An Assessment of the ECMWF Model over the Arctic Land Using Observations from the Mixed-Phase Arctic Cloud Experiment

Xie, S., Lawrence Livermore National Laboratory

General Circulation and Single Column Models/Parameterizations

Cloud Modeling

Xie, S., S. A. Klein, J. J. Yio, A. C. M. Beljaars, C. N. Long, and M. Zhang, (2006): An Assessment of the ECMWF Model over the Arctic Land Using Observations from the ARM Mixed-Phase Arctic Cloud Experiment, JGR, 111, D05107, doi:10.1029/2005JD006509.


Figure 1. Time-height cross sections of (a) the ARSCL, and (b) model clouds (%) at Barrow during M-PACE.


Figure 2. Time series of the domain-averaged observed and model-produced surface downwelling longwave radiative fluxes during M-PACE.


Figure 1. Time-height cross sections of (a) the ARSCL, and (b) model clouds (%) at Barrow during M-PACE.

Figure 2. Time series of the domain-averaged observed and model-produced surface downwelling longwave radiative fluxes during M-PACE.

This paper uses the measurements available from the ARM Climate Research Facility (ACRF) Mixed-Phase Arctic Cloud Experiment to assess the European Centre for Medium-Range Weather Forecasts (ECMWF) model Arctic performance.

The model-measurement differences are typically the smallest when the comparison is made over the analysis domain, suggesting the discrepancy between model and observations can be an overestimated when comparing model data with single point measurements, especially at stations near the coastline where the nearby model output grid points may cover both ocean and land.

It is shown that the model analysis is able to represent well the temporal evolution and vertical structure of the large-scale synoptic systems during M-PACE. The model does exhibit a near-surface warm bias (less than 2O°C) where boundary layer clouds were observed, and a small systematic dry bias (less than 5%) in the entire troposphere. This dry bias could be partially alleviated by allowing ice supersaturation in the model.

The ECMWF model does well in capturing various cloud types observed during M-PACE. However, the model boundary layer clouds contain much less liquid water than the observed, which may be due to the model difficulties in simulating the microphysics of mixed phase. This problem in simulating the mixed-phase boundary layer clouds directly results in a substantial underestimation of the surface downwelling longwave and an overestimation of the TOA outgoing longwave in the model. Due to the underestimation of the surface downwelling longwave, the model shows a much larger energy loss (-20.9 Wm-2) than the observations (-9.6 Wm-2) at the surface during the M-PACE period.