developing cloud paramet erisat ions t he role of
play

Developing Cloud Paramet erisat ions - t he Role of Observat ions - - PowerPoint PPT Presentation

Developing Cloud Paramet erisat ions - t he Role of Observat ions - Clemens Simmer Met eorological I nst it ut e Universit y Bonn I nit ial t hought s... Cloud paramet ersat ions ... ... simulat e sub-scale cloud ef f ect s (geomet rical ext


  1. Developing Cloud Paramet erisat ions - t he Role of Observat ions - Clemens Simmer Met eorological I nst it ut e Universit y Bonn

  2. I nit ial t hought s... Cloud paramet ersat ions ... ... simulat e sub-scale cloud ef f ect s (geomet rical ext ensions+microphysics f or radiat ion and precipit at ion. ... were never developed direct ly f rom observat ions, ... are der ived f rom concept ual ideas about clouds (e.g. non-precipit at ing clouds exist ; t here are t rigger mechanisms f or convect ion, ...) ... are at best calibrat ed t o very limit ed observat ions Clouds are dif f erent (see classical cloud t ypes) ... some simple clouds led t o cloud par amet erisat ion concept s ... ... cloud par amet erisat ion relat e t o special cloud t ypes ... and must be biased when used in a generalized manner, as t hey are. Clouds are an int egral part of t he st at e of t he at mosphere ... ... but are t reat ed as an added-on, re-act ing phenomenon. ... inst ead cloud par amet erisat ions should be t wo-way-coupled wit h large- scale st at e, t urbulence, convect ion and radiat ion processes. -> isolat ed cloud paramet erisat ions are always incomplet e (meaning t hey hard t o validat e).

  3. Cont ent s • What are cloud paramet erisat ions? • How can observat ions be used t o calibrat e cloud paramet erisat ions? • What kind of observat ions do we really need, t o subst ant iat e or calibrat e cloud paramet erisat ions? • Are t here ot her ways t o use observat ions?

  4. What are cloud paramet erisat ions? ... calibrat ed f ormalized concept ual models about cloud processes and st ruct ur es at scales below t he models grid and t emporal resolut ions ... ... in order t o diagnose f ract ional cloud cover, cloud microphysical paramet ers (part icle number concent rat ions, cloud bulk densit ies, part icle size dist ribut ions,...) ... t o allow calculat ion of radiat ive ef f ect s, ... t o allow f or microphysical processes, i.e. precipit at ion simulat ions.

  5. Connect ions bet ween cloud and ot her paramet ersat ions (complet e physics package) Convect ion paramet ersat ion diagnose at mospheric mot ion ef f ect s, like energy, moment um, and mass f luxes on sub-grid scales cloud and convect ion paramet erisat ions must be physically very st rongly relat ed, but t hey are t radit ionally t reat ed independent ly. The same holds f or t urbulence and radiat ion modules, and also includes t he core model.

  6. Types of cloud paramet erisat ions • Det erminist ic schemes gridscale values lead t o unique set s cloud paramet ers • probabilist ic schemes gridscale values lead t o dist ribut ions of cloud paramet ers

  7. Cont ent s • What are cloud paramet erisat ions? • How are observat ions used t o aid cloud paramet erisat ions? • What kind of observat ions do we really need, t o subst ant iat e or calibrat e cloud paramet erisat ions? • Are t here ot her ways t o use observat ions?

  8. Ways t o aid t he development of cloud paramet erisat ions • Proof of concept ...needs dedicat ed experiment set ups • Calibrat ion of f ormalized concept s ... can of t en be achieved wit h t radit ional experiment set ups and long-t erm measurement s (but you need t o very caref ul)

  9. Cont ent s • What are cloud paramet erisat ions? • How are observat ions used t o calibrat e cloud paramet erisat ions? • What kind of observat ions do we need, t o subst ant iat e or calibrat e cloud paramet erisat ions? • Are t here ot her ways t o use observat ions?

