day a storm hits. A storm impacting on a Saturday, Sunday, or Monday has brought the most profit. These factors combined go into creating an index value for Lake Effect Snow.

Nor’easters

Nor’easters are snow storms that tend to form in the Gulf of Mexico or off the Atlantic Cost and travel up the eastern seaboard. These types of snow storms are also very likely to impact business at Weis Markets and are factored into the PROFIT index. Several variables help determine the strength of Nor’easters. We have accounted for the following: the lowest observed pressure, wind speed, air temperature, current snowfall, forecasted amount of snow or ice, probability of precipitation, and day of impact.

The lowest pressure observed at the center of a low is very important. It is helpful to think about it as if it were a hurricane. The lower the central pressure, the stronger the storm will be. For example, the infamous March 1993 blizzard had a central low pressure of 960mb. To put this in perspective, the central low pressure of Hurricane Irene was 942mb. These are very low pressures considering the average sea-level pressure is 1013mb.

The wind speed the storm produces is also taken into consideration. Often the wind speed of a snow storm will make it all the more dangerous. The stronger the winds, the likelihood for downed electric wires and power outages can increase. These factors could be a drain on resources, or a valuable business opportunity for Weis Markets if armed with the proper information.

Air temperature inside the storm is important for determining the strength of Nor’easters as well. Temperatures above 32 degrees Fahrenheit in the storm would produce other kinds of precipitation that are not as likely to cause damage or panic. Temperatures at or below 32 degrees Fahrenheit will produce snow and ice. These are the conditions that will likely have people running to the store for supplies. Ice will be accounted for separately because even very small accumulations (less of ¼ of an inch) can have a significant effect.

8 | P a g e

The current snowfall being produced by the storm is also important. If a storm left behind a foot of snow in one location and is headed your way that is information that will become valuable.

Similarly to Lake Effect Snow, the Nor’easter part of the index will include the forecasted snowfall amount, the mean snowfall for each region, the probability of precipitation, and the day the storm will hit each location.

Excessive Heat

Excessive heat can be very dangerous. Too many consecutive days with above average temperatures and relative humidity can have an impact on your business. With everyone trying to keep cool, ice and water would become top sellers in a heat wave. Two variables that are especially important are the departure from the average temperature and the relative humidity.

It is important to know not only how warm it is but also to know whether or not that is normal. For example, what may be considered hot in New York would not be considered hot in Maryland. It is this distinction that made us use the departure from average temperature rather than just the high temperature when creating this index.

In addition to the temperature, the relative humidity is also very important during a heat wave. The relative humidity represents how much moisture is in the air relative to the temperature. Our bodies cool by evaporating sweat that perspires from our skin. When the relative humidity is very high, our sweat will not evaporate and our bodies are unable to cool. This combined with the high temperature are what go into the heat index, and are what we will use to determine the risk of excessive heating. Because our research has shown that the day of impact is also important for your business, we have factored that in the PROFIT index as well for all event types.

Flooding

Flooding can be the most dangerous part of a storm, causing extensive damage and sometimes occurring with little to no warning. Flooding becomes more likely as the rainfall rate and the duration of rainfall both increase above a certain threshold. It is important to note that there are many other factors that contribute to potential flooding that are not meteorologically related. Once an event becomes more imminent, we will provide National Weather Service flood warnings for Weis Market locations.

The rainfall rate is very important because when significant rain falls in a very short amount of time, the ground does not have a chance to absorb it. This increases run-off, and causes rain to build up in certain areas causing dangerous flooding. Generally, when the rainfall rate is at or exceeds an inch per hour, flooding is likely to happen. However, this will vary for each location and depends upon how dry or wet the soil is before the event.

Duration of rainfall in a storm also influences flooding. The longer it rains, the more rain will accumulate and cause flooding. Generally, heavy rain for an hour or more could potentially cause flooding to occur.

9 | P a g e

Since these factors vary by location, we have taken this into account. The Hydrometeorological Prediction Center will predict flood stages for nearby water sources about four days in advance. At this point, the PROFIT index will take this forecast into account as it will become very important if heavy, long lasting rainfall is predicted. The website shows the following map with flood locations and will forecast a flood stage for each area that will go into the index.

Figure 4 shows an image of river forecasts

In addition to the above variables, the probability of precipitation was also taken into account when building the PROFIT index. It is always important to know how likely an event is to occur, which is why this variable is factored in.

Tropical Storms/Hurricanes

Despite being in the Mid-Atlantic region, this area is still vulnerable to Tropical Storms and Hurricanes, as we saw recently with hurricane Irene in 2011. It is because of this risk that Decision Dynamics has included Tropical Storms and Hurricanes into the PROFIT index.

The variables taken into account are whether a watch or warning is issued, and whether or not a location lies within the cone of uncertainty. This cone shows the uncertainty of the predicted path. It is represented as a cone shaped zone that widens with time and distance from the current location of the center of the Tropical Storm or Hurricane. The cone of uncertainty will become narrower as the storm approaches, showing more certainty of the storm’s path. It is also very important because it shows the projected path and updates as the storm nears.

