© Crown copyright Met Office The whole atmosphere project for the Earth: goals and challenges David Jackson General Circulation Models applied to Exoplanets

© Crown copyright Met Office

The whole atmosphere project for the Earth: goals and challenges

David JacksonGeneral Circulation Models applied to Exoplanets. Exeter Uni. Workshop, 5-7/12/2011


Overview

• What is Space Weather?

• How well can we forecast it and what can we do to improve forecasts?

• What is “whole atmosphere” modelling?

• Motivation

• Options for modelling the surface-thermosphere

• Goals and Challenges

• R&D roadmap – how to make the research tractable

• Summary


What is Space Weather?

• Severe disturbances (“solar storms”) in the upper atmosphere and near-space environment that can disrupt technology

• Space weather impacts are generally higher around solar maximum

Geomagnetic storms – related to CMEs which enhance solar wind and overcome magnetospheric shielding to reach the ionosphereSolar energetic particle events – bursts of high energy charged particles - leading to enhanced radiation levels in Earth’s atmosphereRadio blackouts (solar flares) – ionospheric disturbances via solar X-ray, UV, EUV emissions


Space Weather Impacts

•Power grids (Geomagnetically Induced Currents damage transformers)

•Aviation (avionics, radiation, GNSS, HF)

•Satellites (hardware damage, drag)

•Other GNSS / HF applications, or which may be affected by GICs:• Surveying, Pipelines, Railways, Finance, etc, etc

Worst Case… (Carrington storm)

$1-2 Trillion – Cost of blackout

4-10 years - Recovery time

•Many parts of UK power grid out for 12 hrs (or many weeks if supplies of spare transformers run out)

•30% of comms satellites knocked out


Space Weather Forecast Requirement

•Accurate forecasts with long enough lead times for users (eg power generation companies) to respond is highly important

•But is this available now?


Forecast Quality Depends on Space Weather Type

Storm Type Transit time from Sun

Current Predictive Capability

Potential for future improvement

Geomagnetic 20 h – 4 days

Good but limited – Solar wind model + ACE obs + iono. nowcast

Good - Coupled models (solar wind (quite mature) / geospace – surface)

SEP events 30 mins – 1 day

Fair – obs + iono. nowcast

(less) Good - Coupled models (energetic particle transport model (immature) / geospace – surface)

Radio blackouts 8 minutes Poor Fair- Coupled models (solar irradiance prediction model (immature) / geospace – surface)


Solar Wind Disturbance Propagation Model (1-4 days lead time)

Geospace Model (lead time as above)

Coupled “Sun to Earth” models offer great scope for better Space Weather Predictions

Whole Atmosphere Model (surface to Thermosphere / Ionosphere)

(lead time as above)

Solar GCM (> 4 days lead time)

Solar Wind Disturbance Propagation Model (1-4 days lead time)

Ionosphere model (nowcast)

No prediction of solar events

No indication of CME “geoffectiveness”

ACE observations (L1) – “geoffectiveness” (30-45 mins lead time)

No lower atmosphere coupling (~20% of ionospheric variability)

FUTURE NOW

( )

(Geomagnetic storms example)


Ionosphere & Thermosphere

• Such a coupled model system is complex to develop. But we can start by focusing on the more tractable parts – developing a “whole atmosphere model” that spans the surface to the thermosphere.

• Assessment of many Space Weather impacts requires knowledge of the state of the ionosphere & thermosphere – so this is a logical place to start

The thermosphere and ionosphere are collocated, but are defined by temperature and ionisation, respectively.


What is “whole atmosphere” modelling?

• Model spanning the surface to thermosphere

• Height range limited to region where Navier-Stokes equations are valid – the exobase (~600-650 km). Above here molecules follow ballistic trajectories and air no longer can be treated as continuous fluid

• Models usually evolve from NWP/Climate models (full dynamics, physics) with additional thermospheric dynamics and physics (eg WAM derives from GFS)

• Coupled ionosphere (on “flux tubes” following field lines). Integrated on neutral atmosphere grid is another option.


Motivation

• Non-migrating tides (DE3) important for w4 structure in EIA

• non-migrating tide generation linked to zonally-asymmetric forcing due to tropical convection?

• EIA w4 pattern weaker when DE3 weaker

Impact of lower atmosphere on thermosphere & ionosphere

IMAGE composite of 135.6 nm O airglow (350-400km) in March April 2002 and modelled diurnal temperature oscillation at 115 km - Immel et al, 2006


Motivation

• SEPs lead to NOx increase and O3 decrease in mesosphere / stratosphere

• Impact on stratospheric radiation budget and climatology in longer simulations

• Stratospheric ozone changes may affect tropospheric forecasts at medium/extended range, but further research is needed to understand this (eg Simpson et al, 2010 J. Climate)

• Solar cycle influence on European winter climate (Ineson et al, 2011, Nature Geosci.)

Impact of thermosphere & ionosphere on

lower atmosphere

Ozone changes above 60°N magnetic latitude. Below each representation of the measurements the corresponding model results are shown. (Rohen et al, 2005 JGR)


Goals

• Develop a Whole Atmosphere Model of the surface to thermosphere (plus coupled ionosphere model)

• Longer term – Incorporate Whole Atmosphere Model into “Sun to Earth” system (other UK / international partners working on the other models)

• Even longer term – develop data assimilation for the coupled system – necessary for the eventual development of an operational forecast system


Challenges

• The Thermosphere is different! – how will model dynamics deal with lower density and higher theta?