  10. Opt imal measurement s • Cont inuous long t erm high t emporal resolut ion measurement s are indispensible f or st at ist ical reasons • Surf ace radiat ion measurement s (F) • Temperat ure (T(z) ) and humidit y prof ile (RH(z) ) f rom radiosondes ( RS ), radioacoust ic sounding syst ems ( RASS ), or microwave prof ilers ( MWP ) • Passive microwaves ( MWR ) f or t ot al wat er vapour (W) and cloud liquid wat er pat h (LWP) • Precipit at ion radar ( PR ) f or in-cloud precipit at ion-size part icles (RRt y(z)) • FMCW-radar ( MRR ) f or precipit at ion part icle size dist ribut ions (N(DRRz)) • Cloud radar ( CR ) and laser-ceilomet er ( LC ) f or cloud cover (N(z)) • MWR + CR + LC f or LWC(z), I WC(z) • Aircraf t measurement s f or cloud microphysiscs, wat er vapor variat ions, t urbulence... • Scanning wat er vapor lidar t o det et ct cont inuously spat ial and t emporal wat er vapor variat ions • High t emporal f ields of cloud paramet ers f rom sat ellit es • High qualit y f orcing f ields (analysis)

  11. Similar t o BBC(1) + anot her aircraf t + Raman lidar + micro rain radar(s) + (growing like ...)

  12. What did we learn f rom CLI WA-Net • Percept ion/ assumpt ions of clouds f rom modellers and observers can be very dif f erent (e.g. LWP wit h/ wit hout drizzle or rain, what is a cloud, what is cloud cover). • Modellers always t hink, t hat measurement s have no errors, at least t hey assume, t he are gaussian). When t hey lear n about errors t hey t end t o discard any measurement s. • Bot h models and obser vat ions are biased in very dif f erent ways (daily variat ions, pr ecipit at ion) leading t o dif f erent ly biased st at ist ics. • The impact of measurement s on paramet erisat ions was nil, unt il conf idence was est ablished bet ween modeller s and observers (modellers need t o underst and measurement s and vice ver sa).

  13. Specif ic result s of CLI WA-Net • LWP-f ields wit h reasonable error (30%) f rom sat ellit es available f or model comparisons • High-qualit y (Low-cost ) prof iler (radiomet er) available f or ground-based LWP net work • Algorit hms f or condensed wat er prof iles f rom ground-based synerget ic measurement s (cloud radar + microwave prof ile + laser ceilomet er) • Assessment of cloud paramet ers f rom st at e-of - t he–art at mospheric models • Preliminary quant if icat ions of model short comings and errors in assumpt ions in cloud paramet erisat ions (e.g. cloud overlap assumpt ions)

  14. BBC-Cabauw: Measured and model predict ed ver t ical dist ribut ion of liquid wat er cont ent , LWC(z)

  15. 1 August 2001

  16. 13 August 2001

  17. 14 September 2001

  18. RCA model: Impact of vertical model resolution: 24Levels 40Levels 60Levels

  19. Budgets and fluxes I Reference values: relative deviations 7km run without runs with 7 , 2.8 and Example: 13. April 2001 convection scheme 1.1 km grid spacings average over model domain and 24h 200 150 Water vapour Results: 100 Results: 6.5kg/m 2 50 0 • water vapour, 200 • water vapour, Ice 150 200 100 cloud cover and 3.4g/m 2 cloud cover and 50 150 surface fluxes 0 surface fluxes Cloud 100 300 250 remain unchanged 50 Liquid water cover 59% remain unchanged 200 0 7.6g/m 2 150 • LWP and rain rate • LWP and rain rate 100 50 increase due to increase due to Radiation 0 Heatflux refinement Evapo. 80W/m 2 refinement Rain 36W/m 2 99W/m 2 0.9mm/d 200 200 200 200 150 150 150 150 100 100 100 100 50 50 50 50 0 0 0 0

  20. Nonlinear LWP-rain relation grid refinement more LWP variations nonlinear LWP-RR relation Result of LM cloud scheme more RAIN ! using idealized cloud profiles

  21. 200 175 150 LWP histograms 125 100 Example 13 April 01 75 50 25 0 averaged rainfall Domain average Cabauw Probably poor statistic!

  22. Cont ent s • What are cloud paramet erisat ions? • How are observat ions used t o calibrat e cloud paramet erisat ions? • What kind of observat ions do we really need, t o subst ant iat e or calibrat e cloud paramet erisat ions? • Are t here ot her ways t o use observat ions?

  23. Are t here ot her ways...? • Proof of exist ing concept s ->dedicat ed experiment s (or dedicat ed analysis of exist ing experiment s) f or clearly def ined cloud t ype concept s in order t o prove or even bet t er t o f alsif y t he concept • St at ist ical- probabilist ic approach (e.g. neural net works) wit hout init ial concept s ->very many dat a, very long t ime series, analysis might , or might not lead t o new (or old) concept s

Recommend


More recommend