Fine-Scale Analysis and Nowcast System (FINECAST)


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.

The core technology of FINECAST® is an advanced 4-dimensional variational (4DVar) data assimilation system constrained by a convection-permitting model.

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.