How can I forecast tomorrow s weather?

Transcription

1 How can I forecast tomorrow s weather? A laboratory experiment from the Little Shop of Physics at Colorado State University CMMAP Reach for the sky. Overview While there are considerable difficulties in forecasting the day-to-day weather accurately in a quantitative sense, it is relatively easy to make your own qualitative forecast. A relatively good forecast can be made one day in advance based on knowledge of the weather conditions the day before. By completing the basics of this exercise, you should be able to make an informed decision as to whether it will rain, the relative intensity of the rain, and whether it will be hot or cold. The more advanced areas of the following activities will help you to develop a more quantitative forecast. Theory Most information for a qualitative forecast can be seen on a simple weather map created by taking observations at many different places at the same time. If you see a low pressure area that has been moving eastward toward your current location, what type of weather might you expect? If a cold front is moving southward out of Canada, what would you soon expect the temperature to do? If it is going to be cloudy tomorrow, will it be warmer or colder than it was today? Often the best qualitative forecast follows from persistence. If it is warm and sunny today, and it was warm and sunny yesterday, the odds are pretty good that it will be warm and sunny tomorrow, that is, unless you see something else heading in on your weather map. You can do a lot to forecast tomorrow s weather just by looking at a weather map. Necessary materials: NASA Hurricane tracking activity packet An example station plot map with sea level pressure contours Access to the past few days of weather analyses, including surface data with station plots as well as upper-level data Access to current local model output statistics (MOS) The example station plot map can cover various US locations and can be created at Recent weather analyses can be found at the above address or at Local MOS can be obtained from 1 In order to predict specific quantities in a forecast, such as temperature, humidity, and rainfall, we rely on computer models. Numerical weather prediction (NWP) uses a system of equations that describe the be-

2 havior of the atmosphere. NWP models read in the current state of the weather from observations as an initial condition and then calculates the solutions to the equations that will tell us what the weather will be a few seconds into the future. Then it uses that information and repeats the calculations many times until it gives a quantitative forecast for a few days into the future. Meteorologists like to run many versions of the same forecast with slight differences in the initial conditions to account for a lack of perfect observations. The multiple forecast runs are then averaged together as a summary of the predicted state of the atmosphere. Added to this information is information about how the weather acted the last time that similar weather conditions existed. So if the weather five years ago was very close to what it is like today, it is likely that tomorrow s weather will be very much like what happened five years ago, too. This summary is known as Model Output Statistics (MOS). MOS is often called model guidance because forecasters will use the generated numbers as a guide to make their forecast. The results are never perfect, and different models give different forecasts, too. Therefore, forecasting usually takes some intuition, which is why most people can read the different MOS data sets and come up with reasonable forecasts. Doing the Experiment This experiment includes a number of activities that progressively lead from qualitative to quantitative forecasting. Part I: Hurricane Tracking In this section, students will use satellite images to predict when and where a hurricane will hit. Big idea: Observations can lead too future predictions. The hurricane tracking packet includes a hurricane tracking chart and satellite images of Hurricane Georges moving through the Caribbean Sea at 8 different points in time. Begin by placing an X on the hurricane tracking chart for the first 4 positions of the hurricane. Connect the Xs with a dark or colored line. Based on what you have just drawn, see if you can predict where and when the hurricane will hit the United States. Add the next day s position to your chart and reevaluate your prediction. Has your predicted time and location of landfall changed? Keep adding positions, and see where and when the hurricane made landfall. For further inquiry: Ask the students how this image was made. Is it radar, visible, infrared? Discuss the structure of the hurricane. Are the eye, eye-wall, or spiral rain bands visible? Where would you expect the most damage from this storm? i.e., where is the storm surge strongest? If hurricanes are a particular threat in your area, a discussion of the differences between a hurricane watch and warning may be warranted. A few examples are listed on the hurricane tracking chart. If societal impacts are a focus of the course, research the history of this storm and its aftermath and others for comparison. Part II: Analyze Your Own Weather Map From Station Plots Obtain a map of station plots as directed under Necessary Materials. Explain the coding of the station plots to your students. Details can be found at Read the data off of the station plot map and determine the location of cold and warm fronts based on noted precipitation, wind shifts, pressure troughs, and rapid spatial temperature changes. 2

