The building maintenance scheduled for Friday February 27th at 5:00pm MST has been postponed. It is rescheduled for March 6rd.
 

Datasets for evaluating clouds, precipitation, and radiation in climate models

Overview

The page is the distribution point for datasets related to the paper "Evaluating the present-day simulation of clouds, precipitation, and radiation in climate models" by Robert Pincus, Crispian Batsone, Patrick Hofmann, Karl Taylor, and Peter Gleckler. The paper describes methods for evaluating the performance of climate models by comparing the simulation of the present-day distribution of clouds, radiation, and precipitation to global observations.

From this web page you may download the metrics that appear in the paper, as well as the data from which the metrics were computed. See the paper for technical details.

Metric scores for CMIP3 and other models

We computed metrics for three radiative quantities (net, longwave, and shortwave cloud radiative effect), surface precipitation, and column-integrated cloud fraction. We have also computed the scores for top-of-atmosphere net, longwave, and shortwave fluxes, though we think these are poor measures of model skill for reasons explained in the paper.

Five metrics are computed for each variable: mean bias, RMS error, centered RMS error, correlation coefficient, and the ratio of standard deviations. Computations are done on a 2.5 degree grid using the mean seasonal cycle.

Scores are available for all the models participating in the CMIP3 comparison that formed the basis of the Fourth Scientific Assessment of the IPCC. We also include scores for the ERA-40 reanalysis and a recent version of the forecasting system at the ECMWF, for the "super-parameterized" NCAR CAM, and for the "CMIP3 all-model mean" described in the paper. There are two classes of model runs: one set in which the sea surface temperature is specified (AMIP) and one in which the atmospheric model is coupled to an ocean model (20th C). Some models provided only one run.

Each metric is computed twice using different verifying observational data sets. You may download the scores as a netcdf file.

Model and observational climatologies

We are also providing the climatologies used to compute the performance metrics. These have been constructed by computing a set of twelve average months on a uniform 2.5 degree grid. You may

Citation

If you use this data in publication or presentations, please cite