Influence of IASI Data on the forecast of ET events. Doris Anwender, Nadia Fourrie, Florence Rabier, Philippe Arbogast

Transcription

1 Influence of IASI Data on the forecast of ET events Doris Anwender, Nadia Fourrie, Florence Rabier, Philippe Arbogast

2 Overview Arpege The experiments Nord2 Global results Results for ET Error reduction Case study: Hanna Ike Impact during ET Hanna 14 Gustav 7

3 Motivation Heterogeneous in-situ data coverage in ET events High density observation network benefitial during ET to describe transition from small to large scales? Improvement of moisture processes in the model by additional water vapor observations during ET? => downstream development (flooding)

4 Experiments Measurement of radiances by satellites Monochromatic radiance, measured at the satellite based measurement system, is connected to variables like temperature and moisture by radiation transfer equation Rν = I ν τ ν z B ν { T z } z dτ ν z dz dz weighting function I ν τν z B ν {T z } T z Irradiance from the surface at height z Vertical transmission from height z into space Corresponding profile of the Planck function Vertical temperature profil

5 Experiments IASI (Infrared Atmospheric Sounding Interferometer) Swath Developed by the french space agency CNES swath: 2 km, 3 scan positions, global coverage within 12 hours height: 837 km 1 orbit: 99 min ~ 14 orbits / day

6 Experiments IASI (Infrared Atmospheric Sounding Interferometer) IASI is a hyperspectral measurement instrument infrared: /cm with a spectral sampling of,25 / cm; IASI provides information on temperature and humidity, horizontal resolution 12 km at nadir No cloud affected channels Swath Developed by the french space agency CNES swath: 2 km, 3 scan positions, global coverage within 12 hours height: 837 km 1 orbit: 99 min ~ 14 orbits / day

7 Experiments IASI weighting functions IASI operationel: 68 channels for temperature Channels: Measurement frequencies Weighting function: Determines from which layers the radiation is emitted to space

8 Experiments Global model ARPEGE (Action de Recherche Petite Echelle Grande Echelle) Resolution: T538, stretching factor: 2.4, layers in hybrid coordinats (surface up to.1 hpa) Data assimilation: 4D-Var with two non-stretched minimisation loops Close cooperation with ECMWF

9 Experiments Schematic of 4D-Var data assimilation In Meteo France ARPEGE (similar to ECMWF IFS) c ba k analysis nd u gro observation truth h 6h 6h assimilation window time 12 6h

10 Experiments Overview over the experiments Time period: 1 2 September 28; contains 3 ET cases with probable impact on Europe / Mediterranean: Gustav, Hanna, Ike Experiments with the data assimilation system x x x x x x Control experiment DENS: operational data density x 4 NOIASI: denial of all IASI data IASIWV: 9 additional water vapor channels for IASI

11 Global results Global impact: High density 5 hpa relative humidity, 24 h forecast dens North2 oper Tropics Error South Verification time Positive global impact of high density significant for ~ 48 h; in tropics ~ 84 h 7 14

12 Global results Global impact: Denial of all IASI data 5 hpa temperature, 24 h forecast noiasi dens North2 South2 Error Verification time Impact of denial negative on both extratropic hemispheres positiv in tropics (already corrected, not shown) significant for South: ~ h, North: ~ 24 h, Tropics: ~ 48 h Positive impact of IASI on northern and southern extratropics and weak signal in tropics also found by Guidard et al. (21)

13 Experiments IASI temperature and water vapor channels assimilated in the experiments 64 channels for temperature 9 channels for humidity

14 Global results Global impact: IASI water vapor channels 85 hpa relative humidity, 24 h forecast iasiwv North2 dens Tropics South Error Verification time Significant positive impact on Southern hemisphere for ~ 24 h Neutral from 24 h on in Northern hemisphere and Tropics Significant positive impact for North and Tropics from 24 h (not shown)

15 Results for ET Impact of additional water vapor channels over the Atlantic and Europe Change of the absolute error Variable: total energy E x = fcst exp an EC fcst ctrl an EC i Atlantic Europe i i i

16 Results for ET ET Ike ET Hanna ET Gustav Influence of additional water vapor data: Atlantic

17 Results for ET Influence of additional water vapor data: Atlantic Medium range: h Short range: 3 h 3e E(x) -3e+5 1e E(x) -1e+6 Initialization date 3e+6 Initialization date Longer range: h E(x) -3e+6 Initialization date

