Metar Quiz

    
    KTTN 051853Z 04011KT 1/2SM VCTS SN FZFG BKN003 OVC010
    M02/M02 A3006 RMK AO2 TSB40 SLP176 P0002 T10171017=
    
    

How does it work?

All the questions on metarquiz are created and graded with AI. This is interesting because it means the quiz accepts "close enough" answers. For example, if the METAR included the airport code KJFK, all the following variations would be accepted for the question "Which airport is this METAR for?"

• KJFK
• JFK airport
• john f. kennedy airport

Here's another example:

I'm using OpenAI's GPT-3.5 to grade the questions. 3.5 is very fast and smart enough to know when an answer is close enough.

Every 5 minutes, a new question is generated with GPT-4. This is the prompt I send to the AI:

    
    "You are a helpful bot that helps users learn about aviation METARS. You create quizzes that users can take to test their knowledge."

    Create a new METAR question.

    The question should be in a JSON format:

    {
    "metar": "KLAX 052356Z 27015KT 10SM SCT030 20/15 A2988 RMK AO2 SLP121 T02000150 10244 20200 58010=",
    "content": "What is the reported wind direction and speed?"
    }

    Make sure that the METAR's are realistic. Also, vary the questions. Here are some examples of questions:

    "Which airport's weather is this METAR reporting?"
    "At what speed is the wind blowing according to this report?"
    "What does 'SCT020 BKN250' signify?"
    "Can you tell the atmospheric pressure from this METAR?"
    "How would you interpret '28/18' in this METAR?"
    "What does 'AO2' imply in this report?"
    "Can you decode 'T02830183'?"
    "What does '53006' stand for in this report?"
    "What kind of weather conditions would a pilot expect based on this METAR?"
    "How would you interpret '10306 20283' in this METAR?"
    "Can you tell the reported temperature in Celsius?"
    "What does 'A2996' represent?"
    "What is the purpose of 'RMK' in this METAR?"
    "What does 'AO2' imply in this report?"
    "How would you decode 'SLP143' in this METAR?"