Microwave observations: cloud and precipitation; applications - Alan Geer - ECMWF May 6, 2021 - ECMWF ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Microwave observations: cloud and precipitation; applications Alan Geer © ECMWF May 6, 2021
Information content in cloud and precipitation © ECMWF May 6, 2021
Effect of hydrometeors in microwave window channels Increasing frequency [GHz] (h = horizontal polarization) Observed TB [K] 3
Effect of hydrometeors in microwave window channels Increasing frequency [GHz] (h = horizontal polarization) Observed TB [K] Hydrometeor effect: observed TB – Simulated clear-sky TB [K] Rain (absorption, increases TB) Cloud (absorption, Snow/graupel/hail (scattering, EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS increases TB) decreases TB) 4
Effect of hydrometeors in microwave sounding channels Temperature sounding: Water vapour sounding: Lower troposphere Mid troposphere Mid troposphere Upper troposphere Observed TB [K] 5
Effect of hydrometeors in microwave sounding channels Temperature sounding: Water vapour sounding: Lower troposphere Mid troposphere Mid troposphere Upper troposphere Observed TB [K] Hydrometeor effect: observed TB – Simulated clear-sky TB [K] Cloud and rain (absorption, Cloud and snow/ice/graupel Cloud (absorption, pushes up weighting function (absorption and scattering, increases TB) altitude, decreases TB) decreases TB) 6
Effect of hydrometeors Size parameter • 30 GHz frequency ↔ 10mm wavelength (λ) 2pr x= l Particle type Size range, r Size parameter, x Cloud droplets 5 – 50 μm 0.003 – 0.03 Drizzle ~100 μm 0.06 Rain drop 0.1 – 3 mm 0.06 – 1.8 Ice crystals 10 – 100 μm 0.006 – 0.06 Snow 1 – 10 mm 0.6 – 6 Hailstone ~10 mm 6 EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 7
Wavelength Cloud and 100 mm 10 mm 1 mm precipitation Slab cloud at 283K optical properties: above a 280K the microwave surface (solid) spectrum Slab cloud at 283K above a 100K surface (dashed) Geer et al. (2021, GMD, submitted) EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 8
To sensor Strong scattering at 91 GHz Reverse Monte- [km] Carlo radiative transfer solver Snow Cloud Rain [km] EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 9
Scattering radiative transfer Quick summary… © ECMWF May 6, 2021
Schwarzchild’s equation Planck function Radiance Change in radiance per = ! ( − ) distance Absorption coefficient Loss through − ! absorption + ! Gain through thermal emission EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 11
Adding scattering Extinction coefficient " = ! + # Scattering coefficient (describing the amount of scattering out of the beam) Loss through extinction − " Gain from scattering +? into the beam + ! Gain through thermal emission EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 12
Change in coordinates: optical depth Change in optical depth = − ! d in a non-scattering atmosphere Change in optical depth = −( ! + # ) = − " d including extinction by scattering EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 13
The full scattering radiative transfer equation Direction vector Integral over all solid angle - (Ω) # - = Ω − 1 − # − - Ω′) 3 (Ω, - - Ω′ Ω′ - 4 $% Single scattering Scattering phase Change in albedo (SSA) function, often simplified optical depth and summarised by ‘g’ – # asymmetry parameter # = ( ! + # ) • Without scattering, just integrate this equation along the path travelled by the radiation (Tony’s first lecture) • With scattering, this can be complex to solve: Ω$ , the radiance in one direction, depends on radiance from all other $ directions: Ω′ and all levels depend on each other EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 14
Fast • For each hydrometeor type (e.g. cloud, rain etc.), compute single-species bulk approximate optical properties as a function of mixing ratio solver: • Sum up single-species optical properties along with gas optical properties RTTOV- • Assume a simplified scattering phase-function (Henyey-Greenstein) SCATT • Assume no azimuth dependence of the radiance field • Assume radiance is a simple function of zenith angle (theta): = & + ' cos (Eddington approcimation, similar to the two-stream approach) • delta-scaling (treat sharp peaks in the scattering phase function as no scattering at all) • Set up coupled equations describing radiance all levels simultaneously; solve using matrix algebra for the approximate radiance field I at each level • Integrate the radiative transfer equation along the line of sight, using the delta- Eddington solution to provide the ‘source terms’ - (Ω) # - = Ω − 1 − # − - Ω′) 3 (Ω, - - Ω′ Ω′ - 4 $% EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 15
Optical properties of hydrometeors in RTTOV- SCATT: at 183 GHz Lookup tables for snow hydrometeors as a function of snow water content Geer et al. (2021, GMD, submitted) 16 EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS
