A Numerical Prediction Experiment Involving Data Paucity and Random Errors

500-mb prognoses computed with a barotropic model from progressively lessening amounts of data, showed increasing statistical differences. These differences also grew with the passage of time. With increasing paucity of data, the smaller meteorological features disappeared, while larger features were smoothed and suffered a change in speed of motion.

Random errors, which simulated a natural error source, were added to the data. This appeared to intensify and decelerate the features on the prognostic charts.