A watch within the context of meteorology means that conditions are favorable for an event to

10 | P a g e

occur in a given area. A warning means the event is imminent or already occurring. When one of these is issued for a Tropical Storm or Hurricane, our research has shown that the majority of people will start preparing for the storm. This would be a very good business opportunity for Weis Markets and is why we have chosen to include it in the index.

How Index Values Are Calculated

The program will first be presented with a list of weather event types, and based on the forecast will choose one of these events. Once an event type has been chosen by the program, it will run through the variables for that event and will calculate an intensity value between 0 and 10. This is done by assigning certain values in each variable a ranking. These rankings are then weighted based on their importance and are added together to give an intensity value. Once this value has been obtained, the other variables in each category are also weighted, such as the day of impact and probability of precipitation, and are then added together to obtain a final index value between 0 and 10.

Lake Effect Snow

Table 1 shows how the previously mentioned variables for Lake Effect Snow were ranked to obtain an intensity value. The variables in the following equation represent the ones in the table below and are labeled A through D in order from left to right, and the equation will calculate a value between 0 and 10. Should any value in the below table rank at 0, the intensity ranking will be given a zero.

Lake Effect SnowTemperature Difference

Between the Lake and

850mb (˚C)

Variable Ranking

Fetch (km)

Variable Ranking

Directional Shear

(Degrees)

Variable Ranking

Wind Speed at 850mb (m/s)

Variable Ranking

T<13 0 5-20 1 >60 0 <5 213<T<14 3 21-65 5 30<S<60 5 5<W<7 514<T<16 6 66-75 7 <30 10 >8 1017<T<19 7 >75 10

T>20 10 Table 1 shows variables and assigned rankings used to evaluate the intensity of the Lake Effect Snow

The lake temperatures will be obtained from NOAA’s Coastwatch website and will be updated daily. Air temperature, wind direction, and wind speed can be taken from short range forecast models and atmospheric soundings. The long range National Center for Environmental Prediction (NCEP) models will be used to obtain values beyond four days. The fetch will be determined by the air’s motion over one of The Great Lakes and can be obtained by simply looking at the length or width of the water source.

Event Intensity=A(.25)+B(.25)+C(.3)+D(.2)

11 | P a g e

The variable rankings in the table are values of 0 through 10. These values are based on whether or not the range of values for that variable will contribute to the strength of the storm. A variable is given a zero when the range of values makes it impossible for the event to occur. Lower rankings are given when the range of values in a certain variable will contribute to a storm that will not be severe enough to affect potential profit at Weis Markets.

The highest rankings are for values that will contribute to significant Lake Effect Snow events. These variable ranges are based on previous meteorological research. The weightings that go into this portion of the equation are also based on their relative importance for initiating and maintaining significant Lake Effect Snow. They are all weighted relatively close because these variables together are what make this event possible. The wind speed is given the lowest weighting because a low value can be overcome by the other variables.

Other variables that are not included in the above table are accounted for separately. The value that is calculated from the table is used to assess the storm’s possible intensity. The intensity of the storm is treated as its own variable that will go into the final index equation for Lake Effect Snow.

As was discussed previously, the other variables that will go into the final equation are the forecasted amount of snow, the probability of precipitation, and the day of impact.

The forecasted amount of snow is broken down in the following way: forecasted snowfall - .5(monthly average snowfall). This is done because what is considered a major snow storm in one area is not considered a major snowstorm in another. This will also allow for negative values. For example, the average monthly snowfall for Binghamton, New York in December is eighteen inches. If six inches of snowfall is forecasted, this will not be a major event for the area. We have chosen to cut the monthly value in half because receiving half of an area’s expected snowfall amount in one storm can be considered significant. For the example given, the forecast value will be a -3 and will lower the final value of the PROFIT index. This will also reflect the relative significance of a major snowstorm giving a higher final index value for areas that are not used to so much snow.

The probability of precipitation will be given in percent and divided by ten in order to obtain a value between 0 and 10 to match the index. For example, an 80% chance of precipitation will be given a value of 8. This is weighted slightly less than the other variables because the probability of precipitation will change greatly as the event approaches. When a forecast is made ten days in advance, this variable will be less reliable. However, as the weather event nears, it will become more accurate and is why it is ranked closely but slightly lower than the other variables.

The intensity of the storm is ranked highly because of the impact a strong storm will have on business. However, we have decided that it will be slightly less valuable than the forecasted snowfall amount. We have chosen to do this because even if the forecast ends up being incorrect, if a severe snow storm is forecasted, people will still feel the need to be prepared and this will positively impact business.

12 | P a g e

0-8 1 <15 1 >35 1 1<x<3 19<x<13 3 15-25 3 32-34 4 4<x<6 514-19 5 26-35 5 25-31 7 7<x<11 820-25 7 36-45 8 20-24 9 >12 1026-30 8 >45 10 <19 1031-40 9>40 10

Variable Rankings

Nor'easterLowest Pressure (1013mb-p)

Variable Ranking

Wind Speed (m/s)

Variable Ranking

Air Temperature

Variable Ranking

Current Snowfall

The final variable taken into account is the day the storm hits. It has been found, based on a conversation with a Weis store manager, that the day a storm hits will affect sales positively. Storms hitting during the days of Saturday, Sunday, and Monday, are the most valuable. This will be given the lowest weighting in the final equation because regardless of what day the storm hits, there will still be an impact on sales even if it is not as extreme. Thus, we have assigned each weekday a ranking of 0 through 10 that will be factored into the final equation. We have assigned the days as follows:

Day of impact:No event:  0Tuesday-Thursday:  5Friday:   8Saturday-Monday:   10

The final equation (highlighted below) will provide a value of 0 through 10.