3 For further inquiry: Have students contour the sea level pressure or temperature surfaces before analyzing the map for fronts. Part III: Tracking Day-to-Day Weather In this section, you will compile a few days of national weather maps and satellite images (optional) in order to make a qualitative forecast for a location of your choosing. Collection of the maps can be done on a daily basis, making predictions for the next day based on new observations or you can collect archived maps. Show the students each weather map in order to get a feel for how weather systems move across the United States. Pay particular attention to cold and warm fronts, wind pattern, pressure patterns, and any precipitation. Make a prediction based on the movement of the weather like you did in Part I for the hurricane. Make sure to verify your prediction the following day. For further inquiry: Look at upper-level wind and height charts and determine wether surface cyclones will intensify based on upper-level convergence and divergence. Part IV: Decoding Model Output Statistics Obtain MOS from at least two different models from the website listed under Necessary Materials. Explain the MOS coding system to the students. For details on this visit: Compare the statistics given by each the models. How are they similar? How do they differ? Based on the information obtained in Part III and from the MOS data, make a quantitative forecast for tomorrow s high and low temperatures, averaged sustained and gusting wind speeds, wind direction, sky conditions, and chance of precipitation. What are the chances of a thunderstorm tomorrow? Summing Up After completing this set of exercises, you should have a feel for how meteorologists and you can forecast tomorrow s weather. It turns out that all you really have to do is read a map or some model data. For More Information CMMAP, the Center for Multi-Scale Modeling of Atmospheric Processes: Little Shop of Physics: 3

4 Weather Forecasting - Qualitative Most information for a qualitative forecast can be seen on a simple weather map If you see a low pressure area that has been moving eastward towards Colorado, what type of weather might you expect? If a cold front is moving southward through Wyoming, what would you expect the temperature to do? If it s s going to be cloudy tomorrow, will it be warmer or colder than it was today? Often the best forecast is persistence: if it s s warm and sunny today, and it was warm and sunny yesterday, the odds are pretty good that it will be warm and sunny tomorrow (unless you know something else)

5 How can I forecast tomorrow s weather? Related Questions Part I: Hurricane Tracking In this section, students will use satellite images to predict when and where a hurricane will hit the coast. BIG IDEA: Observations can lead too future predictions. The hurricane tracking packet includes a hurricane tracking chart and satellite images of Hurricane Georges moving through the Caribbean Sea at 8 different points in time. 1. Begin by placing an X on the hurricane tracking chart for the first four positions of the hurricane. 2. Connect the Xs with a dark or colored line. 3. Based on what you have just drawn, see if you can predict where and when the hurricane will hit the United States. Prediction #1: 4. Keep adding positions from positions 5-8 one at a time, and see where and when the hurricane made landfall each time you update your tracking chart. Prediction #2: Prediction #3: Prediction #4: Prediction #5: 5. Has your predicted time and location of landfall changed? 6. How this image was made. Is it radar, visible, infrared? 7. What do the colored parts of the hurricane indicate?

6

7

8

9 Temperature ( o F) Station Plots G23 Wind speed, direction, and peak gust Station Pressure Visibility (miles) Current Weather 9 45 FNL hour pressure change Dewpoint ( o F) Sky Cover Station ID 3 hour precipitation Reading Station Plots Pressure 000 to 499: mb to mb 500 to 999: mb to mb Pressure Tendency In tenths of a mb change in 3 hours E.g., -32 is a drop of 3.2 mb in 3 hrs Temperature in o F Reading Station Plots

10 Reading Station Plots

11 Station Plot

12

13

14 Weather Forecasting - Quantitative In order to predict specific quantities in a forecast (temperature, humidity, rainfall) we rely on computer models Numerical weather prediction (NWP) uses a system of equations that describes the behavior of the atmosphere NWP model uses the current state of the atmosphere as its initial condition and steps through a small time step, recalculating every number for each step until the forecast time is reached So why aren t t forecasts always right? Observations aren t t good enough! A model is only as good as its initial conditions Even having an observation for every square meter of the planet would leave out smaller details and and we don t t have even close to that many observations Computers aren t t fast enough! In order to truly create a perfect forecast, one would have to use the exact equations on a really, really small spatial scale In order to create a model that will create a 12 hour forecast in less than 12 hours, we must approximate certain parts of equations and run the model on a grid with spacing of multiple kilometers ~1.2 million grid points x thousands of calculations > 1 Billion calculations per time step More than 1 Quadrillion calculations per time step at 1 meter resolution (10 12 ) More than 4 Quadrillion calculations to simulate 12 hours Forces us to make approximations to the equations that govern the atmosphere Little bit of error added at each time step Chaos reigns supreme! Ever heard of the butterfly effect? It s s more than just a movie