• Meridional winds (often associated with tides) large

• Vertical velocities often similar to horizontal


Challenges• Flares and geomagnetic storms can cause rapid

changes (50-300%) in neutral density


Challenges

• Need to represent the ionosphere and its interactions with the thermosphere

• The O/N2 ratio influences the plasma density of the F-region;hence regions of enhanced O/N2 tend to have higher plasma densities, and vice-versa

• Therefore, seasonal-latitudinal and longitudinal variations in O/N2 ratio also tend to be reflected in F-layer plasma densities.

• Need to represent constituents:– •separate equations above the turbopause.

•Chemical changes important


Challenges

• Model design:

• Simply couple existing models (UM, CMAT2) or merge them?

• What dynamics, physics, ionosphere to use?


Current UK Models

Model Met Office Unified Model (UM) CMAT2

Range Currently 0- ~80-85 km. 15 or 80 km - ~300-450 km (neutral atmosphere)

Physics No upper atmosphere physics (eg non-LTE radiation) or ionosphere model

Thermosphere physics / chemistry plus coupled ionosphere model

Dynamics Dynamical formulation suited to upper levels

Non-hydrostatic, deep atmosphere formulation (g(z), r≠a)

But: 1) Need to test robustness of solver, model formulation (ρ`,Θ`) 2) New equations for constituents needed above turbopause

Hydrostatic, shallow atmosphere formulation

Robustness

Used operationally at Met Office and other NMSs; widely used in academic community

Questions on robustness / portability. No operational usage, limited success in running on a supercomputer


Options for modelling the surface-thermosphere• a) couple the existing UM to CMAT2

• on the face of it seems a fairly cheap and easy option.

• However, CMAT2 is a hydrostatic model, whereas the UM is non-hydrostatic, so there would be a fundamental incompatibility in the coupling.

• The two models are the wrong way round: you would want the deep-atmosphere (non-hydrostatic) model on top of the (hydrostatic) shallow-atmosphere one, rather than the reverse.

• Met Office view: non-starter of an approach, which would be costly to develop and, scientifically, fundamentally flawed.

• b) extend the UM up to the thermosphere, with a top level of ~600 km

• based on the above, this is the way to go


Extended UM Option

• But..Why even do this? Could use existing whole atmosphere models (eg WAM (NOAA))

• Proposed next generation UM (EndGame) is potentially very suitable for this work - (non-hydrostatic, deep atmosphere). WAM has hydrostatic formulation (less preferred)

• after a sudden intense enhancement of high-latitude Joule heating, the magnitude of the vertical wind perturbation increases with altitude and reaches 150 (250) m/s at 300 (430) km during the disturbance. These large vertical winds are not typically reproduced by hydrostatic models of the thermosphere and ionosphere. The results give an explanation of the cause of such strong vertical winds reported in many observations (Deng et al, 2008)

• Strategically and scientifically it’s better to have a wider range of models, and especially a UK one


Physics, Chemistry & Ionosphere options

• Thermosphere-specific physics and chemistry

• Non-LTE radiation

• Joule heating & ion drag

• representation of E field

• N2, O2, O (and other constituents) chemistry

• Ionosphere

• Usually coupled to neutral atmosphere model

• Model co-ordinates

• At this stage – just import from eg CMAT2 or WAM. Further developments (in assoc. with collaborators) can be incorporated later, once “community model” is built


R&D RoadMap: How to make the development more feasible

Focus on Dynamical Core to start with

1) Develop TestBed - stripped down EndGame (runs easily on a desktop):

•No Physics – so very quick and simple•Use WAM initial conditions and dT/dt from WAM as forcing •Later, use range of ICs from MSIS and dT/dt model (based on Newtonian cooling, etc)•Develop standard dynamics tests (eg acoustic wave resolution, Held-Suarez) as needed.


2) Examine existing dynamical core

• Setup and run a range of trials with extended model (aim for ~600 km upper boundary):

• robustness to resolution, upper boundary, time stepping

• if robust:

• trial for different seasons and solar and geomagnetic forcings

• run dynamics tests (eg Held-Suarez test)

• if not robust:

• Diagnose problems. Propose and implement solutions (eg add molecular diffusivity to implicit timestep, option for implicit horizontal diffusion at upper levels)


3) Develop thermosphere-specific code

• Above turbopause (~120km) turbulent diffusion no longer dominates: separate equations for each constituent are required

• Set up framework and code for main constituents (O, N2, O2) to start with.

• Add simple chemistry, mutual molecular diffusion, turbulent mixing to tracer transport equations

4) Transfer to full EndGame model

5) Add in existing (full) physics / chemistry / ionosphere


Summary

• For improved space weather forecasts in the long term, a coupled Sun-to-Earth modelling system is needed

• An important component is a Whole Atmosphere model – enables the lower atmosphere – thermosphere / ionosphere coupling to be well represented.

• Initial model development should focus on dynamical issues:

• Use Test Bed first

• Check robustness of dynamical core, and modify if needed

• Add in thermospheric constituent equations and trial full model

• Add in comprehensive physics, chemistry and coupled ionosphere model later


Questions and answers

Documents

© Crown copyright Met Office The whole atmosphere project for the Earth: goals and challenges David Jackson General Circulation Models applied to Exoplanets