Lake Effect Snow = Event Intensity (.3)+Forecast(.35)+Precipitation(.25)+Day of Impact (.1)

The decimal values next to each variable represent the relative importance of that variable. Event intensity is 30% of the final index value, the forecast is 35%, probability of precipitation is 25%, and the day of impact is 10%.

Nor’easters

Table 2A below shows how the previously mentioned variables for Nor’easters were ranked to obtain an intensity value. The variables in the equation below represent the ones in the table below and are labeled A through D in order from left to right, and the equation will calculate a value between 1 and 10.

Table 2A shows variables and assigned rankings used to evaluate the intensity of the Nor’easter.

Event Intensity = A(.3)+B(.25)+C(.2)+D(.25)

The pressure at the center of the low can be obtained from the GFS model in the short term and from medium range NCEP models beyond four days. Wind speed, air temperature, and snowfall

13 | P a g e

in the short term can be determined from surface maps, while longer term will be determined through medium range models.

The variable rankings in the table are values of 1 through 10. These values are based on whether or not the range of values for that variable will contribute to the strength of the storm. A variable is given a lower ranking when the range of values in a certain variable will contribute to a storm that will not be severe enough to affect potential profits at Weis Markets. The highest rankings are for values that will contribute to a significant Nor’easter. These variable ranges are based on previous meteorological research into this weather event.

The weightings that go into this portion of the equation are based on their relative importance for strong Nor’easters. They are all weighted relatively closely because it is all of these variables together that make a strong Nor’easter. The lowest pressure is ranked slightly higher because the stronger this storm is, the lower its pressure it has, and that is extremely important for significant snow events. The air temperature is ranked slightly lower because if the other variables are very high, then a strong storm is likely to occur.

As was discussed previously, the other variables that will go into the final equation are the forecasted amount of snow, the probability of precipitation, and the day of impact. The forecasted amount of snow in this scenario will be substituted with a forecasted amount of ice accumulations should that occur instead of snow.

Forecasted Amount of Ice (inches)

Variable Ranking

<.1 0.1-.25 4.26-.75 7.76-1 8

>1 10Table 2B shows the rankings for forecasted amounts of ice.

It is generally considered that an ice storm has occurred when the storm produces a quarter of an inch of ice or more. We have decided to give a moderate ranking for above a tenth of an inch because this will still affect travel, and therefore will affect the transportation of products to Weis Markets. Above a quarter of an inch is considered significant and is reflected in the above table. These rankings will be substituted in for the forecast in the final equation (highlighted below) should the forecast call for ice. Should the forecast call for both ice and snow, the rankings above will be averaged with the value calculated for the forecasted snowfall to obtain a final ranking. Should there be no snow produced by the storm, the current snowfall value in table 2A will be replaced with a current ice accumulation and will be ranked the same as in table 2B.

The day of impact is broken down the same way as it was for Lake Effect Snow. The probability of precipitation will be given in a percent and divided by ten in order to obtain a value between 0 and 10 to match the index. For example, an 80% chance of precipitation will be given a value of 8. The final equation will provide a value of 1 through 10.

14 | P a g e

Nor’easter = Event Intensity(.3) + Forecast(.35) + Precipitation(.25) + day of impact(.1)

The decimal values next to each variable represent the relative importance of that variable. Event intensity is 30% of the final index value, the forecast is 35%, probability of precipitation is 25%, and the day of impact is 10%. The reasons for these weightings are the same as for Lake Effect Snow. The type of snow storm does not change how it will be evaluated in the index.

Excessive Heat

Table 3 shows how the previously mentioned variables for excessive heat were ranked to obtain an intensity value. The variables in the following equation represent the ones in the table and are labeled A through B in order from left to right, and the equation will calculate a value between 1 and 8. Our research has shown that fewer people will react to excessive heat than other weather events. Therefore the index will account for an extreme case of excessive heat, and treat most other cases as moderate or weak since there is still a chance for some profit during a heat wave, but not as much as for another event.

Excessive HeatDeparture from

Average Temperature (˚F) Over 3 Days

Variable Ranking Relative Humidity Over 3 Days (%)

Variable Ranking

<5 0 <40 16-8 3 40-60 49-11 5 61-80 512-15 7 >80 7>15 8    

Table 3 shows variables and rankings for excessive heat

The values in the table above are ranked based on how much each range of values will affect the intensity of the event, with lower rankings indicating a lower impact. It is generally considered significant when the temperature is 10 degrees above normal.

The temperature and relative humidity can be obtained from MOS and GFS models in the short term. For beyond four days, the NCEP long range model will be consulted.

