Modelización de la Atmósfera en Alta Resolución High Resolution Numerical Weather Prediction

Transcripción

1 Modelización de la Atmósfera en Alta Resolución High Resolution Numerical Weather Prediction Isabel Martínez Marco Jefe del Área de Aplicaciones AEMET

2 OUTLINE Introduction Forecast Ranges Operational Activities Observations Data Assimilation Numerical Formulation Physical Parametrizations Numerical Models of the Atmosphere ECMWF Model HIRLAM Models HARMONIE Model

3 Numerical weather prediction The behaviour of the atmosphere is governed by a set of physical laws which express how the air moves, the process of heating and cooling, the role of moisture, and so on. Equations cannot be solved analytically, numerical methods are needed. Given a description of the current state of the atmosphere, numerical models can be used to propagate this information forwards to produce a forecast for future weather. Additionally, knowledge of initial conditions of the system is necessary. An incomplete picture from observations can be completed by data assimilation. The resolution of the model is determined by available computing resources. It does not correspond to any natural scale separation.

4 Numerical Weather Prediction Processes not resolved by the model must be parametrized. Effective resolution is not the same as model grid spacing. Numerical algorithms are compromise between accuracy and speed; care needed to ensure numerical stability. Interactions between atmosphere and land/ocean are important.

5 Forecast ranges Short-range weather forecast (0-2 days ahead) Detailed prediction - regional forecasting system It produces forecast few hours after observations are made Medium-range weather forecast (2 days - 2 weeks ahead) Less detailed prediction - global forecasting system It produces forecast up to several hours after observations are made Long-range weather forecast (more than 2 weeks ahead) It predicts statistics of weather for the coming month or season Climate prediction It predicts the climate evolution on the basis of pre-defined scenarios (CO2, O3, )

6 Operational activities Observations Acquisition/Pre-processing/Quality control/bias correction Data assimilation Dynamical fit to observations Forecasts Product dissemination and archiving Verification Operational/Pre-operational validation Data Monitoring

7 Observations Conventional Surface: Weather stations (land+sea) Profiles: Radiosondes, UHF/VHF profilers Altitude: Aircraft (Ocean: TAO/PIRATA) Satellite Imagery -> Winds Radiances -> Temperature, Humidity, Sea wind (Precipitation) Ozone Scatterometers -> Sea wind

8 Conventional observations used SYNOP/METAR/SHIP: MSL Pressure, 10m-wind, 2m-Rel.Hum. DRIBU: MSL Pressure, Wind-10m Radiosonde balloons (TEMP): Wind, Temperature, Spec. Humidity PILOT/Profilers: Wind Note: We only use a limited number of the observed variables; especially over land. Aircraft: Wind, Temperature

9 Satellite data 13 Sounders: NOAA AMSU-A/B, HIRS, AIRS, IASI, MHS 3 Scatterometer sea winds: ERS, ASCAT, QuikSCAT 5 imagers: 3xSSM/I, AMSR-E, TMI Geostationary, 4 IR and 5 winds 4 ozone 2 Polar, winds: MODIS 6 GPS radio occultation

10 Significant increase in the number of observations assimilated Conventional and satellite data assimilated at ECMWF

11 Data Assimilation Observations measure the current state, but provide an incomplete picture Observations made at irregularly spaced points, often with large gaps Observations made at various times, not all at analysis time Observations have errors Many observations not directly of model variables The forecast model can be used to process the observations and produce a more complete picture (data assimilation) Start with previous analysis Use model to make short-range forecast for current analysis time Correct this background state using the new observations

12 Data Assimilation Observations= Background values = Analysis values = Analysis 00 UTC 5 May 12 UTC 5 May 00 UTC 6 May 12 UTC 6 May

13 Data assimilation for weather prediction The FORECAST is computed on a grid over the globe. The meteorological OBSERVATIONS can be taken at any location in the grid. The computer model s prediction of the atmosphere is compared against the available observations, in near real time. A short-range forecast provides an estimate of the atmosphere that is compared with the observations. The two kinds of information are combined to form a corrected atmospheric state: the analysis. Corrections are computed and applied twice per day. This process is called Data Assimilation.

14 4D-Var implementation All observations within a 12-hour period (~9,000,000) are used simultaneously in one global (iterative) estimation problem Observation minus model differences are computed at the observation time using the full forecast model. 4D-Var finds the 12-hour forecast evolution that optimally fits the available observations. A linearized forecast model is used in the minimization process based on the adjoint method. It does so by adjusting surface pressure, the upper-air fields of temperature, wind, specific humidity and ozone. 09Z 12Z 15Z 18Z 21Z

15 4D-Var is using more a-synoptic data than 3D-Var 4D-Var SYNOP Screening 3D-Var SYNOP Screening 4D-Var is using more data from frequently reporting stations. The plots show the use of SYNOP surface pressure observations. Column height gives the number of observations available, while the black part displays those actually used in the assimilation. 3D-Var is like 4D-Var without the time dimension. The analysis is performed at synoptic times only (0000, 0600, 1200 and 1800 UTC). Mostly only data valid a synoptic time is used.

