A suite of six ECMWF AMIP runs (differing only in their initial conditions) have also been examined. The Indian monsoon rainfall exhibits a consistent response during 1987 and 1988, while during other years differences are simply not very predictable, possibly because of internal chaotic dynamics that are associated with intraseasonal monsoon fluctuations and/or land surface process interactions. In this case the Reproducibility is poor indicating that the average intramodel spread is greater than the temporal variability of the ensemble mean. For the Sahel and the Nordeste the Reproducibility increases to 2.4 and 9.0 respectively indicating a robust response to the boundary conditions for this model.
Five SUNY/NCAR Genesis low resolution initial condition realizations were also examined. In this model the Nordeste rainfall also exhibited the largest Reproducibility, however for all regions the Reproducibility was smaller than that for the ECMWF model. The relationship of the all-India and Sahel rainfall/SST teleconnections with horizontal resolution, convective closure and numerics has been evaluated. Models with resolution >=T42 performed more poorly than lower resolution models. The higher resolution models were predominantly spectral. At low resolution, spectral vs. gridpoint numerics performed with nearly equal verisimilitude. At low resolution, moisture convergence closure was slightly preferable to other closure techniques. At high resolution, the models that utilized a moisture convergence criteria performed very poorly suggesting that moisture convergence may be inappropriate for models with horizontal resolution >=T42. (pdf file)