Inez Fung University of California, Berkeley. Center for Multiscale Modeling of Atmospheric Processes UC Berkeley, January

Size: px

Start display at page:

Download "Inez Fung University of California, Berkeley. Center for Multiscale Modeling of Atmospheric Processes UC Berkeley, January"

Elizabeth Abigayle McCarthy
7 years ago
Views:

1 Inez Fung University of California, Berkeley Center for Multiscale Modeling of Atmospheric Processes UC Berkeley, January Chabot Space and Science Center October

2 Only half of the CO 2 produced by human activities is remaining in the atmosphere Where are the sinks that are absorbing over 40% of the CO 2 that we emit? Land or ocean? Eurasia/North America? Why does CO 2 buildup vary dramatically with nearly uniform emissions? How will CO 2 sinks respond to climate change?

3 1.1±0.7 PgC y ±0.1 PgC y -1 47% Radia6vely ac6ve 7.7±0.5 PgC y PgC y -1 27% Calculated as the residual of all other flux components 26% 2.3±0.4 PgC y -1 Average of 5 models Global Carbon Project 2010; Updated from Le Quéré et al. 2009, Nature Geoscience; Canadell et al. 2007, PNAS

4 NOAA- ESRL ~ 100 sites at remote marine loca6ons, bi- weekly flasks, 2m Long- term increase 2-3ppm/yr Seasonal cycle ~O(10 ppm) N- S gradient ~ 1% of global mean

5 5

6 Height SF6 (ppt) Latitude Denning et al. Tellus 1999

7 Height surface Model Sources only obs Infer Carbon Flux Aircraft obs ppm obs Tropical land NH land Aircraft obs Carbon Sink inferred as that required to match the observa6ons; Underestimate Overestimate flux Stephens et al., 2007, (Science) 7

8 Assimilate meteorological observations together with satellite and surface CO2 observations into a carbon-climate model Goals: improve representation of carbon processes in carbon-climate model Improve estimation of carbon sources/sinks Articulate uncertainties in carbon sources/ sinks due to uncertainties in transport Junjie Liu, Inez Fung (University of California, Berkeley) Eugenia Kalnay, Ji-Sun Kang (University of Maryland) Supported by DOE All the computa9ons were carried out in DOE NERSC

9 Ensemble analyses (ini6al states) Ensemble forecasts Observa6ons t=0hr t=06hr t=12hr Background error changes with 6me; Obtain ensemble analyses. 9

10 u' v' T' q' Ps' b x : ensemble mean. K 1 Pb = (x bi x b )(x bi x b )T K 1 i =1 u ʹ u ʹ u ʹ vʹ u ʹ T ʹ u ʹ qʹ u ʹ Psʹ vʹ u ʹ vʹ vʹ vʹ T ʹ vʹ qʹ vʹ Psʹ = T ʹ u ʹ T ʹ vʹ T ʹ T ʹ T ʹ qʹ T ʹ Psʹ qʹ u ʹ qʹ vʹ qʹ T ʹ qʹ qʹ qʹ Psʹ Psʹ u ʹ Psʹ v' Psʹ T ʹ Psʹ qʹ Psʹ Psʹ u' v' T' q' Ps' Run K ensemble members- > xib; i=1..k std dev in u etc error of the day Large std dev atm is dynamically unstable 10

11 Run K ensemble members- > x ib ; i=1..k x b : ensemble mean. = P b = uʹ uʹ vʹ uʹ T ʹ u ʹ qʹ uʹ Psʹ uʹ 1 K 1 uʹ vʹ vʹ vʹ T ʹ v ʹ qʹ vʹ K i=1 Psʹ v' (x i b x b )(x i b x b ) T uʹ T ʹ vʹ T ʹ T ʹ T ʹ qʹ T ʹ Psʹ T ʹ uʹ qʹ vʹ qʹ T ʹ q ʹ qʹ qʹ Psʹ qʹ uʹ Psʹ vʹ Psʹ T ʹ P sʹ qʹ Psʹ Psʹ Psʹ u' v' T' q' Ps' u' v' T' q' Ps' Propagate info from the dynamical variables with observa6ons to the dynamical variables with no observa6on. From loca6on with observa6on to loca6ons with no obs. 11

12 1) Global 6 hr ensemble forecast star?ng from the analysis ensemble 2) Choose the observa?ons used for each grid point 3) Compute the matrices of forecast perturba?ons in ensemble space X b 4) Compute the matrices of forecast perturba?ons in observa?on space Y b 5) Compute P b in ensemble space space and its symmetric square root 6) Compute w a, the k vector of perturba?on weights 7) Compute the local grid point analysis and analysis perturba?ons. 8) Gather the new global analysis ensemble. Ensemble forecast Obs. operator Ensemble forecast at obs. locations LETKF Ensemble analysis GCM Observations 12

13 Community Atmospheric Model (fvcam 3.5) (2.5x1.9x26) Ocean carbon flux (Takahashi et al. 2002) CO2, winds, q, T, Ps Fossil fuel emission Photosynthesis Respira6on Ocean Land CO2 is transported as a tracer in CAM 3.5. Land carbon flux: 6- hourly flux from biogeochemical model. Model produces CO2 distribu6on that matches major features in surface CO2 obs Time period: the first half year of

