87th AMS Annual Meeting

Monday, 15 January 2007: 11:45 AM
Assessing consistency between EOS MLS and ECMWF analyzed and forecast estimates of cloud ice
214B (Henry B. Gonzalez Convention Center)
Jui-Lin Li, JPL, Pasadena, CA; and D. E. Waliser, J. H. Jiang, and A. M. Tompkins
Measurements from the Earth Observing System's (EOS) Microwave Limb Sounder (MLS) are used to examine the fidelity of the ECMWF integrated forecasting system in representing upper-tropospheric ice water content (IWC). Comparisons are made with ECMWF analyses and forecasts for the period August 2004 to July 2005. To account for the sampling differences, the ECMWF data are sampled along the MLS measurement tracks and have a lower-limit threshold applied to account for the MLS instrument/algorithm sensitivity. From the annual mean values at 147 hPa, the spatial agreement between MLS and sampled ECMWF analyses is quite good, in particular over the oceans, but the analyses are biased high relative to MLS by about 10%. In contrast, over tropical landmasses, including the maritime continent region, ECMWF analyses are biased low relative to MLS by up to 50%. This underestimation grows in the forecasts, with a further ~40% reduction in IWC at 147 hPa across the global tropics by day 10, and with greater reductions occurring over the warm pool region. At 215hPa, the global average IWC undergoes very little systematic change, although the spatial distribution changes with IWC decreasing over the warm pool region as well as over the Americas. The temporal evolution of these biases, along with a systematic decrease (~50%) in the large-scale upward vertical velocity over the warm pool, indicate a lack of deep convection over the maritime continent and equatorial Africa and America. Comparisons of precipitation and radiative fluxes to observations appear to confirm this postulation. Half of the reduction in IWC occurs in the first 24 hours of the 10 day forecasts. Since the forecast biases reflect those in the analyses, it suggests that the analyses biases are at least in part due to shortcomings intrinsic to the model physics, in the assimilation system itself. These results demonstrate the potential usefulness of the MLS data set for evaluating GCM performance and will be disused in the meeting.

Supplementary URL: