Super Typhoon Nepartak struck the southeastern coast of Taiwan on 7 July 2016 with maximum sustained winds of 150 mph and a central pressure around 930 mb. 16 hours prior to landfall, the National Weather Service (NWS)’s Hurricane Weather Research and Forecasting (HWRF) model output projected radar reflectivities for the storm, anticipating the rainfall intensities that would be associated with Nepartak’s arrival. This output was stored in GRIB2 format (a World Meteorological Organization standard gridded binary format) and made available at the National Oceanic and Atmospheric Administration (NOAA) Operational Model Archive Distribution System (NOMADS) website.
STORM Virtual Globe HWRF (STORM VG HWRF) shows how HWRF data can be displayed in an interactive and visually appealing way. Matt Lammers, web analyst/developer for the NASA Precipitation Processing System (PPS), used Python to convert the GRIB2 files for the first 36 forecast hours into JSON files containing the latitude, longitude, elevation, and reflectivity values for each point with a reflectivity greater than 30 dBZ. Once these JSON files are ingested into the website, PointPrimitives are created for each grid point. Using Cesium’s ScreenSpaceEventHandler, mouseover events on each point are captured. The moused-over point is “highlighted” (border color is changed and size is expanded) and its raw value and elevation are displayed in a text field in the top-right menu.
For more detailed investigation, users can add or remove levels of points based on their reflectivity values, enabling them to look more closely at the location and shape of the most intense portions of the storm. Using the time controls, users can advance in three-hour intervals through the forecasts. Since destroying PointPrimitives moving from one time interval to the next is one of the more time-consuming processes when displaying 100,000+ points at one time, the PointPrimitives are simply relocated and given a new color, with extra unused PointPrimitives being hidden from view.
While the other PPS Cesium demos, GPM NRT Viewer and STORM Virtual Globe, visualize satellite data, it is important to recognize the interplay between observed values and computer model forecasts. Cesium has the ability to enhance the ways in which gridded atmospheric model output can be presented and investigated online. Despite advancements in high-resolution modeling of extreme weather events, it remains common to look at the model output with static two-dimensional imagery. STORM VG HWRF shows that not only can three-dimensional interactive visualization easily be done with this sort of data, but that it can be done online and shared among the entire scientific community.