Alternative methods for specification of observed forcing in single-columnmodels and cloud system models

We discuss alternative methods for prescribing advective tendencies in sing le-column models (SCMs) and cloud system models. These include "revealed fo rcing," in which the total advective tendency is prescribed from observatio ns; "horizontal advective forcing," in which the horizontal advective tende ncies are prescribed, together with the observed vertical motion which is c ombined with the predicted sounding to determine the tendencies due to vert ical advection: and "relaxation forcing," in which the horizontal advective tendencies are computed by relaxing the sounding toward the observed upstr eam sounding, with a relaxation timescale determined by the time required f or the wind to carry parcels across the grid column. When relaxation forcin g is used, the horizontal advective tendencies can be diagnosed from the mo del output and compared with the corresponding observed tendencies. We pres ent SCM results to illustrate these three forcing methods, based on data fr om several field experiments in both the tropics and the midlatitudes. Each method is shown to have its strengths and weaknesses. Overall, the results presented here do not show unambiguous differences between revealed forcin g and horizontal advective forcing. The two methods appear to be generally comparable. Revealed forcing may therefore be preferred for its simplicity. Relaxation forcing guarantees realistic soundings of the state variables b ut can produce large errors in parameterized processes which are driven by rates (e.g., fluxes) rather than states. In particular, relaxation forcing gives large errors in the precipitation rate in this model. We demonstrate that relaxation forcing leads to unrealistically high (low) precipitation i n versions of the model which tend to produce unrealistically dry (humid) s oundings. The observed horizontal advective tendencies in the tropics are s o weak, especially for temperature, that small absolute errors in the diaba tic tendencies diagnosed with relaxation forcing can lead to large relative errors in the diagnosed horizontal advective tendencies.