Why does NWP have so much sensor latency?
Sensor latency in meteorology refers to the amount of time between the actual measured weather phenomena and the publishing of weather information for consumption. A weathervane senses speed and wind direction and if it's read on the spot, the sensor latency is zero. Ground radar systems collect data which is presented on the screens of connected users almost immediately, but this information is also packaged up and published for use in other weather applications where the sensor latency can be 15 min or 15 hours. So, sensor latency can be caused by processing, packaging, or purely by the period of the application.
if we want the best, most current information, we need low sensor
latency. But weather forecasts look into the future, and they need lots
and lots of data to feed their super-sophisticated physical models of
the world. The supercomputers used by National Institutes for Weather
take up to 6 hours to crunch through all that sensor input in order to
create what's known as a Numerical Weather Prediction (NWP) that
forecasts weather for up to 14 days.
it takes 6 hours to run and because weather reality is changing all the
time, NWP updates are published every 6 hours. And while the
supercomputers are crunching away, all the sensor input is being
collected for the next update - for 6 hours. That means that by the
time an NWP update is published, the sensor latency for the inputs used
in that forecast is anywhere from 6 hours to 12 hours!
For near-term decision making (in aviation, this is known as the tactical decision making timeframe), 6-12 hours of sensor latency introduces a potential for error in the forecast that can be quite serious for critical decisions (whether they are risk mitigations or opportunities). And this is why critical decision makers develop skills in interpreting weather forecasts. Nowcasting with very low sensor latency can reduce that potential for error substantially, and make data-driven decision making much more viable.