Departure from average temperature and relative humidity are both averaged over three days in the table. We have chosen to do this in order to show that it is representing an extended period of excessive heat rather than just one abnormal day. These two variables are given an equal weighting because they both have a very strong effect on heating, or rather both will make it very difficult to cool down.

Event Intensity= A(.5)+B(.5)

The other variable taken into account for this event is the day of impact. The day of impact is broken down the same way as it is for lake effect snow and nor’easters. It is given a lower

15 | P a g e

weighting because the other variables are far more important when considering the intensity and possibility of excessive heating.

Excessive Heat = Event Intensity(.9) + Day of Impact(.1)

Flooding

Table 4 shows how the previously mentioned variables for flooding were ranked. The variables in the following equation represent the ones in the table and are labeled A through C in order from left to right, and the equation will calculate a value between 0 and 10. The variables are ranked in terms of their contribution for a flood. A ranking of 0 means a flood is very unlikely to occur for the given range of values. If the rainfall rate is above an inch per hour and lasts for more than an hour, flooding is likely to occur.

Flooding

Average Rainfall Rate (in/hr)

Average

Rainfall Rate

(in/hr)

Average Rainfall

Rate (in/hr)

Average

Rainfall Rate

(in/hr)

Average Rainfall

Rate (in/hr)

Average Rainfall

Rate (in/hr)

<.7 <.7 <.7 <.7 <.7 <.7.7-.9 .7-.9 .7-.9 .7-.9 .7-.9 .7-.91-2 1-2 1-2 1-2 1-2 1-2>2 >2 >2 >2 >2 >2

Table 4 shows variables and rankings for flooding

The flood stage variable would be specific for any river or water source closest to the desired location. This variable would be found on the Hydrometeorological Prediction Center’s website and updated daily. This variable is only valid less than four days before a potential event. With more than four days, the other variables would be gathered from long range NCEP models. With less than four days, the variables will be gathered from short range ensemble forecast models and short range GFS models.

It is important to note that there are many other factors that contribute to potential flooding that are not meteorologically related. Once an event becomes more imminent, Decision Dynamics will provide you with National Weather Service warnings as well as more specific information from the Hydrometeorological Prediction Center.

Long Range Flooding (5-7 days out) = A(.4) + B(.3) + Precipitation(.2) + Day of Impact (.1)

Short Range Flooding (4 days or less) = A(.25) + B(.2) + C(.25) + Precipitation(.2) + Day of Impact (.1)

This part of the index has been separated into two equations because the flood stage value cannot be predicted more than four days out, but it will be very important as the event draws closer. For more than four days, the rainfall rate and the duration of rainfall are given the highest weightings

16 | P a g e

because we cannot use the flood stage variable. Once we reach four days, the variables are weighted almost the same.

Tropical Storm/Hurricane

Table 5 shows how the previously mentioned variables for a Tropical Storm or Hurricane were ranked to obtain an intensity value. The variables in the following equation represent the ones in the table below and are labeled A through B in order from left to right, and the equation will calculate a value between 0 and 10. Our research has shown that the majority of people will react to a warning or watch for a Tropical Storm or Hurricane. Since warnings are more serious than watches, we have ranked warnings higher than watches, but watches are still considered extremely important and are ranked accordingly. The cone of uncertainty is ranked in terms of whether or not a location falls within the cone. The cone of uncertainty locations can be gathered from the National Hurricane Center’s website, and the watches or warnings will be issued by the National Weather Service. This will only go out five days and will update several times every day up to the event.

Tropical Storm/Hurricane

Watch/Warning Watch/Warning Watch/Warning Watch/Warning

None Issued None Issued None Issued None IssuedWatch Watch Watch Watch

Warning Warning Warning Warning Table 5 shows variables and rankings for a tropical storm/hurricane

Event Intensity=A(.5)+B(.5)

The two variables are given equal weightings because they are both equally important when looking into the threat of a tropical storm or hurricane. The final equation will also include the day of impact, and is broken down as it is for all of the other event types. The day of impact is given a much lower ranking because our research has shown that this event is something most people will be cautious of.

Tropical Storm/ Hurricane = Event Intensity(.9) + Day of Impact (.1)

Interpretation of the PROFIT Index

Each color on the index map shown on Table 6 will represent an intensity value for a particular Weis Market location. The colors and values on the map are to be interpreted as follows:

17 | P a g e

PROFIT IndexWeather Impact

No Impact 0-2Weak 3-4

Moderate 5-6High 7-9

Extreme 10 Table 6 is a legend for the PROFIT Index map below (Figure 5)

Once the user selects a location, an individual index value 0 through 10 will be displayed for that location. An explanation of what that value means, what event type the value is for, and when it is expected to occur will also be displayed. Green on the map, values 0 to 2, indicates that there is either no weather event expected over the next ten days or it is one so mild it is not expected to have a material effect on people. Yellow on the map, values 3 to 4, indicates the event will have a weak impact on sales. This could mean a light snowstorm, a weak heat wave, etc. Orange on the map, values 5 to 6 indicate a moderate impact on sales. This means the event is significant enough that there will potentially be an impact on sales, and supplies that may be specifically related to the event should be stocked. Red on the map, values 7 to 9 indicate a high impact on sales due to a severe weather event. This could mean a possible blizzard, tropical storm, extreme heat wave, or other major event that should be taken very seriously. Should the map be colored purple, it would indicate an extreme impact on sales and a very severe weather event is imminent. This value is reserved for the most severe cases.

