Into the eye of the storm: NEXRAD Level II open data
Last month Hurricane Douglas quickly intensified to a Category 4 tropical cyclone as it passed Hawaii Islands. Using radar data, Allan Walker built a 3D model looking into the eye of Douglas, using Mapbox GL JS — including bathymetric data and topological data for the surrounding Islands. The animation follows the track of the storm, with the the sky changing colour depending on the local time.
The visualization comprises of three sources:
- NEXRAD II Radar messages downloaded from Amazon S3 and converted to geojson using the NOAA’s Weather and Climate Toolkit
- Bathymetry and Topology data from the School of Ocean and Earth Science and Technology, The University of Hawaii at Manoa was converted from a NetCDF raster grid to contours
- Meteorological data (such as Wind Speed and Pressure) with storm positions from IBTrACS was used to generate the storm track (Note: you can view a map of the Douglas track at NOAA Historical Hurricane Tracks and additional animations at the NOAA National Hurricane Center archive)
The NEXRAD Level II data is generated for the four radar stations in the Hawaii Island Chain for the 27th of July 2020. It comes in roughly 5 minute “time slices” and each “slice” is also chopped up per “sweep” (of the radar). Therefore the data has to be unioned and joined together on a timestamp to synchronize the four radar stations, “stitching” together a view of the Hurricane’s path. NEXRAD II data has two metrics to visualize: Reflectivity, expressed as dBZ; and height Above Sea Level (in meters). We use the plasma color ramp to encode Reflectivity, and extruded the polygons using the height Above Sea Level attribute.
The Bathymetry and Topology data is converted from a raster to contours as geojson using the SAGA Contour tool in QGIS, firstly as lines, which are then transformed into polygons. We choose two color palettes from cpt-city: “ibsco-bath” and “wiki-knutux” respectively.
Using a preview build of GL-JS, the contours are draped over 3D terrain. We also use a sunrise and sunset calculator by Cambridge in Color to build a palette for the sky fade colors, allowing us to simulate colors with the local time of day (or night).
To allow a camera to follow and enter the eye of the hurricane, we use the storm track positions, find the bearing between each point and then synchronize timestamps with the radar data. For greater context, we add the wind speed (Knots) and pressure (Millibars) information for each point along the track.