18 Results for ET Influence of additional water vapor data: Atlantic Medium range: h Short range: 3 h 3e E(x) -3e+5 1e E(x) -1e+6 Initialization date 3e+6 Initialization date Longer range: h E(x) -3e+6 Initialization date

19 Results for ET Influence of additional water vapor data: Atlantic Medium range: h Short range: 3 h 3e E(x) -3e+5 1e E(x) Slightly improved forecasts -1e+6 when water vapor channels are added Initialization date Initialization date around ET times 3e+6 Longer range: h E(x) -3e+6 Initialization date

20 Results for ET ET Ike ET Hanna ET Gustav Influence of additional water vapor data Europe

21 Results for ET Influence of additional water vapor data: Europe Medium range: h Short range: 3 h 3e E(x) -3e+5 1e E(x) -1e+6 Initialization date 3e+6 Initialization date Longer range: h E(x) -3e+6 Initialization date

22 Results for ET Influence of additional water vapor data: Europe Medium range: h Short range: 3 h 3e E(x) -3e+5 1e E(x) -1e+6 Initialization date 3e+6 Initialization date Longer range: h E(x) -3e+6 Initialization date

23 2 hpa horizontal wind >= 3 m/s (shaded) surface pressure (blue contours), 2 hpa geopotential (black contours) 3 Sep Z Gustav 6 Sep Z Hanna 4 Sep Z 7 Sep Z 5 Sep Z 8 Sep Z

24 Case study Hanna 4 September 28 UTC background error correction analysis error correction 8 2 dens iasiwv 2

25 Case study Hanna 4 September 28 UTC background error correction analysis error correction dens 2 Slightly improved correction of background and analysis error iasiwv 2

26 Case study Hanna 4 September 28 UTC background error correction analysis error correction ctrl 2 Smaller analysis errors in Hanna's outflow 2 hpa horiz. wind >= 3 m/s (shaded) sfc press (blue), 2 hpa geopotential (black) iasiwv 2

27 Case study Hanna Difference propagates into the midlatitudes Diff: 2 hpa horizontal wind iasiwv - dens + h + 12 h dens + 24 h Total precipitation + 36 h iasiwv

28 Case study Hanna Smaller errors in the analysis do not yield smaller forecast errors Total energy RMSE difference + h Fcst initialized on 4 Sep 28 UTC + 12 h + 24 h Smaller fcst errors in midlatitude trough from 24 h tropical cyclones Hanna and Ike

29 Case study Hanna Influence on midlatitudes Smaller forecast errors during ET of Hanna expand in scale Total energy RMSE difference Fcst initialized on 4 Sep 28 UTC + h + 72 h + 84 h + 12 h

30 Case study Hanna Improved representation of jet does not yield strong improvement in surface pressure forecast 12 h forecast: 2 hpa horizontal wind (shaded), mslp (blue contours), 2 hpa geopotential (black contours) Analysis Forecast dens + 12 h Forecast iasiwv Diff: 2 hpa horizontal wind iasiwv dens (shaded) + 12 h + 12 h

31 Summary Global impact: High density: Positiv over the whole globe, strongest impact on southern extratropics Denial of all IASI data: Negativ on both extratropic hemispheres IASI water vapor channels: Positiv on southern extratropics; weakly positive on northern extratropics and tropics for a very short time Impact of water vapor channels during ET: Notable influence (mainly positive) for initialization around ET times Better correction of background and analysis error Improvement of forecasts can be partly assigned to improved representation of ET events, however improvements are small

32 Summary Global impact: High density: Positiv over the whole globe, strongest impact on southern extratropics Denial of all IASI data: Negativ on both extratropic hemispheres IASI water vapor channels: Positiv on southern extratropics; weakly positive on northern extratropics and tropics for a very short time Impact of water vapor channels during ET: Notable influence (mainly positive) for initialization around ET times Better correction of background and analysis error Improvement of forecasts can be partly assigned to improved representation of ET events, however improvements are small

33 Merci pour votre attention!

34 observations dens - ctrl Absolute difference Absolute difference observations iasiwv - dens Dates September 28 Dates September 28

35 Results for ET Influence of additional water vapor data Atlantic

36 Influence of additional water vapor data: Atlantic Short range: 3 h Medium range: h Longer range: h

37 Results for ET Influence of additional water vapor data Europe

38 Influence of additional water vapor data: Europe Short range: 3 h Medium range: h Longer range: h