Apples and oranges

It is often implicitly assumed that over suitably long periods the mean of observations and models should be comparable, even if they have different temporal sampling. We have assessed the errors incurred due to ignoring temporal sampling and show they are of similar magnitude as (but smaller than) actual model errors (20–60%).

Using temporal sampling from remote sensing datasets, the satellite imager MODIS (MODerate resolution Imaging Spectroradiometer) and the ground-based sun photometer network AERONET (AErosol Robotic NETwork), and three different global aerosol models, we compare annual and monthly averages of full model data to sampled model data in Schutgens et al. Atmos. Chem. Phys. 2016.

Figure: Zonal and annual averages of AOT from the satellite sensor MODIS Aqua (black) or the ECHAM-HAM model (red). The model average is either a normal average (solid) or an based on model data collocated with the observations (dotted).

Figure: Comparison of model errors (red), temporal sampling errors (blue) and observational errors (grey) across different time-scales. Statistics based on global data from MODIS Aqua and the model ECHAM-HAM.

Not only do our results show the importance of sub-sampling model data to observations to prevent unnecessary noise in the evaluation, they also point to substantial differences in the temporal evolution of AOT across models that have similar annual means. This suggests new ways to evaluate models, potentially allowing process studies.

Figure: Relative contribution of the diurnal cycle to temporal variability in AOT across three models.