Initialization of cloud water in a numerical weather prediction model

Abstract

In recent years many studies have shown the importance of treating condensation processes in a consistent manner in numerical weather prediction models. Among emerging improvements is the explicit treatment of cloud water, and in some cases precipitating water. An unresolved problem then is how to initialize the cloud water, especially since this quantity is not treated in the most commonly used analysis schemes. In this study, a method for initializing the cloud water in a numerical weather prediction (NWP) model will be presented and tested. The implications for the model's spin-up are investigated. Information from an earlier run (“first guess fields”) is used, together with satellite data. If necessary, humidity enhancement is performed where clouds are indicated by those sources. The results indicate that initialization of the cloud water field by itself does not have a large effect on the spin-up of precipitation and clouds. A much larger effect is obtained when the humidity field is enhanced. The spin-up time for precipitation is then reduced from 12 to 6 hours, while for cloud cover it is reduced to only 1–2 hours. The method is computationally very efficient, and is particularly useful over data-sparse areas, such as the oceans. An investigation of the different terms in the cloud water tendency equation is done and the results interpreted in terms of spin-up of cloud parameters. These tests confirm that the cloud water field only accounts for a small part of the spin-up effect. These also show that the production of cloud water per time step increases throughout the simulation.