Transient Tropical Diabatic Heating and the Seasonal Mean Response to ENSO

Is the seasonal mean response to ENSO only determined by the seasonal mean redistribution of tropical diabatic heating? To address this, large ensembles of boreal winter simulations of the Community Atmospheric Model version 4.0 carried out using observed sea surface temperature fields from three El Niño events of 1982/83, 1991/92 and 1997/98 (CTL runs) were compared to four suites of simulations in which the diabatic heating rate Q across the Indo-Pacific region is specified from a filtered version of the CTL heating rates (Q_CTL) at every grid point, vertical level, and day. Each suite contains runs parallel to the CTL runs. In the first suite Q is replaced by the time invariant seasonal - ensemble mean of Q_CTL for each ensemble member (suite FIX); in the second suite Q is replaced by the seasonal mean Q_CTL from the appropriate ensemble member of the CTL (suite EFIX); in the third suite Q is replaced by the seasonal mean plus low-frequency component of Q_CTL (suite ESUBFIX); and in the fourth suite Q is replaced by the daily means of Q_CTL (suite DAYFIX). The mid-latitude ENSO anomalies of the seasonal mean upper level height field and time filtered meridional wind variance are enhanced in the FIX, EFIX and ESUBFIX suites, with little change in patterns, compared to CTL anomalies. The enhancements have a smaller magnitude in ESUBFIX and especially in DAYFIX; qualitative differences are seen in DAYFIX. These differences are due to (i) the required set-up time for mid-latitude response; (ii) the altered relationship between vertical structure and vertically integrated heating; and (iii) the lack of mid-latitude interactive influence on tropical heating. Some evidence indicating forcing of the tropical circulation by midlatitude fluctuations is seen.