How ARM observations are improving cloud, precipitation and radiation prediction in the ECMWF global NWP model

Poster PDF

Description

The focus of the ASR-funded work at ECMWF in recent years has been on the process-oriented model evaluation of the global Integrated Forecast System (IFS) with ARM observations. A key advantage of the ARM program is the synergy of cloud, precipitation and radiation observations it provides, which are well suited to identify systematic and compensating errors in the model's parameterizations of cloud and radiative processes. A global weather forecast model has to be able to represent clouds and their radiative impacts for all meteorological regimes. It is therefore important to evaluate different cloud regimes in different regions of the world in order to identify viable improvements for the model parameterizations. Some of the key results from recent evaluation of the IFS with ARM data are summarised here. (a) A study of continental low clouds at the SGP site revealed compensating surface radiation biases in overcast and broken low cloud conditions and showed that the model overestimates the liquid effective radius. This sparked a review of the triggering and interaction of the boundary layer and shallow convection schemes. (b) In the Arctic, a case study from MPACE served to test and evaluate how a new prognostic microphysics scheme performed for mixed-phase clouds, and informed further changes to the treatment of ice deposition. The model's representation of supercooled liquid layers at the top of boundary layer cloud was significantly improved leading to reduced errors in 2m temperatures over the northern hemispheric wintertime continents and reduced radiation errors over the Southern Ocean, a region that many GCMs have difficulty representing correctly. (c) The long term observations from the mobile facility on Graciosa Island provided an opportunity to tackle the overestimate of light rain (drizzle) occurrence, another issue common to many GCMs. New parametrizations of autoconversion, accretion and evaporation were tested, leading to a reduction in occurrence of drizzle and improved prediction of low cloud cover. (d) Ice cloud retrievals from the Tropical West Pacific sites were used to assess the model's tropical ice cloud occurrence and ice water contents. The model's treatment of precipitating ice and the sub-grid distribution of cloud ice and snow was found to be a source of large uncertainty in the evaluation. Indeed, deficiencies in the treatment of spatial heterogeneity in the model emerged as a common theme from all of the evaluation studies.