These plots use the same data provided here. All anomalies are with respect to the lead-time dependent model climatology (1999-2010). The run date on the image should be yesterday’s date.
Weeks 3 and 4 can be expected to have much lower skill, much weaker signals, and much larger run-to-run variability than weeks 1 and 2. They should be interpreted in a more probabilistic sense.