Maps similar to the one in Figure 5 will be emailed to Weis Markets daily starting with a map indicating a risk for the next ten days. Should an event be detected it will be followed up on from seven days out until the event occurs. For example, should Weis Markets receive the following map and select a store location, the index will give a value and an explanation for that location

Figure 5: Theoretical PROFIT Index map ahead of snow storm. Blue dots show Weis Markets locations

18 | P a g e

The index would read as follows:

“A PROFIT index value of 8 has been calculated for the York, Pennsylvania area. There is a high risk for a major nor’easter five days from today on January 10th. We are expecting strong winds and heavy snowfall amounts up to a foot of snow. Please take precautions when transporting goods near that location. You may want to stock up on necessary supplies before the event occurs. There will be an event update tomorrow. Please feel free to contact Decision Dynamics with any questions or concerns.”

PROFIT Index: February 5-6, 2010 Snowstorm

A strong Nor’easter slid past the East Coast on February 5-6, 2010 causing a major disruption in the Mid-Atlantic area with very heavy snowfall and strong winds. Assuming the storm was well forecasted, the following is a simulation of our index.

To create the maps, we must calculate the index for locations across the Weis Markets coverage area. Here, we will use Philadelphia, Pennsylvania and Williamsport, Pensylvania as an example.

Philadelphia, PA

The first variable in the index is the Event Intensity. This takes into account lowest sea level pressure of the storm, highest wind speed/gust, air temperature, and current snowfall.

Lowest pressure: 1003mbHighest wind speed/gust: 41mph (18.3 m/s)Mean air temperature: 25°FCurrent Snowfall: Greater than 12 inches

Table 7 shows an example of the selection of rankings for variables in nor’easter intensity

After we found the rankings for the various variables above, they are factored into the event intensity equation.

Event Intensity = A(.3)+B(.25)+C(.2)+D(.25)Event Intensity = 3(.3) + 3(.25) + 7(.2) + 10(.25)Event Intensity = 5.55

19 | P a g e

Already, this value is associated with a “moderate” category on the PROFIT Index. Before the final value is calculated, the forecasted snowfall, precipitation, and day of impact must be factored in.

Forecast: 28 inches of snow (subtract 0.5*6.6 to take out half of average monthly snow, which accounts for the area’s climatology)Precipitation: 100% chance (value of 10)Day of impact: Saturday (value of 10)

After the final variables are calculated, they are factored into the final equation.

Nor’easter = Event Intensity(.3) + Forecast(.35) + Precipitation(.25) + Day of Impact(.1)Nor’easter = 5.55(.3) + 24.7(.35) + 10(.25) + 10(.1)Nor’easter = 13.81

A value of 13.81 surpasses the threshold of 10 for the “extreme” category on the PROFIT Index. This corresponds well with the actual conditions. Philadelphia experienced well over their yearly total snowfall in this one storm, paralyzing the city for days.

Williamsport, PA

Lowest pressure: 1008mbHighest wind speed/gust: 22mph (9.8 m/s)Mean air temperature: 27°FCurrent Snowfall: Greater than 12 inches

Table 8 shows an example of the selection of rankings for variables in a nor’easter.

After we found the rankings for the various variables above, they are plugged into the event intensity equation.

Event Intensity = A(.3)+B(.25)+C(.2)+D(.25)Event Intensity = 1(.3) + 1(.25) + 7(.2) + 10(.25)Event Intensity = 4.45

20 | P a g e

Already, this value is associated with a “weak” category on the PROFIT Index (rounded down to a value of 4). Before the final value is calculated, the forecasted snowfall, precipitation, and day of impact must be factored in.

Forecast: 5 inches of snow (subtract 0.5*9.7 to take out half of average monthly snow, which accounts for the area’s climatology)Precipitation: 80% chance (value of 8)Day of impact: Saturday (value of 10)

After the final variables are calculated, they are factored into the final equation.

Nor’easter = Event Intensity(.3) + Forecast(.35) + Precipitation(.25) + Day of Impact(.1)Nor’easter = 4.45(.3) + 0.15(.35) + 8(.25) + 10(.1)Nor’easter = 4.39

A value of 4.39 was calculated, which is on the border of “weak” and “moderate”. This corresponds well with the actual conditions. Williamsport experienced a snowfall around 5 inches that was a disturbance, but given their climatology was not a significant storm.

Graphical Product

After the PROFIT Index values are automatically generated for sites across the Weis Markets coverage area, a contour map similar to Figure 6 will be created with all 161 locations layered on top. Each blue dot represents a Weis Market location, with larger dots indicating larger locations. This will allow for quick and easy viewing on the impacts of the impeding storm on every particular Weis location. Maps, updated daily, will be distributed on our private website as well as emailed.

21 | P a g e

Figure 6: The PROFIT Index’s graphical product for the February 5-6, 2010 winter storm during the Day 1-3 forecast period.

