Lucas Ehinger

Lucas Ehinger

Experimental Nuclear Physics PhD Student at MIT

New Hampshire Weather

Model Details Historical Weather
?
Select a summit. Quick links and plots will automatically update.
Model:
?
Toggle which models' traces appear; plots with no data for the selected models are hidden automatically. HRRR, NAM, GFS and RAP cover every field; ECMWF and NBM only publish some (e.g. temperature, ceiling, visibility), so they populate just the relevant plots. "All" is the combined-parameters undercast model.
Units:
?
Switch between metric (C, m) and imperial (F, ft) for all plots.

Quick Links

?
Links to weather forecasts and tools
for specified forecast location.
Some weather modeling I've done to determine whether it'll be undercast in the mountains. The model details page covers how the models are built and how well they actually do (honestly: not that well — undercast is rare and hard to predict). TLDR; it's undercast only a handful of times per year on Mt. Washington.
Model Explanations & Details
This site uses atmospheric model data from NOAA's HRRR, RAP, NAM, and GFS models, the ECMWF open data, and the National Blend of Models (NBM) to generate forecasts for undercast probability at several summits in New Hampshire. Forecast resolution, accuracy, and update frequency vary between the models.
Model Forecast Range (Hours) Spatial Resolution Strengths Weaknesses
HRRR0-483 kmHigh temporal resolution, good for short-term forecastsLimited range, may struggle with longer-term trends
NAM0-603 kmGood for medium-range forecasts, covers larger areasLower resolution compared to HRRR, less accurate for local events
GFS0-12013 kmLong-range forecasts, global coverageCoarser resolution, may miss localized weather phenomena
RAP0-4813 kmHourly regional refresh (HRRR's parent model)Coarser than HRRR, short range only
ECMWF0-48~0.25° (~28 km)Highly skilled global dynamics and thermodynamicsNo cloud fields in open data; 3-hourly steps
NBM0-48~2.5 kmCalibrated blend; direct ceiling & visibility, incl. probabilisticSurface fields only (no upper-air); blends can smooth extremes
The forecasts you see on NWS and other sites are often based on a combination of these models along with observational data and expert analysis. Trust them, before you trust my plots, which come from individual models and are interpreted by a dumb particle physicist.

I built this to visualize data not readily available from the NWS website. While exact wind speeds from an individual, non-local, coarse-grain model may be inaccurate, larger trends such as sky coverage and cloud elevation can be informative. Full model details can be viewed here.


Individual Parameters

Atmospheric models produced and provided by the National Oceanic and Atmospheric Administration (NOAA), and accessed via the Herbie python package. Machine learning datasets utilized Mt. Washington webcams (MWOB).