Once manual exploration stops scaling, the forecast has to arrive in your own system. There are two routes: pull the research datasets directly from Google's cloud products, or call an interface that returns forecast answers for a location and a time range. This page covers both, and the download guide covers what the data actually contains.
7-day free trial, then $19/mo · API keys are included on every plan

Four steps take you from a manual forecast to an automated one.
Sign in with Google, then pick a plan. Every monthly plan starts with a 7-day free trial and includes forecast credits.
Create and label a key in your settings so you can rotate or revoke it without disturbing other integrations.
Ask for a location, a variable set and a time range. Requests take coordinates or place names and return the hourly fields you selected.
Schedule the call so forecasts land in dashboards, alerting rules or planning tools before anyone has to ask for them.
The interface returns the same hourly forecasts you see in the workspace, which means one integration covers planning, alerting and reporting rather than three separate data pulls.
Hourly series for a coordinate or named place, ready to join with an asset register.
Precipitation, temperature, humidity, wind, radiation and cloud cover, chosen per request.
Request the window you plan over instead of pulling a fixed daily summary.
Usage draws on the same monthly forecast credits as the workspace, so cost stays in one place.
If you need the underlying fields rather than an answer, Google publishes forecast data through its own cloud and developer products.
Query forecast tables directly and join them with your operational data in SQL.
Combine forecast layers with satellite imagery and land data in geospatial analysis.
Retrieve bulk files when your pipeline needs raw arrays rather than query results.
Developer-facing weather data for products that already sit on Google's platform.
One key and one contract from this platform instead of separate integrations per source.
Variable names, units, time handling and credit behaviour, so your integration matches how the model actually reports.
What developers ask before putting a forecast into production.
Related: what can be downloaded and how Google exposes the model.
Start the trial, create a key and make your first call today. If you are still weighing the routes, the download guide compares data channels side by side.