The LaggedLinear Regression (LLR) method is a statistical forecast that uses information from the previous 15 days as input. Based on historical independent data, relative weights are determined for each daily forecast lead for various combinations of daily lag and mode. The two modes are those outlined in WH2004 (RMM1 and RMM2) and 15 daily lags are used. The principal component time series (PCs) for each mode are forecast using the following relationship: The forecast in varying displays are shown below with recent and historical verification information.

Phase Diagram 

Phase diagram for last 40 days of observations with the LLR forecast (red line) for the next 15 days appended to the time series. The thick line is for the first 7 days while the thin line is for the remaining period.

Spatial OLR 

Spatial OLR map reconstructed from the forecast of RMM1 and RMM2 for the next 15 days using the LLR method. Blue shaded show
negative OLR anomalies and enhanced convection and yellow/red shades show positive anomalies and suppressed convection.

Recent Verification 

Phase diagram illustrating the recent verification of the LLR forecast. The solid line is the forecast from the date on the figure and the thin lines are the subsequent observations during the period. Correlation during the period is shown in the corner.

Historical Verification 

Historical verification of the LLR method via anomaly correlation for leads from 1 to 15 days. The validation period is from 19791989
forecast data.
