Analysis and hindcast experiments of the 2009 sudden stratosphere warming in WACCMX+DART

Transcription

1 Analysis and hindcast experiments of the 2009 sudden stratosphere warming in WACCMX+DART Nick Pedatella 1,2, Hanli Liu 1, Daniel Marsh 1,3, Jeffrey Anderson 4, and Kevin Raeder 4 1 High Altitude Observatory, NCAR 2 COSMIC Program Office, UCAR 3 Atmosphere Chemistry Observations and Modeling, NCAR 4 Computational Information Systems Laboratory, NCAR

2 Motivation: Sudden stratosphere warming events drive large uch variability is disturbances in the middle and upper atmosphere s observed over nalyzed the fof2 ined with the iodoing a similar observed similar time values after before the SSW entage wise) are ferences indicate ges, the shape of he ISR electron then we proceed The he observaiking differences r the winter and Change in F-region Vertical Plasma Drift Velocity Jicamarca, Peru (75W, 12S) Temp. (90 N, 10hPa) SSW 5 ΔTEC Figure 7. Similar to Figure but for Note that the JRO drift enhancement is delayed 4 5 days with respect to the SSW peak temperature. 6 of 8 To study the middle-upper atmosphere variability during specific events it is necessary to constrain the model meteorology the F regio average TE daytime 38% and (8 8 semidiurna [15] The tude (top p suggests th E region d suggests in major com the increas variation, t related prim density afte the F regio daytime (8 semidiurna tude (top p

3 Motivation: Nudging towards reanalysis can lead to large differences at higher altitudes to the sensitivity of the mesosphere GW drag Journaldue of Geophysical Research: Space Physics to /2013JA Zonal Mean Temperature N Constrained Domain (Pedatella et al., 2014) Figure 1. Zonal mean temperature averaged between 70 and 80 N for (a) GAIA, (b) HAMMONIA, (c) WAM, (d) WACCM-X, and (e)

4 Geophysical Research Letters /2015GL Differences in modeled MLT dynamics influence nitric oxide descent NO at 80 N, SD-WACCM Nudged Region NO at 80 N, NOGAPS-WACCM Nudged Region (Funke et al., 2017) Figure 2. Observed and modelled NOx 2008 March Figure 1. (a and c) Calculated nitric oxide given in log mixing ratio units during (parts per November volume; the 30 ppbv contour referred to M Direct assimilation of lower, middle, upper atmosphere observations in the text!7.5 the figure) at 80 N two versions ofand the SD-WACCM model, nudged by ppbv MERRA (Figure 1a) andto Figure 1. (aisand c) in Calculated nitric oxidefrom given in log mixing ratio units locations (parts per volume; the 30 referred and times ofcontour the observations is one approach to improving simulations of MLT dynamics nudged byisnogaps-alpha (Figure 1c). from (b andtwo d) Corresponding calculated temperatures. in the text!7.5 in the figure) at 80 N versions of the SD-WACCM model, nudged by MERRA (Figure 1a) and 0.1, 1, and 10 ppmv. White regions refl (Siskind et al., 2015) nudged by NOGAPS-ALPHA (Figure 1c). (b and d) Corresponding calculated temperatures.

5 WACCMX+DART Framework for WACCMX+DART is identical to WACCM+DART (next slide) Same observations are assimilated in the troposphere, stratosphere, and mesosphere. Main change between WACCMX+DART and WACCM+DART is increased damping in WACCMX. This was necessary for model stability, and to ensure that mixing from small scale waves introduced by the data assimilation do not excessively reduce thermosphere O/N 2 and electron density. Changes made for model stability tend to damp tidal amplitudes, and have a slight negative impact on performance of the data assimilation in the troposphere-stratosphere. Troposphere humidity is biased by ~20-30% due to model physics issue when using a 5 min time-step. We have performed initial WACCMX+DART analysis and hindcast simulations for the 2009 SSW time period.

6 WACCM+DART WACCM+DART provides an atmospheric reanalysis from the surface to the lower thermosphere (~145 km). Conventional Lower Atmosphere Observations: Aircraft temperature and wind Radiosonde temperature and wind Satellite drift winds COSMIC GPS refractivity Sparse Middle/Upper Atmosphere Observations: TIMED/SABER Temperature ( hpa) Aura MLS Temperature ( hpa) 500 hpa Geopotential Height 0000 UT 15 Nov., 2008 NCEP Reanalysis WACCM+DART Typically use a 40-member ensemble, which is a tradeoff between computational expense and having a sufficiently large ensemble to capture a variety of atmospheric states. WACCM+DART is useful for correcting model biases, studying dynamical variability due to sudden stratosphere warmings, and short-term tidal variability Pedatella, N. M., K. Raeder, J. L. Anderson, and H.-L. Liu (2014), Ensemble data assimilation in the Whole Atmosphere Community Climate Model, J. Geophys. Res., 119, doi: /2014JD

7 Middle Atmosphere Variability in WACCMX+DART and SD-WACCMX SD-WACCMX: Specified Dynamics WACCMX constrained to MERRA meteorology up to 50km

8 NO descent following the SSW (Funke et al., 2017) Figure 2. Observed and modelled NOx during November 2008 March M locations and times of the observations 0.1, 1, and 10 ppmv. White regions refl

9 Semidiurnal Migrating Tide (10-4 hpa) (Pedatella et al., 2014)

10 GPS TEC Variability in WACCMX+DART and SD-WACCMX

11 2009 SSW Hindcast Experiments - Initialize 40-member ensemble forecasts (hindcasts) of the 2009 SSW on January 5, 10, 15, 20, and Ocean SSTs are specified as the true values (i.e., not forecasted) - Solar activity is specified by using 27-days prior solar activity

12 2009 SSW Hindcasts: 70-90º N Temperature

13 2009 SSW Forecasts: TEC at 75W

14 Summary Recent developments in WACCMX support whole atmosphere data assimilation, providing a global view of the troposphere, stratosphere, mesosphere, thermosphere, and ionosphere state Middle atmosphere chemical and dynamical variability are generally well reproduced in WACCMX+DART. Tidal amplitudes are generally too weak in WACCMX+DART, indicating the need to determine a better method for filtering small-scale waves introduced by DA. Ionosphere variability during the 2009 SSW is reproduced in WACCMX+DART. Forecast experiments for 2009 SSW show that middle-upper atmosphere variability can be qualitatively predicted ~5-10 days in advance of the SSW. Work is ongoing to improve ionosphere-thermosphere analysis fields through assimilation of ionospheric observations

15 TEC hindcasts with fixed solar forcing

16 Diurnal Migrating Tide (Pedatella et al., 2014) Figure 9. DW1 amplitude of temperature at 0.01 hpa ( 80 km) for (a) GAIA, (b) HAMMONIA, (c) WAM, and (d) WACCM-X Figures 9a 9d except for the DW1 phase.

17 2009 SSW Forecasts: Solar+Lunar Semidiurnal Migrating Tide

18 Impact of DA noise on the ionosphere electron density