15 Chaos Theory Also known as the Butterfly Effect,, coined by Dr. Edward Lorenz in the 60 s s (he s s still teaching at MIT) Basically says that even the most insignificant change to initial conditions will magnify into drastic changes The smallest disturbance will eventually grow into a large difference this limits the range of forecasts to just a few days Even if the initial conditions and computing power were perfect,, chaos theory would limit us to a reasonable range of about 2 weeks Computers cannot possibly predict the movement of a butterfly, or how hard you step on the gas pedal So How far have we come? 100 years ago we just looked to the west. The first attempt at numerical weather prediction by Lewis Fry Richardson was done by hand. On a WWI battlefield as part of an ambulance unit in Northern France Just wanted to predict the pressure change over the next six hours Calculation took him six weeks! And he was off by over 140 mb Richardson s s Forecast Factory

16 How far have we come? In 1937, the US started using weather balloons. It wasn t t until WWII that the existence of a jet stream was confirmed. In 1948, ENIAC was put together by John von Neumann in a 30 by 50 room. In April 1950, the first 24-hr forecast was attempted. Took more than 24 hours due to breakdowns 8 years later, forecasts began to show signs of skill. As it stands today, we can have a great deal of faith in a weather forecast out to about 3 days. Just 30 years ago, we could only do 2 days Beyond that, the accuracy drops dramatically If you see a 15 day forecast don don t t believe it MOS The output from a model can be shown in the form of Model Output Statistics (MOS) MOS is a summary of the predicted condition of the atmosphere at each forecast time MOS is often called model guidance,, because forecasters will use the generated numbers as a guide to make their forecast MOS isn t t perfect though, and forecasting takes some intuition

17 The premise behind MOS The (preceding) models produce output describing the weather over No. America and vicinity at grid points or in wave space. Conditions for this time of year at a specific city may have occurred similarly in the historic past (i.e., old climate data are used). Forecasts of current conditions can be made for a city using the current model output and based on the historic weather outcomes. Downfalls of MOS There isn t t just one model that is used for NWP Multiple models are used that have differences in resolution and in the equations used and assumptions made The models never agree on everything A good forecaster will look at multiple model predictions and have a feel for which model performs the best under certain circumstances Station Name Artesia, NM Date and Time the Model was run

18 Decoding MOS Max and Min Temperature KMIA AVN MOS GUIDANCE 9/04/ UTC DT /SEPT 4 /SEPT 5 /SEPT 6 / HR X/N TMP DPT CLD BK BK BK OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV BK WDR WSP P P Q Q T06 33/ 0 64/ 0 21/ 0 19/ 0 42/ 0 80/ 0 37/ 0 37/ 0 39/ 0 22/ 0 T12 66/ 0 42/ 0 93/ 0 58/ 0 66/ 0 CIG VIS OBV N N N N N N N N N N N N N N N N N N N N N Decoding MOS 3 hourly temperature forecasts KMIA AVN MOS GUIDANCE 9/04/ UTC DT /SEPT 4 /SEPT 5 /SEPT 6 / HR X/N TMP DPT CLD BK BK BK OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV BK WDR WSP P P Q Q T06 33/ 0 64/ 0 21/ 0 19/ 0 42/ 0 80/ 0 37/ 0 37/ 0 39/ 0 22/ 0 T12 66/ 0 42/ 0 93/ 0 58/ 0 66/ 0 CIG VIS OBV N N N N N N N N N N N N N N N N N N N N N Decoding MOS Cloud Cover KMIA AVN MOS GUIDANCE 9/04/ UTC DT /SEPT 4 /SEPT 5 /SEPT 6 / HR X/N TMP DPT CLD BK BK BK OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV BK WDR WSP P P Q Q T06 33/ 0 64/ 0 21/ 0 19/ 0 42/ 0 80/ 0 37/ 0 37/ 0 39/ 0 22/ 0 T12 66/ 0 42/ 0 93/ 0 58/ 0 66/ 0 CIG VIS OBV N N N N N N N N N N N N N N N N N N N N N