Forecast text for the Day 1-3 period:

A Nor’easter will be taking shape off the North Carolina coast, allowing snow to develop through Maryland by 10am-12pm, southern/central Pennsylvania by 2pm-4pm, and northern/northeastern Pennsylvania by 9pm-11pm on Friday, February 5. Eighteen inches or greater of snow accumulation is expected in most purple (“extreme”) regions, eight inches or more of snow in most red (“high”) regions, four inches or more of snow in most orange (“moderate”) regions, up to four inches in most yellow (“weak”) regions, and little to no accumulation in green (“no impact”) regions. Stores around State College, Pennsylvania and south of Allentown, Pennsylvania in the red and purple regions will need to prepare for a significant snowstorm, especially toward far southern Pennsylvania, Maryland, and West Virginia locations where snowfall is expected to exceed twenty inches. Scattered power outages can be expected in these regions. Wind gusts will be strong, particularly in Maryland where gusts to forty miles per hour are expected. Areas to the north, including Scranton, Pennsylvania and Binghamton, New York can expect little to no snowfall with relatively low PROFIT Index values.

This product would have allowed for efficient planning before the storm. Transportation could be avoided in the High and Extreme areas. Stores in a higher impact zone would be able to properly stock their shelves and schedule employees. Stores in lower impact zones would avoid costly preparation work. The ultimate goal of saving money and providing better customer service would be achieved.

22 | P a g e

Various Uses of PROFIT Index: Labor Costs

The PROFIT Index, which warns of hazardous weather up to seven days in advance, can be used to save on labor costs.

In each individual Weis Markets store, managers are responsible for scheduling staff. According to Matt Dixon, manager at the Weis Markets store on Martin Street in State College, Pennsylvania, his scheduling is typically done at least one week in advance. Once the scheduling is done, changes are rarely made. If a winter storm or other serious weather event is suddenly forecasted to strike in less than seven days, he is forced to add staff by paying employees overtime in many cases.

With the PROFIT Index, Matt and other managers would know within seven days if a storm is likely or at least possible, allowing him to better plan his staffing levels. With that extra information, overtime hours could largely be avoided.

The values presented in Table 9 are averages only, meaning that the expected savings of $220,634.40 per year may vary by year. However, the assumptions made in arriving at the averages are conservative, suggesting that more savings might be realized.

COSTAverage Hourly Wage for Weis Markets Employee1 $7.61Average Hourly Overtime Wage for Employee $11.42Average Number of Overtime Employees per Event per Store 5Average Hours of Overtime per Employee 8Overtime Cost per Store per Event $456.80Average Overtime Events per Store per Year 3Number of Stores 161Amount Saved with Decision Dynamics $220,634.40

Table 9: Potential labor cost savings with PROFIT Index

For example, the 46 largest stores in the Weis Markets network (55,000 to 70,000 square feet) may need more than five overtime employees ahead of a major weather event. Of course, the number of major weather events warranting overtime also varies greatly both by year and store. We used three to account for one major warm weather storm such as a flood or tropical system, and two winter storms.

The days on which these storms hit also would impact how much is saved in overtime. For example, Matt Dixon pointed out that storms hitting on a Saturday, Sunday, or Monday bring in the most extra business. As a result, even more staff would be required. That would increase the savings further.

Various Uses of PROFIT Index: Transportation of Goods 1 Obtained at GlassDoor.com

23 | P a g e

The PROFIT Index would not only help individual stores better staff major weather events, but it would aid in the dissemination of goods to various stores.

First of all, the visually appealing mapped version of our index values in Figures 5 and 6 will quickly give decision makers a means of viewing what regions of the network will be impacted by a major weather event. For example, if a major snow event is predicted in the Mid-Atlantic as we saw in 2010, the PROFIT Index values would show executives which stores should be better stocked for the event, and which travel routes will need to be avoided by distribution vehicles.

An independently organized survey of 100 people by Decision Dynamics shows that when the average shopper hears of a major weather event seven days away, they wait until two to three days before the event to stock up at the grocery store (Figure 7). With that in mind, the seven day PROFIT Index outlook would give corporate decision makers enough time to decide which stores should receive increased supplies of things like milk, bread, and eggs, as well as non-perishable items before the snow hits and shoppers raid the shelves.

11%

28%52%

9%

When Customers Stock Up(Days Before a Storm)

6 or 74 or 52 or 31

Figure 7: How many days before a storm customers stock-up

Furthermore, the PROFIT Index would give a clear indication of which regions will need to be avoided during and just after the storm to prevent accidents and limit transport delays.

According to the 2011 Weis Markets Press Release, the company owns and operates 131 tractors and 420 trailers. Their vehicles operate from the main distribution center in Milton, Pennsylvania. With a location just off of Interstate 80 (I-80), this roadway is a primary travel route for Weis Market vehicles. However, the Federal Motor Carrier Safety Administration (FMCSA) says it is also one of the most hazardous routes for truckers.

24 | P a g e

Overall, I-80 extends from eastern New Jersey to California and shows the greatest concentrations of weather related fatal accidents, including a diversity of winter and non-winter events, and wet pavement events. I-80 lies at the boundary of polar air masses to the north and warm, moist air masses from the south, thus producing more inclement weather.

