As Time Goes By in Weather Forecasting: For Teachers

This WebQuest is designed to help students learn some basic concepts in statistics using real-time, self-collected weather forecast data. The project involves research and reporting, web searching and data gathering, mathematical calculation, and interpretive reasoning. The project is designed as a group activity. The suggested evaluation rubric includes components for a student’s individual contribution, as well as for the group effort.

Grade Level

This WebQuest is appropriate for grades 5-12.

Objectives

At the end of this activity, students will have accomplished the following:

  • Gathered data using the web
  • learned the statistical concepts of mean, error, bias and lag
  • compared the accuracy of weather forecasts using the concepts of ‘mean absolute errors’ and ‘mean relative errors’
  • evaluated the accuracy of weather forecasts as a function of lag time
  • prepared individual and group reports using data gathered form the web
  • referenced information sources
  • collaboratively analyzed an environmental problem (the accuracy of weather forecasts)
  • learned to make interpretations of statistical results

Requirements

An Internet-connected computer. Students may use home computers if they wish. (If this is not available, the project can still be accomplished using traditional media: newspapers, television and radio.)

Time

This project requires three weeks.

  • Week 1: data gathering (collect forecasts)
  • Week 2: data gathering (collect forecasts)
  • Week 3: calculations, analysis and preparation of reports

Other Ideas

  • Each group could evaluate weather forecasts for a different city. Results from different groups could be prepared to answer the following questions:
    • Does a certain network (ABC, FOX, etc.) tend to produce more accurate forecasts?
    • Are high temperature forecasts more accurate in certain cities? (In other words, are the mean forecast errors smaller for certain cities?)
    • Does the deterioration of forecast accuracy with increasing lag times change for different cities?
  • Discuss the difficulties in comparing 1-day forecast results with 4-day forecast results. How does the different sample size (more 1-day forecasts than 4-day forecasts) affect the comparison? How might the problem of different sample sizes be eliminated?
  • Examine the interpretation of different values of relative errors.
    • Discuss the concept of bias (relative error consistently positive or negative).
    • Discuss how a relative error near zero does not necessarily indicate a good forecaster (for example, very high overpredictions could balance out very low underpredictions).
    • Discuss how the absolute error does not provide any information about bias.
  • Examine the difference between relative and absolute errors.
    • What’s better – a small mean relative error or a small mean absolute error? Why?
    • Why isn’t just one type of mean error (relative or absolute) enough to describe the accuracy of a forecaster?
    • Why is the following equation incorrect?:
      AAE=|ARE|

      (in other words, why isn’t the average absolute forecast error simply the absolute value of the average relative forecast error?)

  • Examine some relevant meteorological considerations.
    • What might be some reasons for differing quality of forecasts for different cities? For different Lags? Is the weather more complex (harder to forecast) at certain cities? Why?
    • Do you think the results might change if the forecasts were collected during a different two week period? Why?
  • Why isn’t just one type of mean error (relative or absolute) enough to describe the accuracy of a forecaster?
  • Would the results be different if the forecasts were made in degrees Celsius instead of degrees Fahrenheit? Why or why not?
  • This WebQuest addresses the issue of ‘Who is the best weather forecaster’ in any particular location. This is a relative question. In other words, we’re only comparing the forecasters to each other. It might be interesting to ask the question in an absolute sense: Are the forecasters really any good? This question can be addressed by adding a new forecaster, called “Persistence”. A persistence forecast works like this: tomorrow’s high temperature will be the same as today’s high temperature. A good forecaster should achieve results that are consistently better than persistence forecasts.

Standards

This WebQuest specifically addresses the following National Science Education Standards, but may also touch on other standards not listed:

Science as Inquiry – Content Standard A

  • Use appropriate tools and techniques to gather, analyze and interpret data.
  • Think critically and logically to make the relationships between evidence and explanations.
  • Recognize and analyze alternative explanations and predictions.
  • Use mathematics in all aspects of scientific inquiry
  • Understandings about scientific inquiry

Earth and Space Science – Content Standard D

  • Structure of the earth system
    • atmosphere
    • clouds
    • global patterns of atmospheric movement

Science and Technology – Content Standard E

  • Developing student abilities and understandings
    • straightforward activity with only a few well-defined ways to solve the problems involved
    • only one or two science ideas involved in the task
  • Implement a proposed design
  • Understandings about science and technology

History and Nature of Science – Content Standard G

  • Science as a human endeavor
  • Nature of science

Coordination of Science and Mathematics – Program Standard C

  • Use estimations
  • Identify and use functional relationships
  • Develop and use tables, graphs and rules to describe situations
  • Use statistical methods to describe, analyze, evaluate and make decisions

Source: National Science Education Standards, 1996, National Academy Press, Washington, DC, 262 pp., www.nas.edu.