Improving Quantitative Precipitation Forecasts using Radar Data Observations
Quantitative precipitation estimation (QPE) still remains a challenging task despite the advances in numerical models and data assimilation systems. One of the main reason for that is the difficulty in capturing the inherent space–time variability of the hydrometeorological process. In that sense, radar observations are valuable source of information for improving prediction of convective events in NWP because they provide detailed information on its spatio-temporal evolution.
Although all radar data assimilation techniques have demonstrated some success, there is no obviously superior scheme. Notwithstanding all the efforts in mesoscale radar data assimilation, improvements only last for a few hours. These limitations can be due to the assimilation method itself as well as predictability limits of the meteorological situation at the mesoscale.
In order gain understanding on why improvements are short-lived, a simple humidity nudging assimilation scheme, Radar Mapping, is used in idealized experiments. Radar Mapping is a technique where the humidity in a convective parameterization scheme is adjusted in order to match precipitation observations. The convective triggering is altered by modifying the water vapor. Since the nudging is done in the parameterization the balance of the model is not altered. Short model runs transmits this information to the other variables of the system in an iterative manner. The underlying hypothesis is that the changes in model dynamics will result in a state closer to the nature. The improvements in QPE forecasts and the transmission of information to other variables than precipitation will be discussed.