16 4D-Var versus 3D-Var and Optimum Interpolation 4D-Var compares observations with background model fields at the correct time. 4D-Var can use observations from frequently reporting stations. The dynamics and physics of the forecast model are in an integral part of 4D-Var, so observations are used in a meteorologically more consistent way. 4D-Var combines observations at different times during the 4D-Var window in a way that reduces analysis error. 4D-Var propagates information horizontally and vertically in a meteorologically more consistent way. 4D-Var more complex: it needs linearized perturbation forecast model and its adjoint to solve the cost function minimization problem efficiently.

17 Numerical Formulation The Model equations Gas Law Hydrostatic equation Equation of Continuity Equations of Motion Thermodynamic equation Conservation Equation for moisture Conservation Equation for liquid water The Numerical Formulation Horizontal grid: finite differences, finite elements or spectral representations Vertical grid: finite differences or finite elements schemes Advection scheme: Semi-Lagrangian

18 Orography HIRLAM 0.16 (~16km) HIRLAM 0.05 (~5km) HARMONIE 2.5km

19 Vertical Resolution L60 L Position of levels and pressure layer thickness of L60 (blue) and L91 (red) Pressure (hpa) Level number levels 91 levels

20 Why parametrization Small scale processes are not resolved by large scale models, because they are sub-grid. The effect of the sub-grid process on the large scale can only be represented statistically. The procedure of expressing the effect of sub-grid process is called parametrization.

21 What is parametrization and why it is needed The standard Reynolds decomposition and averaging leads to co-variances that need closure or parametrization Radiation absorbed, scattered and emitted by molecules, aerosols and cloud droplets play an important role in the atmosphere and need parametrization. Cloud microphysical processes need parametrization Parametrization schemes express the effect of sub-grid processes in resolved variables. Model variables are U,V,T,q, (l,a)

22 Reynolds decomposition e.g. equation for potential temperature: ) ( z y x Q x w y v x u t advection source molecular diffusion Reynolds decomposition: '. ', ', ', w w W v v V u u U Averaged (e.g. over grid box): ) ' ' ( ) ' ' ( ) ' ' ( z w z y v y x u x Q x w y v x u t : source term (e.g. radiation absorption/emission or condensation) Q ' ' w : subgrid (Reynolds) transport term (e.g. due to turbulence, convection)

23 Physical processes

24 Importance of physical processes General Tendencies from sub-grid processes are substantial and contribute to the evolution of the atmosphere even in the short range. Diabatic processes drive the general circulation. Synoptic development Diabatic heating and friction influence synoptic development. Weather parameters Diurnal cycle Clouds, precipitation, fog Wind, gusts T and q at 2m level. Data assimilation Forward operators are needed for observations.

25 Numerical Models of the Atmosphere Horizontal resolution Vertical resolution Time range Climate models 200 km 500 m 100 years Global weather prediction Limited area weather pred. Cloud Resolving Models (CRM) Large Eddy Simulation (LES) 20 km 200 m 10 days 10 km 200 m 2 days 500 m 200 m 1 day 50 m m 5 hours Different models need different level of parametrization

26 Importance of physical processes and initial conditions - + Climate Models General Circulation Global Numerical Models Limited Area Numerical Models High-Resolution Meso-scale Models + - Initial Conditions External Forcings

27 The ECMWF Numerical Weather Prediction (NWP) Model High-resolution model T1279 spectral resolution 16 km global grid 91 hybrid levels from the surface to a height of 80km Variables at each grid point Wind Temperature Humidity Cloud water, ice, cloud fraction Ozone Pressure at surface

28 HIRLAM (High Resolution Limited Area Model) Model Formulation: Horizontal resolution: 0.16º latxlon (ONR) and 0.05º latxlon (HNR) Boundary Conditions: ONR: from ECMWF with 0.25º HNR and CNN: from ONR with 0.16º (nesting models) Analysis: 3-dimensional variational method (3D-VAR) The Resolution in space Vertical Resolution: 40 hybrid levels Horizontal grid: regular rotated longitude/latitude

29 Integration Domains

30 HIRLAM (High Resolution Limited Area Model) In development: HARMONIE Hirlam Aladin Regional/Meso-scale Operational NWP In Europe A new model formulation: Horizontal resolution: 2.5km Vertical resolution: 65 hybrid levels Analysis: 4-dimensional variational method (4D-VAR) Horizontal grid: spectral representation Vertical grid: finite differences Non-hydrostatic dynamical kernel from ALADIN Model

31 Productos de Predicción: