Weather Forecasts Help to Understand Climate Model Biases

Klein, S., Lawrence Livermore National Laboratory

General Circulation and Single Column Models/Parameterizations

Cloud Modeling

Klein, Stephen A., X. Jiang, J. Boyle, S. Malyshev, and S. Xie, 2006: Diagnosis of the summertime warm and dry bias over the U. S. Southern Great Plains in the GFDL climate model using a weather forecasting approach. Geophys. Res. Lett., 33, L18805, doi:10.1029/2006GL027567.

Jiang, X., N.-C. Lau, and S. A. Klein, 2006: Role of eastward propagating convection systems in the diurnal cycle and seasonal mean of summertime rainfall over the U. S. Great Plains. Geophys. Res. Lett., 33, L19809, doi:10.1029/2006GL027022.

Figure 1

Figure 2

Figure 1

Figure 2

Two of the most important simulated variables of a climate model are the surface air temperature and precipitation. Climate models exhibit important biases relative to observations in both quantities yet an understanding of the causes of the biases is often lacking. The climate model of the Geophysical Fluid Dynamics Laboratory in Princeton, New Jersey has a warm bias of surface air temperature over the central United States in summer. At the same time the climate model strongly underestimates the amount of precipitation. Both biases are large at the location of the ARM Southern Great Plains site (indicated by the X in the figure) suggesting that ARM observations could aid in understanding the cause of the model bias.

Yet understanding the cause of these biases is difficult because of feedbacks between the land surface and atmosphere. Reduced precipitation may result if evaporation is suppressed due to below normal soil moisture. But, below normal precipitation may also be the cause of the below normal soil moisture. Having a means to separate initial errors from amplifying feedbacks would be useful.

For this reason, the approach of weather forecasting is attractive. If the state of the atmosphere and land model can be initialized with observations, it may be possible to diagnose the process behind the drift towards a biased climate. Researchers at Lawrence Livermore National Laboratory working in the ARM-sponsored CAPT project performed a series of weather forecasts using the GFDL climate model initialized with analyses from a weather prediction center. Each forecast was 3 days in length and a forecast was begun every day in the months of June and July 1997 when ARM conducted an intensive observing period at its SGP site.

These forecasts showed that the precipitation bias of the model was present in the first day of the forecasts where as the temperature bias was much smaller. At the SGP site, the model does not do a bad job of predicting the timing of precipitation but it fails to simulate the strong precipitation events observed by ARM.

ARM observations also showed that the model overestimated the solar radiation reaching the surface which may be explain some of the initial positive temperature bias.

The reason that the temperature bias is larger in climate than in the weather forecasts is that the underestimate of precipitation in the forecasts causes the soil to dry out unreasonably which then causes the soil to warm unrealistically when evaporation of soil moisture is no longer possible to balance the incoming solar radiation. Thus, the weather forecasts have shown that the precipitation bias is not primarily the result of land surface feedbacks but is mostly present before these feedbacks can operate.

The technique of weather forecasting is valuable in deciphering the way that climate model biases develop. For the GFDL model, the underestimate of precipitation is the key factor explaining the overestimate of surface temperature.

More work is necessary to determine why the GFDL model underestimates the precipitation. Because much of the precipitation is at night associated with propagating convective systems, climate models will need to improve the simulation of these systems if they wish to improve their simulation of summertime climate over North America.