The FMCSA report goes on to say that rainfall and wet pavement are the most common types of adverse weather to affect Commercial Motor Vehicles (CMVs) over ten thousand pounds between 1975 and 2006. Rainfall and wet roads are blamed for transport delays and an increased risk of collisions. Figure 8 shows the mean number of days with precipitation in the United States, with the number of fatal CMV accidents plotted underneath. The Weis Markets store network lies in the circled area, noted as the most dangerous area east of the Mississippi River by the FMCSA because of the high number of rainy days.

25 | P a g e

Figure 8: Mean number of days with measurable precipitation, with fatal CMV accidents (2001-2006) plotted underneath

In the colder months, the Weis Markets network is also exposed to some of the most dangerous trucking weather in the country, according to the FMCSA. Heavy snow, which is prevalent in the Northeast and Mid-Atlantic, causes transport delays, road closures, loss of traction, loss of visibility, and other driver control problems. The FMCSA report points out that areas on the eastern side of The Great Lakes, which again encompasses the Weis Markets network, are particularly prone to dangerous snow squalls that can drop visibility instantly. CMVs have significant vulnerabilities to reduced visibilities because they take 40% longer to stop compared to a typical automobile. Figure 9 (below) again shows that parts of Pennsylvania and New York, home to Weis Markets locations, are in the most vulnerable area for fatal CMV accidents. When compared to Figure 10, which shows the average number of days on which one inch or more of snowfall accumulates, this makes sense.

Figure 9: Plot of fatal CMV accidents, 2001-2006

26 | P a g e

Figure 10: Plot of mean days with snowfall greater than 1"

Even prolonged episodes of extreme heat can impact the transportation of goods. The extreme temperatures can degrade both safety and performance characteristics of CMVs, specifically including the engine, tires, and hydraulic systems.

Winds over 25 miles per hour can also be dangerous to truck drivers, as speeds of that magnitude can inhibit maneuverability and stability to high profile vehicles. Strong headwinds can also increase fuel consumption.

According to the FMCSA, weather hazards increase transportation costs by approximately 18%. With that in mind, we are confident that the PROFIT Index could cut that value down to 6%, saving Weis Markets a significant sum of money as shown in Table 10.

COSTAnnual Weis Markets Shipping Costs $60,499,470Average Daily Weis Markets Shipping Costs $165,751.97Average Network Days with >1” Snow or >1” Rain2 25Daily Cost on 25 Bad Weather Days (18% inc.) $195,587.33Daily Cost on 25 Bad Weather Days (6% inc.) $175,697.08Amount Saved with Decision Dynamics $497,256.05

Table 10: Transportation cost savings by using PROFIT Index. Source: Weis Markets 2011 10-K

Again, some assumptions are made in this table. However, we feel that conservative numbers were used, and that savings could be even higher than this. The number of days with snow over one inch was averaged out over the network, and assumed to be 15 as per Figure 10. The number of days with over one inch of rain was calculated from Climate Prediction Center data.

2 http :// www . cpc . ncep . noaa . gov / products / outreach / research _ papers / ncep _ cpc _ atlas /1/ fig 14. html

27 | P a g e

Realistically, rainfall of less than an inch can still cause accidents, so this would be one category where our values are conservative.

Finally, it is worth noting that Figure 10 only accounts for savings by using PROFIT to keep trucks out of hazardous weather. We also feel that we could prevent at least one weather related accident per year. According to the FMCSA, the cost of the average truck accident is $91,112. If that amount is included with Figure 10, the likely savings increases to: $588,368.05

Various Uses of PROFIT Index: The Public

We feel that Weis Markets would realize the largest financial benefit by using the PROFIT Index in-house to better staff, stock, and transport ahead of major weather events. However, the easy to understand PROFIT Index may also be beneficial to the consumer.

If the color coded index, or mapped version of the index, were published on social media websites, it could act as a free form of advertising. It would also help the consumer better understand when he or she should shop ahead of the storm. Keep in mind it should be re-branded for the public to reflect your own brand, with a catchy name such as the ‘Weis Weather Watcher’. For example, two or three days before a storm, when customers are most likely to visit their local grocery store, a Weis Markets employee in charge of social media could post something along the lines of “The Weis Weather Watch has gone red, stop in to pick up your storm supplies. All registers open!”

This would hopefully draw in customers who may be searching for storm supplies from a business that is well stocked and staffed. Plus, it would increase brand recognition and loyalty if the customers come to use the index regularly.

If the PROFIT Index were to be rebranded and used as an in-house marketing device on social media, where customers could readily share it with friends who may find it useful as well, it is likely that advertising in other sectors could be reduced. For example, non-circular ads in newspapers could be trimmed, as newspapers are not always timely sources of information. In many cases, newspaper ads are created weeks or months ahead of time according to the store manager in State College. That would limit the ability to advertise the current index rating ahead of a storm, while fast and free social media sites include no such limitation and have a wider reach.

