High Resolution Analysis
Rapid updated high-resolution meteorological analysis is one of the key requirements for the improvement of forecasting various hazardous weather events. Operational Doppler radar network in the U.S. and many other developed countries are able to provide temporally and spatially high-resolution observations, but these remote sensing observations are limited only to radial wind and reflectivity. The complete meteorological variables of 3-dimensional wind, temperature, and microphysics must be retrieved from these limited observations.
At the National Center for Atmospheric Research (NCAR), scientists and engineers have endeavored to develop a fine-resolution analysis and nowcasting system, named FINECAST® (preiously known as Variational Doppler Radar Analysis System), based on high-density and -frequency observations such as those from Doppler radars and surface networks. The core technology of FINECAST is an advanced 4-dimensional variational (4DVar) data assimilation system constrained by a convection-permitting model. By making use of the observations at more than one time levels, the model trajectory within a specified time window is fitted to the observations and meteorological analyses consisting of observed and unobserved variables are obtained. These analyses cannot be obtained from operational NWP models because they are primarily based large-scale observations.
FINECAST is a result of many years of scientific research that was conducted at NCAR. It has been used by the research community to study convective weather initiation mechanisms as well as by the operational community to improve nowcasting. Since 1998, FINECAST has been implemented in real time for the applications of severe weather nowcasting, wind energy prediction, hazardous chemical detection, and model initialization. The code is fully parallelized by MPI (Multiple Processing Interface) to handle the intensive computation demand of the 4DVar scheme, such that each analysis can be produced in a few minutes with an update frequency of 10 minutes to meet the nowcasting requirement.
3D Picture of Convective Storm
A critical step toward improved nowcasting capability for convective storms is to understand their initiation and evolution mechanisms. FINECAST provides a unique tool that helps researchers and forecasters identify thunderstorm predictors and develop conceptual models for convective weather nowcasting. By optimally fitting model trajectories to observations using the 4DVar technique, dynamically consistent meteorological fields are retrieved. From the retrieved fields, dynamical and thermodynamical diagnostic fields, such as divergence, updraft/downdraft, CAPE, CIN, vertical wind shear, perturbation temperature (cold pool), helicity, etc, can be derived, which are important for characterizing and predicting convective storms. These diagnostic fields provide depictions of the mesoscale environment, the triggering mechanism, and the structure of convective storms. Some of these fields are being used in the automated heuristic nowcasting system AutoNowcaster to identify regions of convective initiation. Studies are also being conducted to examine their statistical correlations with observed reflectivity by radar for the purpose of nowcasting convective precipitations.
Nowcasting Wind and Precipitation
Other than using FINECAST analyses by forecasters or observation-based automated nowcasting systems, the analysis fields produced by FINECAST are also being utilized to initialize Numerical Weather Prediction (NWP) models for the improvement of precipitation and wind nowcasts. FINECAST analyses are essentially initial conditions for the cloud model used in FINECAST that constrains the 4DVar analyses, from which nowcasts can be launched by directly integrating the cloud model forward. FINECAST analyses have also been used to initialize other non-hydrostatic models, such as WRF (Weather Research and Forecasting) model.
Studies are being conducted to assimilate FINECAST analysis fields into WRF via WRF data assimilation schemes.
In collaboration with wind power industries, FINECAST is recently being examined for its ability in improving the challenging problem of nowcasting wind ramp events. Research is also being conducted to evaluate the capability of FINECAST in nowcasting convective precipitation. Our goal is to expand FINECAST into a new-generation of nowcasting system that is model-based and initialized by assimilating multiple platforms of high-resolution observations producing rapid updated nowcasts every 10 minutes or less.
On-Going and Future Developments
One of the recent new developments that expand the capability of FINECAST toward a model-based nowcasting system is the inclusion of terrain effect. A terrain scheme based on the Immersed Boundary Method (IBM) is being developed and tested. Recent studies have shown that the IBM terrain scheme is capable of simulating the updraft/downdraft and associated precipitations caused by complex terrain. The assimilation of polarimetric radar observations is another research activity that is believed to have the potential to improve the accuracy of FINECAST analyses and nowcasts. Through an upcoming collaborative project, FINECAST will be implemented with a high-resolution (~250m) configuration to assimilate both radar and lidar observations in an airport.