19 Decoding MOS - Cloud Cover CL = Clear, no clouds SC = Scattered, between clear and sky 1/2 filled with clouds BK = Broken, 1/2 - totally cloudy OV = Overcast Decoding MOS Winds KMIA AVN MOS GUIDANCE 9/04/ UTC DT /SEPT 4 /SEPT 5 /SEPT 6 / HR X/N TMP DPT CLD BK BK BK OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV BK WDR WSP P P Q Q T06 33/ 0 64/ 0 21/ 0 19/ 0 42/ 0 80/ 0 37/ 0 37/ 0 39/ 0 22/ 0 T12 66/ 0 42/ 0 93/ 0 58/ 0 66/ 0 CIG VIS OBV N N N N N N N N N N N N N N N N N N N N N Meteorology Wind Direction 180 = From the South 090 = From the East 270 = From the West 000 = From the North

20 Decoding MOS Chance of measurable precipitation KMIA AVN MOS GUIDANCE 9/04/ UTC DT /SEPT 4 /SEPT 5 /SEPT 6 / HR X/N TMP DPT CLD BK BK BK OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV BK WDR WSP P P Q Q T06 33/ 0 64/ 0 21/ 0 19/ 0 42/ 0 80/ 0 37/ 0 37/ 0 39/ 0 22/ 0 T12 66/ 0 42/ 0 93/ 0 58/ 0 66/ 0 CIG VIS OBV N N N N N N N N N N N N N N N N N N N N N Decoding MOS Quantitative Precipitation Forecast KMIA AVN MOS GUIDANCE 9/04/ UTC DT /SEPT 4 /SEPT 5 /SEPT 6 / HR X/N TMP DPT CLD BK BK BK OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV BK WDR WSP P P Q Q T06 33/ 0 64/ 0 21/ 0 19/ 0 42/ 0 80/ 0 37/ 0 37/ 0 39/ 0 22/ 0 T12 66/ 0 42/ 0 93/ 0 58/ 0 66/ 0 CIG VIS OBV N N N N N N N N N N N N N N N N N N N N N T06, T12: probability of thunderstorms / severe t-storms over 6, 12h Decoding MOS Precip. Amount 0 = no precipitation 1 = 0.01 to 0.09 inches 2 = 0.10 to 0.24 inches 3 = 0.25 to 0.49 inches 4 = 0.50 to 0.99 inches 5 = 1.00 to 1.99 inches 6 = 2.00 inches or greater

21 Decoding MOS Local forecast KMIA AVN MOS GUIDANCE 9/04/ UTC DT /SEPT 4 /SEPT 5 /SEPT 6 / HR X/N TMP DPT CLD BK BK BK OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV BK WDR WSP P P Q Q High Temperature: 89F Low Temperature: 77F Precipitation Category: Cat-4 Decoding MOS Zone Forecast KMIA AVN MOS GUIDANCE 9/04/ UTC DT /SEPT 4 /SEPT 5 /SEPT 6 / HR X/N TMP DPT CLD BK BK BK OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV BK WDR WSP P P Q Q T06 33/ 0 64/ 0 21/ 0 19/ 0 42/ 0 80/ 0 37/ 0 37/ 0 39/ 0 22/ 0 T12 66/ 0 42/ 0 93/ 0 58/ 0 66/ 0 Tonight: Mostly cloudy with scattered showers and thunderstorms. Lows in the mid 70s. Light southeast winds. Chance of rain 30%. Tomorrow: Mostly cloudy with scattered to numerous showers and thunderstorms. Highs near 90. Light southwest winds. Chance of rain 70%. Tomorrow Night: Mostly cloudy with scattered showers and thunderstorms. Lows in the mid 70s. Light southwest winds. Chance of rain 70%. Decoding MOS Zone Forecast KMIA AVN MOS GUIDANCE 9/04/ UTC DT /SEPT 4 /SEPT 5 /SEPT 6 / HR X/N TMP DPT CLD BK BK BK OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV OV BK WDR WSP P P Q Q T06 33/ 0 64/ 0 21/ 0 19/ 0 42/ 0 80/ 0 37/ 0 37/ 0 39/ 0 22/ 0 T12 66/ 0 42/ 0 93/ 0 58/ 0 66/ 0 Tonight: Mostly cloudy with scattered showers and thunderstorms. Lows in the mid 70s. Light southeast winds. Chance of rain 30%. Tomorrow: Mostly cloudy with scattered to numerous showers and thunderstorms. Highs near 90. Light southwest winds. Chance of rain 70%. Tomorrow Night: Mostly cloudy with scattered showers and thunderstorms. Lows in the mid 70s. Light southwest winds. Chance of rain 70%.

22 Description of MOS Alphanumeric Message

23 Definitions of Categorical Elements