Data Source

The weather maps shown here are generated from the NCEP Global Forecast System (GFS) model. GFS is the primary operational model framework underlying U.S. NOAA/NWS weather forecasting. The model is run four times daily on a global T1534 gaussian grid (~13 km) to produce 16-day forecasts. Here, we use 0.25°x0.25° (~30 km) output grids available from NOMADS, and calculate daily averages from eight 3-hourly timeslices starting at 0000 UTC.

Additional Details

• Sea surface temperature (SST) and SST anomaly maps are generated from NOAA Optimum Interpolation SST version 2 (OISST V2). OISST is a 0.25°x0.25° blendend dataset derived from satellite, ship, and buoy measurements. The SST anomaly is based on a 1971-2000 NOAA climatology.


• Temperature refers to air temperature at 2 meters above the surface. The temperature anomaly is made in reference to a 1979-2000 climatology derived from the reanalysis of the NCEP Climate Forecast System (CFSR/CFSV2) model. This climate baseline is used instead of the 1981-2010 climate normal because it spans a period prior to significant warming of the Arctic beyond historically-observed values. For context, see this timeseries plot showing how various climate baselines compare against the NASA GISS 1880-2014 global land-ocean temperature index.


• GFS Model Bias Correction — GFS has a systematic model bias compared to CFSR/CFSV2 in which the diurnal cycle has greater amplitude over areas of snow/ice and hot desert. A correction term, calculated from the difference between GFS and CFSR/CFSV2 on average per month from June 2015 - May 2016, is therefore applied such that regional temperature anomaly values are consistent with CFSR/CFSV2. A timeseries plot showing the bias corrections for each region is shown here. Temperature anomaly maps showing uncorrected and corrected GFS vs. CFSR/CFSV2 can also be found here.

  Note that bias-corrected GFS-CFSR temperature anomalies will still differ slightly from those calculated from CFSV2-CFSR. The most reliable temperature anomaly estimates can be found in the Daily Reanalysis Maps image archive. The archive is updated once or twice per month.

  See this NCEP/NWS discussion for additional information on model bias.