The assimilation of Doppler radial wind data into operational Numerical Weather Prediction models presents formidable scientific challenges

<< Fathalla Rihan>>

Assimilation of Doppler radar data into Met Office Unified Model

Fathalla Rihan, Chris Collier, Sue Ballard & Sean Swarbrick

Introduction:

Numerical Weather Prediction (NWP) is an initial-boundary value problem: given an estimate of the present state of the atmosphere, the model simulates (forecasts) its evolution. The goal of Data Assimilation (DA) is to construct the best of the Initial Conditions (IC), known as the analysis, from which to integrate the NWP model forward in time. The observations are typically irregular distributed in space and time, the DA involves deriving the best current state of the atmosphere by combining the model forecast background (the most recent information) with observations giving each a weighting
which depends upon their characteristics.

A Doppler radar uses electromagnetic waves to investigate atmospheric properties: the amplitude of waves are used to estimate the reflectivity and phase of the waves are used to estimate the radial wind. The radial velocity of scattering particles is determined from their observed phase difference between successive radar pulses.

Limited Area Models (LAMs) require observations with high spatial-temporal resolution to determine the ICs. Doppler radar has the ability to scan large volumes of atmosphere with high resolutions. It provides measurements of radial velocity and reflectivity with high resolution. Doppler winds give extra information in forecasting of quickly developing mesoscale systems. The resolution of radar data is however much higher than the resolution of the NWP models.

Assimilation of Doppler data into operational NWP models presents formidable scientific challenges. The aim of this project is to develop an incremental assimilation of Doppler radar wind data into UM using 3D-Var/4D-Var, leading to an improved representation of the Initial Condition (ICs) for storm-scale forecasting.

The data used are based on PPI volume scans (Each PPI- Plan Position Indicator-is taken at a single, fixed elevation angle, and thus forms a cone of coverage in space) of Chilbolton radar. Reflectivity and radial wind are saved as a function of range, azimuth, and elevation. Some further process is required before assimilation, via averaging of raw data to grid size of the model.

Main Challenges:

Quantitative precipitation forecasts.
Accurate estimate forecasts.
Data assimilation at high resolution.
Loss of vertical resolution of the data.
Nonlinearly and uncertainty of observation operator.
Preprocessing of the Doppler radar data that includes data quality control, generation of super-observations, and error statistics of the doppler radar data.

Observational Needs:

Radar radial velocity and reflectivity data (surface observations).
Suitable filtering technique.

Project Steps:

Pre-processing Doppler data before assimilation: re-scale the radar data at a spatial resolution that corresponds with that used in the NWP model, using super-obbing technique.

Setting up an OSSE (Observing System Simulation Experiment) : assimilating pseudo data generated from a model forecast so that performance of the technique can be assessed by knowledge of truth.

Formulating the Observation Operator for radial velocity and reflectivity and their adjoints and insert them into the Unified Model.

Running Met Office 3D-Var system for Doppler radial velocity, having a time window of 3 hours at 5-km grided data resolution to produce possible best initial conditions for 4-10km resolution forecast.

The ultimate aim is to assimilate 1-5km resolution grided radial velocity and reflectivity data, using 4D-Var technique to produces ICs for 1-4km resolution forecast.

Investigating the impact of assimilating reflectivity combined with radial velocity.

Documentations:

Assimilation of Doppler Radar Data into Met Office UM (as PDF format)
Four-Dimensional Variational Data Assimilation for Doppler Radar Wind Data (as PDF format)

Slides: Assimilation of Doppler Radar Winds into Met Office Unified Model (as PDF format)

Visit to NCAR (Colorado) and Oklahoma Weather Center

Organizational Structure:

This project is managed jointly by Chris Collier (who is the UWERN Principal Investigator, Salford University), and Sue Ballard (JCMM). This work is carried out by Fathalla Rihan (Salford) and Sean Swarbrick (Met Office).

<<Go back to F.Rihan>>