TERMINOLOGY FOR FORECAST SYSTEMS
Observation—measurement of a climate variable (e.g., temperature, wind speed). Observations are made in situ or remotely. Many remote observations are made from satellite-based instruments.
Statistical model—a model that has been mathematically fitted to observations of the climate system using random variables and their probability density functions.
Dynamical or Numerical model—a model that is based, primarily, on physical equations of motion, energy conservation, and equation(s) of state. Such models start from some initial state and evolve in time by updating the system according to physical equations.
Data assimilation—the process of combining predictions of the system with observations to obtain a best estimate of the state of the system. This state, known as an “analysis”, is used as initial conditions in the next numerical prediction of the system.
Operational forecasting—the process of issuing forecasts in real time, prior to the target period, on a fixed, regular schedule by a national meteorological and/or hydrological service.
Initial conditions/Initialization—Initial conditions are estimations of the state (usually based on observational estimates and/or data assimilation systems) that are used to start or initialize a forecast system. Initialization can include additional modification of the initial conditions to best suit the particular forecast system.