Skip to main content

Introducing Habistack

Habistack is the most flexible, cloud-native forecast engine and API for delivering machine learning and statistics-based forecasts with any combination of variables.

Habistack’s cloud-based forecast engine can ingest data from any model or sensor, process that data through a machine learning algorithm and combine with the Fathym Platform to launch applications from ready-made templates auto-populated with Habistack data.

Habistack Architecture

Habistack combines the world’s best weather forecasts with statistics-based, machine-learning techniques to tackle large datasets, including road weather and surface condition forecasts.

Habistack brings together NOAA’s HRRR and GFS as a predictive weather forecast data stream with current conditions data from ground-based weather stations all over the US. This forecast model also incorporates historical sensor data from current-conditions sensors, establishing a combined forecast + IoT sensor machine learning model.

Habistack delivers a unique suite of highly specialized forecast variables derived through statistically-based machine learning models. These derived variables include road temperature, road state/condition, and a delay risk factor for destination arrival estimates.

Habistack Variables

Habistack draws from infrared and temperature sensors on roads to predict a Road Temperature derived variable for road surfaces anywhere in the world.

Combining that temperature prediction with known weather variables such as recent precipitation, it outputs a Road State variable. For example, when road temperatures drop below freezing and precipitation is forecast, a Road’s State may be “Icy.”

Habistack also determines when a road will be too hot or too windy – outputting a derived variable for a road route’s potential Delay Risk.

Habistack data outputs:

  • Point and Multiple Point Forecasts
  • Geo-Fenced Area Forecasts
  • Route and Alternative Route Forecasts
  • Road Temperature
  • Surface / Road Condition
  • Weather Delay Risk
  • 16 Hour CONUS Forecasts at 3-km Resolution
  • 120+ Hour Global Forecasts at 13-km Resolution

The Fathym platform offers open-source, pre-configured app templates and one-click, automated workflows to easily transform forecast data into dynamic apps and dashboards in the cloud.