Estimating Scale-Dependent Temporal Variability

We present a method for the scale distribution of global circulation variability in weather and climate models in comparison with reanalysis data. The method application to the low resolution ERA-20C reanalysis data provides the interannual variability spectrum with around one-quarter of the meridionally and vertically integrated global variability in the zonal mean state. The smaller the scale, the less variability and in scales with zonal wavenumber greater than 10 there is only about 10% of the total interannual variability. The reanalysis variability spectrum is used to validate a simplified general circulation model SPEEDY at the same horizontal and vertical resolution. The results quantify the underestimation of interannual variability in the model at all scales, but especially in largest scales. The scale distribution of normalized cumulative interannual variability for each season and annual mean is about the same as in reanalysis, suggesting it is the climate system property. Deficiencies in temporal variability are accompanied by deficiencies in the spatial variability in the same wavenumbers. The model has relatively less energy in waves and its annual cycle of energy in planetary-scale waves is not simulated well. This is not surprising as the model has been tuned to reproduce well the annual average precipitation and surface temperature fields. The presented method is suggested for climate model validation.