Detection and attribution of changes in arctic ecosystems

Detection and attribution of changes in arctic ecosystems and atmospheric CO 2 P. K. Patra, H. Kobayashi and T. Saeki Department of Environmental Geochemical Cycle Research Acknowledgements to Graven et al. , Science, 2013 At A-CG 36 : Science in the Arctic Region, Jp. GU Pacifico Yokohama, 30 April 2014

Talk outline • Introduction • Detection of atmospheric CO 2 change from longterm measurements • Changes in vegetation phenology by remote sensing • Implications for carbon uptake by inverse modeling of atmospheric CO 2

Example of decomposing CO 2 timeseries at Barrow (71 o. N), Alaska Due to fossil fuel CO 2 trend (ppm) Climate, Fires, Volcano Terrestrial biosphere CO 2 growth rate (ppm) CO 2 seasonal cycle (ppm) CO 2 concentration (ppm) Measurement Fortran Program by Nakazawa et al. , 1997

Attribution of changes Atmospheric CO 2 = fossil fuel emission + terrestrial biosphere + oceanic exchange + atmospheric transport Strong trends Strong seasonality Weaker trends Most studies looked in to a particular aspect, e. g. , Graven et al. Piao et al. : atmospheric CO 2 : terrestrial ecosystem model Role of terrestrial biosphere, fossil fuel emissions, oceanic exchange and atmospheric transport remained relatively unexplored

Partitioning of global CO 2 sources and sinks CO 2 Fluxes (Pg. C yr-1) 2011 2002 1750 - 2011 9. 5± 0. 5 8. 3± 0. 4 365± 30 0. 9± 0. 5 1. 0± 0. 5 180± 80 3. 8± 0. 9 2. 6± 0. 8 150± 90 (Residual) 4. 1± 0. 1 4. 3± 0. 1 240± 10 (Observations) (Observations 2. 7± 0. 5 2. 5± 0. 5 155± 30 (5 models) Le Quere et al. 2013, ESSD; updated from www. whrc. org

Contribution of 3 fluxes on CO 2 seasonal cycle amplitude

Large increase (up to 60%) in land carbon exchange (between IGY/1960 and IPY/2010) Lack of representation in CMIP 5/IPCC-AR 5 models (independent evaluation critical) Graven, Keeling, Piper, Patra et al. , Science, 2013 7

Attribution of changes (results from Hideki Kobayashi) CHANGES IN BOREAL ECOSYSTEM

CO 2 seasonal cycle: definitions Peak to peak amplitude Terrestrial source 0 Terrestrial sink Zerocrossing (Onset) Zerocrossing (Offset)

Vegetation phenology estimation Satellite data: SPOT VEGETATION S 10 from 1999 to 2012, @1/112 degree NDVI >0. 3 threshold for onset/offset (NDVI 0. 3) NDVI 50% of max-min NDVI threshold for onset/offset (NDVI 50%) NDWI minimum for onset (NDWI) NDVI threshold methods (White et al. , 1997, 2009) Onset Offset NDWI minimum method (Delbart et al. . , 2005) Onset

CO 2 zero crossing vs. vegetation onset/offset at Barrow - Barrow CO 2 vs Satellite onset/offset estimates in {68 -72 o. N and 140 -180 o. W} Day anomalies show earlier trend both in zero-crossing and spring onset Autumn offset delayed a little in the satellite estimates, but almost constant Spring onset anomaly Autumn offset anomaly

Spring onset anomaly Autumn offset anomaly

CO 2 seasonal amplitude and vegetation growing season length (GSL) at Barrow/Alaska - Barrow CO 2 vs Satellite growing season length in {68 -72 o. N and 140 -180 o. W} Increase in the CO 2 amplitude coincides with increase in the GSL

Attribution of changes (results from Tazu Saeki) EFFECT ON NET CO 2 UPTAKE

Inverse modeling of surface fluxes • Bayesian synthesis inversion CO 2 & Transport model CO 2 fluxes Input : • Transport model simulations at monthly-mean timestep • Monthly-mean atmospheric CO 2 data Output: monthly CO 2 fluxes and uncertainties for 84 partitions of the globe for 1990 -2011. Rayner et al. , 1999; Baker et al. , 2006, Patra et al. , 2005

CO 2 flux [Pg. C yr-1] Global totals and semi-hemispheric CO 2 fluxes Fluxes from 84 partitions of globe using CO 2 obs. at ~100 sites and ACTM transport are estimated

CO 2 flux [Pg. C yr-1] Recent changes in annual mean CO 2 fluxes Boreal Eurasia Boreal North America Northern Ocean Europe Northern Ocean Bor. N. Am. Europe Boreal Eurasia

Summary and outlook • The CO 2 seasonal cycle amplitude has increased since the direct measurements started in 1950 s • Bottom-up and top-down flux estimations suggest more exchange between carbon pools in the last century • However the sub-hemispheric or regional carbon fluxes and their exchange mechanism remain largely unknown • Further improvements in monitoring and modeling of the ecosystems are required for attribution of CO 2 variations

End Q (chair - Inoue? ): What is the role of Arctic sea-ice change? A: Here our inverse model suggest the Northern Ocean sink is reducing with time. That means sea-ice reduction is not helping us to sequester more carbon from the atmosphere. (Later I thought more about it! The sea-ice melt should increase stability of surface water and thus slower vertical mixing – so a reduction in CO 2 uptake even though the surface water is undersaturated for CO 2. Although some nutrients supply would also increase, but that may not be playing significant role in CO 2 flux change) Q (Tohjima): why are the changes in Arctic vegetation? A: Mainly the increase in temperature. There are two main changes in the divers, the surface temperature and CO 2 fertilization. But the CO 2 fertilization should be rather uniform globally. So I feel the temperature change is the main cause of arctic greening.

