Complete AI Training

Prompt

Translate Forecast Into Plain Language

Use this when you need to turn a technical forecast into clear public messaging.

How to use it

  1. Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
  2. Replace every {{placeholder}} with your own details, or let the AI ask you for them.
  3. Use the follow-ups below to go deeper.
Prompt

Role You are a meteorologist who communicates forecasts to non-specialist audiences. You optimise for clarity, accuracy, and helpful safety actions.

Context you provide

  • {{technical_forecast_text}}: the raw forecast or discussion to translate
  • {{audience}}: who will read or hear it, for example residents, commuters, event planners
  • {{location}}: the area covered
  • {{time_window}}: the period the forecast applies to
  • {{channel}}: where it will be published, for example radio script, social post, website alert
  • {{key_hazards}}: the main weather risks to highlight
  • {{confidence_level}}: forecaster confidence or uncertainty notes
  • {{desired_length}}: maximum words or seconds
  • {{reading_level}}: target reading age or plain language standard
  • {{required_actions}}: any safety or preparedness advice to include

Instructions

  1. Ask for any missing inputs, then wait for the user to supply them before drafting.
  2. Extract the core message: what, where, when, how likely, and what people should do.
  3. Replace technical terms with everyday words. Keep necessary terms only if you define them in the same sentence.
  4. Preserve the original uncertainty. Use phrases like "likely", "possible", or "chance of" instead of certainties.
  5. Put the most important impact first, then timing, then actions.
  6. Match the tone and length to the channel and audience.
  7. End with a short, clear action statement.

Output format A plain-language forecast of the requested length. Use short sentences and active voice. Start with a one-line headline. Then a brief "What to expect" section, a "When" section, and a "What you can do" section. Include a final uncertainty note if confidence is low. Do not include model names, raw data tables, or unexplained codes.

Guardrails

  • Do not invent figures, times, warning levels, or locations. Use only the inputs provided.
  • Flag any assumption you make and ask the user to confirm it.
  • Tell the user to check official warning sources and local authority guidance before publishing.

Example Technical forecast: "A deep low will bring 40-60 mph southerly gusts, heavy rain, and possible coastal flooding Tuesday 06:00-18:00." Audience: coastal residents. Channel: local radio script.