The three Lorenz (1963) coupled nonlinear ordinary differential equations are integrated forward from a set of initial conditions, to define the reference trajectory. A given number of observations are simulated from the reference trajectory and forecasts from a different set of initial conditions are combined with observational information through an ensemble Kalman filter (EnKF). The EnKF algorithm used here is based on a square root formulation, which avoids the need of perturbed observations. At the end of the assimilation time window, the program performs a forecast (both for the reference and the data assimilation run) to monitor the forecast accuracy as a function of time. The documentation discusses the purpose of various optional parameters and suggests a number of exercises.