Optical properties of hydrometeors in RTTOV- SCATT: across frequencies Lookup tables for snow hydrometeors of water content 10^-4 kg/m^3 as a function of frequency Geer et al. (2021, GMD, submitted) 17 EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS
Wavelength Cloud and 100 mm 10 mm 1 mm precipitation Slab cloud at 283K optical properties: above a 280K the microwave surface (solid) spectrum Slab cloud at 283K above a 100K surface (dashed) Geer et al. (2021, GMD, submitted) EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 18
All-sky assimilation © ECMWF May 6, 2021
• Clear-sky assimilation: Clear-sky or – Remove any cloud-contaminated observations all-sky? – Do not model the effect of cloud on brightness temperatures – Traditionally used for temperature sounding channels (e.g. AMSU-A channel 5) – Extract small signals of temperature forecast errors (order 0.1K) that would be swamped by errors from displaced clouds and precipitation (10-100K) • All-sky assimilation – Model the effect of cloud and precipitation on the observations – Assimilate all data, whether clear, cloudy or precipitating – Initially developed for water-vapour sounding and imaging channels – Use the tracing mechanism of 4D-Var to infer the dynamical state from errors in the location/intensity of water vapour, cloud and precipitation • Forthcoming ECMWF cycle 47r3: AMSU-A will be assimilated in all-sky conditions – Now all but a handful of microwave sensors are assimilated in all-sky conditions – Broadly, clear-sky approach is now outdated at microwave frequencies EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 20
Impact of satellite radiances by sensor (100% = all obs) After moving AMSU-A to all-sky (future 47r3 config.) 21
Time of Start End All-sky Assimilation window observation (08Z) ‘tracer’ effect Single 190 MSLP and GHz snow observation column assimilation test case (FG) Snow column increment Snow reduction at observation time generated by reduction in strength of low pressure area 1000km away, 11h earlier MSLP increment EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 22
All-sky microwave imagers: a unique contribution from precipitation- affected observations Microwave imagers give their largest forecast impact from a small fraction of precipitating scenes. AMSR2, July-August 2016 EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 23
GMI contributes to improved forecasts of hurricane / cyclone Leslie (2018) GMI observations at 24 GHz 06 UTC, 12-Oct-2018 Forecast from 11th TB [K] October, 12 UTC is incorrect - Leslie’s “+” are position is around particularly 1000km out beneficial observations according to Background FSOI Forecast from 12th October, 00 UTC is simulated correct – Leslie hits from model Portugal overnight on 13-14th October Data assimilation shifts the cyclone • Long-lived Atlantic hurricane Leslie transitioned to an extratropical storm and brought strong winds, heavy centre 60 km rain and flooding to Portugal, Spain and France. Analysis towards Portugal • Forecasting the landfall in Portugal remained challenging 60 hours ahead but the 48 hour forecast began to (almost invisible capture the true evolution of the storm. at this scale), and • On this occasion, GMI was in the right place to give the biggest satellite contribution to the improved forecast triggers more (drifting buoys contributed significantly more, but were the only observation giving more impact than GMI). rapid and more • Forecast Sensitivity to Observation Impact (FSOI) shows GMI on average contributes around 1.3% to the NE movement of 24h forecast quality at ECMWF. Around 28 satellite radiance sensors are assimilated overall, along with the cyclone many other types of observation, all of which contribute at one time or another. • GMI all-sky radiances are assimilated over ocean from channels 19v, 19h, 24v, 37v, 89v, 183±3 and 183±7 EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 24
Geer (2021, submitted to AMTD): Parameter estimation for 6 macro- and microphysical variables Physical characteristics of frozen hydrometeors inferred using parameter estimation Costfunction at original config. at best config. EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 25
2. Active 1. Microwave radar . 6. Earth 5. Solar T-sounding low- high-freq. ud ecip dar 3. Sub-mm 4. Infrared o r Li radiation Cl P budget Cloud Cloud ice Cloud fraction, particles, water cloud top lighter height, multiple precip. Ice particle layers size, shape, orientation Vertically Larger Vertically resolved frozen resolved particles, particle Cloud size and particle Size of water Cloud shape cloud particles size and liquid number water 9. Lightning Melting particles Cloud Sub-FOV Rain, including cover heterogeneity particle size & structure 7. Rain gauge 8. Ground radar
Questions? EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS 27
You can also read
