Forecast Model Options Follow

WeatherFlow-Tempest suite of applications (iKitesurf, iWindsurf, SailFlow, Fishweather etc) presents computer-generated weather forecast data for points of interest around the globe. We endeavor to produce and present data that has proven to be the most accurate over time. Since no forecast model can be perfect, power users such as our team of operational meteorologists find it helpful to review multiple models for a given location, which is why we present data from several forecast models within our apps.

All the forecasting models presented are foundationally based on numerical weather prediction (NWP), a method of weather forecasting that employs a set of equations translated into computer code that use governing physical processes combined with initial and boundary conditions (including our observations). The code is run over a specific domain (geographic area) and the resulting output predicts future conditions. In some cases the forecast models we present also use machine learning and other forms of AI to further take advantage of observation data and post-process the NWP model output into a superior forecast.

TRRM - Tempest Real-time Refresh Model

By default, we present the Tempest Real-time Refresh Model (TRRM), WeatherFlow-Tempest’s premium weather forecasting product that updates hourly and has proven to be the most accurate wind prediction model over a time scale of 1 hour to 36 hours. TRRM is unique in that it combines the best-available numerical weather prediction models (NWP) for a given point with well-tuned/trained AI modeling. The TRRM model leverages high-density weather station datasets curated from the WF-ProNet, Tempest networks and other weather station networks. This rapidly growing set of observation data is a powerful tool for training the AI model for areas rich with this data.

Regional NWP Models

WF-WRF - WeatherFlow Weather Research & Forecast Model

The Weather Research and Forecasting (WRF) Model is a state of the art mesoscale numerical weather prediction (NWP) system. It features a data assimilation system and a software architecture supporting parallel computation and system extensibility. As WRF can be run across multiple scales it is especially useful for providing the high resolution data required by our applications. We currently both run WRF operationally and source model runs from third parties, at a resolution as high as 1km. Throughout CONUS we run or source WRF at a minimum resolution of 3km. We are constantly refining our WRF process to source the best available data for our AI enhanced TRRM model.

HRRR - High Resolution Rapid Refresh

The HRRR is National Oceanic and Atmospheric Administration (NOAA) & National Centers for Environmental Prediction (NCEP) operational weather prediction system comprised of a numerical forecast model and an analysis/assimilation system to initialize the model. It is run with 3 km resolution over CONUS every hour, and the high temporal and spatial resolution often yields the best available output for TRRM in certain locations.

ARW 5km - Hawaii Model
This model run from UCAR incorporates codes from several different NWP models in an attempt to get better accuracy in Hawaii and Alaska than is attainable with the Global Models. The 5km resolution enables the model to incorporate the intricacies of islands and varying topography.

NAM - North American Mesoscale Model

The North American Mesoscale Forecast System (NAM) is one of the National Centers For Environmental Prediction’s (NCEP) major models for producing weather forecasts. Based on WRF code, NAM runs 4 times per day and generates multiple grids of weather forecasts over the North American continent at horizontal resolution ranging from 3km to 12km depending on location.  It has proven to be quite accurate over the first 48 hours in areas where spatial resolution is high, and therefore may be selected as the base model for TRRM in some circumstances.

BLEND - Blend of Models

This model combines short and long-range NWP forecast model output for a complete forecast table that extends for 14 days. The BLEND uses the best available high-resolution, hourly updating NWP computer model when available  for forecasting the next 48 hours and blends that output with coarser resolution data for an extended forecast.  With a single data set, BLEND allows you to view both short and long term forecasts, while bringing the accuracy only available in short-term NWP model output.

Global NWP Models

Global NWP Models are characterized by lower spatial resolution but as the name implies these models provide forecast data for the entire globe.   In remote areas this is often the only data available.  Global NWP models are also integral in the initialization of higher resolution models over targeted geographic areas.

GFS - Global Forecast System
The GFS is run four times a day by NOAA and produces forecasts up to 16 days in advance.  It is run at roughly 50km resolution.  However, the spatial and temporal resolution is reduced as the model is further out in time.  It produces a forecast for every 3rd hour for the first 180 hours.

CMC - Environment Canada Global Model
This is another model which will give you a forecast for all points on the globe. The resolution is approximately 100km resolution and runs out 138 hours.