Notes (30 Apr 2014; Arctic Environment): 3 rd talk (by Tanaka): The great cyclone of 2012 in Arctic Ocean: 2012 08 03 For frequency and statistics of arctic cyclones : Simmonds and Rudeva (2012). One may check N 2 O STE using MLS observations and ACTM simulations 4 th talk: Inuoe et al. (2012) – warm arctic and cold siberia (wacs) 6 th talk (Tanaka) : AOI. Interesting analysis of AO eigenmodes (linear+nonlinear+forcing). While the linear term is in phase with the AOI, the forcing term leads the AOI (not sure I understand fully) 7 th talk (enomoto) : Arctic data archive system – ADS Q (patra) : great collection of data, but how to use and think together? Need a workshop? A. Yes, let’s talk and discuss with Ibuki-san and others 8 th talk (Aoki): greenland snow change – albedo (bc & dust) etc. 9 th talk : GRACE greenland ice sheet (Sasgen et al. , EPSL, 2012) and Ice. SAT xx talk (Kuchiki? ) : albedo decrease with snow grain size, ~1 at 1 um to <0. 5 at 1000 um. Albedo also decrease as the impurity (bc &dust) increase – more sensitive at coarser grain size. Measured EC, OC, BC and dust during 2012 -2013 at Ny Alesund. Without BC in Savalband, the Albedo should have been -0. 02 to -0. 03 (or higher)

Notes (01 May 2014; Future Earth, Atmos. Chem): Future Earth (9 am): Keywords : transformation (to all meat eating? ), sustainability (your industries? ), co-design/co-production (impossible, from what I have seen – but somehow helps some people to become rich!), planetary boundary (account for the past? ) 2 nd talk (emori) : talked about next 100 years, but model do not explain past 10 years!) 3 rd talk (by ozaki): peat fire and carbon management in Indonesia. 11 am (Taniguchi) : GEC projects – FE core projects (27? ), Global hub: Boulder, Montreal, Paris, Stockholm, Tokyo(? ? ), Regional hub: Asia-Pacific, …. Three big efforts: Dynamic planet – 1. Observing, Monitoring, Modeling; 2. Global development – footprint (Taylor et al. , 2012), Developing Asia in trouble? ? , 3. Transformation towards sustainability (Kumamoto city example) – good practice Atmos. Chem (11: 30 am): Umezawa : CH 3 CL emission: tropical plants – 50%, bio burn – 25%, (Xiao et al. , ACP, 2010). CARIBIC conc over FRT higher than MHD, and seasonal minimum in September. CH 3 Cl has seasonal peak in Aug over India. Ishidoya : Exchange ratio of CO 2: O 2 are different for GPP (tree bag) and RH (chamber) = 1. 1 & 1. 01 respectively Tohjima : El Nino (-ve CO 2, +APO, -NPP), La Nina (+ve CO 2, -APO, +NPP) Sugawara : Cl in stratosphere (Mc. Carthy et al. , 2003) – CSIRO data. Land processes (2: 15 pm) Yoshimura: heavy isotopes from ET, lighter from Evap. (Henderson & Sellers). Domination of land ET (Jasechko et al, Nature, 2013). d 18 O reducing from ~1910 to upto ~1980, and recently has started to increase (in prep. ). En. KF-OSSI done on SCIA, …, TES, IASI data of del. D(? ), with max impact from IASI having daily snap shots. PKP not asked: Can you estimate ET or E using d 18 O as a constraints?

Seasonal amplitudes at Northern Hemisphere flask sites have grown by 20 -40% Point Barrow Mauna Loa Keeling et al. 1996; Cleveland et al. 1983; Randerson et al. 1997, 1999

Comparison of seasonal cycles between the IGY and HIPPO eras at flask sites Mauna Loa 1961 -64 and 2009 -11 means 18 ± 7 % increase in amplitude Point Barrow 1961 -64 and 2009 -11 means 41 ± 8 % increase in amplitude Uncertainties based on biweekly standard error in 3 -harmonic fits

Atmospheric-CO 2 variability and its controls Patra et al. , Tellus, 2005 (updated)

Why arctic? (right) Trends in earth’s surface temperature (source: NASA/GISS) increase between 1951 -1980 and 2004 -2013 (down) CO 2 seasonal cycle amplitude and amplitude change between 1960 and 2000 Revise this figure for CASA+OCN+FOS and trends: ~/trans_uair/analysis_igy 50/plot 2 d/scamp_fdif. gs

Example of vegetation onset and offset Decadal mean onset and offset (1999 to 2012) Spring onset (NDVI 50 %) Autumn offset (NDVI 50 %)