In 2011, according to Form 10-K, Weis Markets spent $24.7 million on advertising. We propose promoting our PROFIT Index on social media sites that allow for interacting, sharing, and client building in place of stagnant newspaper ads. In doing that, we expect Weis Markets to save at least: $617,500.00

By increasing Weis Markets presence on social media, particularly in regards to promoting the index, we expect Weis Markets to attract new, younger customers that may not have been reached in the newspaper advertisements. This new clientele would not only increase sales, but it would foster a new generation of grocery shoppers. Right now, Weis Markets has 48,200 fans on

28 | P a g e

Facebook, and about 1,100 followers on Twitter. With PROFIT, we feel the combined followers could reach 75,000. If each of those new followers spent just $35 at Weis Markets each year, sales would increase by at least: $897,505.00

Various Uses of PROFIT Index: The Bottom Line

SAVINGSLabor Savings $220,634.40Transportation Savings $588,368.05Advertising Savings $617,500.00Sales Increase $897,505.00Amount Saved with Decision Dynamics $2,324,007.45

Table 11: Total savings by using PROFIT Index

Contract

The cost of the PROFIT Index is $300,000 per year. This will include weekly video conferences (or a total of 52 sessions) to consult about the index. Exclusivity rights in the grocery industry will cost $150,000 per year. As outlined in the report, there is a conservative savings estimate for Weis Markets of over 2.3 million dollars.

A minimum two year contract and 180 days notice of service cancellation is requested.

29 | P a g e

Sources

Committee on Earth Observation Satellites. (2012). Weather Forecasting. http://www.eohandbook.com/eohb05/ceos/part2_6.html

Cox, J. (2000). Weather for Dummies. http://www.books.google.com/books?id=Ecx-ASpqA2QC&printsec=frontcover&hl=en&sa=X&ei=N_6MT7w6w-HRAaGS9YEP&ved=0CEcQ6AEwAA#v=onepage

Environmental Modeling Center. (2012). Model Performance Statistics. http://www.emc.ncep.noaa.gov/gmb/STATS/html/aczhist.html

Great Lakes Environmental Research Laboratory. (2012). Great Lakes Water Temperatures. http://coastwatch.glerl.noaa.gov/marobs/sta0.html

Morningstar. (2012). WEIS Financial Annual Reports. http://quicktake.morningstar.com/stocknet/secdocuments.aspx?symbol=wmk

National Climatic Data Center. (2008). Snowfalls- Average Total in Inches. http://lwf.ncdc.noaa.gov/oa/climate/online/ccd/snowfall.html

National Centers for Environmental Prediction. (2011). Model Analyses and Guidance. http://mag.ncep.noaa.gov/

30 | P a g e

National Weather Service Eastern Region Headquarters. (2012). Middle Atlantic River Forecast. Center. http://www.erh.noaa.gov/marfc/

National Weather Service Eastern Region Headquarters Baltimore/ Washington. (2012). Preliminary Totals Ending February 6, 2010.http://www.erh.noaa.gov/lwx/events/?event=20100206

National Weather Service Eastern Region Headquarters Mount Holly/Philadelphia. (2012). Winter Storm February 5-6, 2010. http://www.erh.noaa.gov/phi/storms/02062010.html

National Weather Service Eastern Region Headquarters State College. (2012). Snow Storm. February 6, 2010. http://www.erh.noaa.gov/ctp/features/2010/02_06/

Tanski, J. (1993) Nor’easters. https://docs.google.com/viewer?a=v&q=cache:hiKhIGOvFYQJ:www.geo.hunter.cuny.edu/~fbuon/PGEOG_334/Literature_pdfs/NEpaper4.pdf+strength+of+a+nor%27easter&hl=en&gl=us&pid=bl&srcid=ADGEESgovpvEImxuJMlgc5Kml2vnEzgUP5KTTIwPAOf6FA4e9ua0IGqeZbsLz9B01Wtb_9ulpC5HnEn6rP9-iu5EZT-qUPP0RukPbb7kFqNe5__0brY756znpoc4JF2KFGhSbDBHtfK7&sig=AHIEtbSCEhlcYtYRfGXBKMterAmj849ZNA

Truck Accident Info. (2012). Truck Accident Info. http://www.truckaccidentinfo.com/

Truckers Report Facts. (2012). Facts About Trucks. http://www.thetruckersreport.com/facts-about-trucks/

US Department of Transportation. (2011). Weather and Climate Impacts on Commercial Motor

Vehicle Safety. http://www.fmcsa.dot.gov/facts-research/research-technology/report/Weather-Impacts-on-CMV-Safety-report.pdf

Zielinski, G. (2002). Classification Scheme for Winter Storms in the Eastern and Central United States with an Emphasis on Nor’easters.https://docs.google.com/viewer?a=v&q=cache:TmP7vWmeYPsJ:www.weatheranswer.com/public/Snow_storm_index_Zielinski_.pdf+nor%27easter+classifications&hl=en&gl=us&pid=bl&srcid=ADGEESh2lg6eYgJqlkNyUlzp44XQnHXLstHJrVydK6BPgiDVUzBsUv0WacmdAqWhPh03iyW6SqTXRYIcUIf1wFUxqwO9x3T7IgWo7n9Vv_crPVtXROu_mpgXNz7PdHR4Q7D4wVQ_8w&sig=AHIEtbRCcFOGmA2mj_MCX6j7ROYsZNz8jg

31 | P a g e

32 | P a g e