Adding Difficulty in Weather Forecasting Challenges to Enhance Learning

A little considered consequence of the weather forecasting information firehose is that for meteorologists, it is increasingly difficult to start from scratch and produce a truly independent expectation for how the atmosphere may evolve. For students attempting to develop forecasting skills, this is difficult; hoards of datasets, other forecasts, and discussions online can contaminate and bias their own developing forecasting skills. Classroom exercises can focus on historical or realtime events to assist students in learning how to develop their own expectation, but sometimes non meteorological context clues about the event or an overwhelming set of data can prevent students from stripping down the process to its bare bones.

In an attempt to overcome some of these issues, students of the Oklahoma Weather Lab were challenged to forecast a random, historical (1948-2016) severe weather event. Python software was created to develop a full set of products for the students to sort through when developing their forecast. Unlike most case studies, this event utilized actual depreciated data sources (e. g. WSR-57 radar), to break students out of typical forecasting habits and force them to find ways to sort through unfamiliar data in order to solve the forecasting problem. Special care was taken to conceal the event type and remove as many non meteorological context clues as possible from the data.

The participants of this challenge enjoyed the unexpected and unfamiliar nature of the forecasting challenge. Many of them were surprised by their own performance and the performance of tools that have long since been replaced in operational forecasting. The success behind this exercise suggests that the element of surprise can be used effectively in creative ways when in teaching forecasting.