NASA - National Aeronautics and Space Administration

An iterative automated classification technique that combines European Centre for Medium-Range Weather Forecasts analysis data and vertically pointing millimeter wavelength cloud radar observations is used to identify commonly occurring atmospheric patterns or states around Darwin, Australia. The technique defines the atmospheric states by large-scale, synoptic variables such that, once defined, these states will be suitable to composite climate model output. Radar observations of clouds are used to test the statistical significance of each state and prompt the automated refinement of the states until each state produces a statistically stable and unique hydrometeor occurrence profile. The technique identifies eight atmospheric states: two monsoon states, two transition season states, and four dry season states. The two monsoon states can be identified as the active monsoon and the break monsoon. Among the dry season states, periods of isolated and suppressed convection can be identified. We use these states as the basis for compositing hydrometeor occurrence, precipitation rate, outgoing longwave radiation, and Madden-Julian Oscillation phase to further understand the meteorology of each state.