14 Raw Meteorology every 6 hours: u, v, T, q, Ps In-situ CO2: continuous nearsurface CO2 at Pt Barrow, AK; Mauna Loa; American Samoa; and South Pole Satellite CO2: AIRS, ~100km, 2/day 14

15 Zonal wind within 500hPa and 600hPa Surface pressure m/s hpa Assimilate all the quality- controlled meteorological observa6ons assimilated in the DOE/NCEP Reanalysis 2 products (Kanamitsu et al. (2002) observa6ons within 6- hour. 15

16 Overpass the same spot twice per day; Footprint is about 100km. 16

17 AIRS averaging kernel AIRS CO2 at 18Z01May2003 (+/- 3hour) o: polar region; +: mid- la6tude; closed circles: the tropics. ppm Averaging kernel: the sensi6vity of AIRS CO2 to the CO2 at each ver6cal level. 17

18 Free run. No assim. CO2 flux prior; CO2 as inert tracer Met run: assim meteorology only. CO2 flux prior; CO2 is inert tracer Compare with free run Compare bias in CO2 if use mean of meteorology Assimilate Met+CO2; CO2 flux prior (not updated) Assim Met+CO2; update CO2 flux 18

19 CAM3.5 6 hour ensemble forecasts (u, v, T, q, Ps) Mean meteorology CAM3.5 LETKF Analysis 64- member ensemble meteorology states CAM3.5 Case 1 Single forecast Ensemble forecasts Case 2 (u, v, T, q, Ps) CO2(met) CO2((met) i, i = 1, K) 19

20 [CO2(met) 1 K January K i=1 CO2((met) i )] at surface June Non-uniform in both space and time, with the difference as large as ± 1.5ppm. The difference is due to the nonlinear interaction in the model alone. 20

21 Free run. No assim. CO2 flux prior; CO2 as inert tracer Met run: assim meteorology only. CO2 flux prior; CO2 is inert tracer Compare with free run Compare bias in CO2 if use mean of meteorology Assimilate Met+CO2; CO2 flux prior (not updated) Assim Met+CO2; update CO2 flux 21

22 Met- run AIRS- run analysis (u, v, T,q, Ps) CAM3.5 6 hour forecast (u, v, T, q, Ps) LETKF CO2 analysis (u, v, T, q, Ps) LETKF CAM3.5 6 hour forecast (u, v, T, q, Ps) CO2 LETKF analysis (CO2) Observa6ons (u,v,t,q,ps) Observa6ons (u,v,t,q,ps) Observa6ons AIRS CO2 AIRS- run: AIRS CO2+met obs; Met- run: only met obs. The first half year of Prescribed surface carbon flux forcing. 22

23 90N CO 2 analysis ensemble spread at observa6on space Average number of AIRS CO 2 observa6ons within 6- hour 55S Analysis ensemble spread is an6- correlated with the the CO 2 observa6on coverage 23

24 AIRS obs Column CO2 Assim Meteorology +AIRS Assim Meteorology only Column CO2 minus AIRS

25 Molokai Island, Hawaii. May 11, 2003 Estevan point, Bri6sh Columbia, Feb 27, 2003 obs Meteor- run: CO2 tracer transported by 64- member ensemble meteorological analyses generated every 6hr - - > precision of CO2 forecast by model AIRS- run: CO2 assimilated along with meteorological obs. 25

26 May 2003: CO2(925hPa)- CO2(500hPa) Met- run (AIRS- run)- (Met- run) ppm In the NH, CO2(925hPa)>CO2(500hPa): fossil fuel+ land carbon source; In the SH, CO2(925hPa)<CO2(500hPa): transported from the NH. Assimilating AIRS CO2 decreases vertical gradient 26

27 momentum energy water vapor CO2 u t + u u + 2Ω u = 1 ρ p + g k ˆ + F + I( u ) T t + u T = SW +LW +SH + LH + I(T) q t + u q = Evap Condensation + I(q) C t + u C = SfcFlux + I(C) I convec6ve mixing

28 T(z), q(z) from meteorologoical assimilation ~= best approx to obs Convective mixing working terms on RHS of equations (in-situ source/sink terms) to produce best T(z), q(z) CO2 has no in-situ source/sink terms CO2(z) is a cleaner diagnostic of sub-grid scale vertical mixing CO2 is the only inert (in atm) tracer with repeating obs in the vertical

29 High resolu6on spectra of reflected sunlight in near IR CO 2 and O 2 bands 3 km 2 footprint at nadir 3 Hz Sun- Synchronous Orbit (7km/s): 16- day repeat Nadir Glint Glint Spot O 2 A- band CO µm Local Nadir CO µm Clouds/Aerosols, Surface Pressure Column CO 2 Clouds/Aerosols, H 2 O, Temperature

30 Ji- Sun Kang, PhD thesis 2009

Integrated Global Carbon Observations. Beverly Law Prof. Global Change Forest Science Science Chair, AmeriFlux Network Oregon State University

Integrated Global Carbon Observations Beverly Law Prof. Global Change Forest Science Science Chair, AmeriFlux Network Oregon State University Total Anthropogenic Emissions 2008 Total Anthropogenic CO 2