Data assimilation (DA) is a method of combining observation data with

model forecast data in order to more accurately predict the state of a system.

One of its most common uses is in numerical weather prediction (NWP).

In NWP, we have observation data about different atmospheric properties,

such as pressure, temperature and wind speed, obtained from various sources,

such as ground stations, radiosondes, aircraft and satellites, at many different

locations. We also have details of the state of the system at an earlier time,

and a model which uses this to predict a forecast state. In the analysis

step of the DA system, this model forecast state is combined with recent

observation data to give an accurate estimate of the state of the system,

called the analysis state. This analysis state is then used to produce the

forecast state for the next analysis time step.