Changes in the seasonal activity of temperate and boreal vegetation The critical role of Autumn temperatures. Shilong Piao, Philippe Ciais, Pierre Friedlingstein, Philippe Peylin, Nicolas Viovy and Peter Rayner LSCE, CEA-CNRS-UVSQ Gif sur Yvette, France Carbon Fusion Meeting 9-11 May 2006
Background delayed fall earlier spring Jan Jul Aug Dec As temperature is rising, the length of growing season usually increases. How will the net Carbon Uptake Period respond to the warming ?
meteorological forcing output variables sensible & latent heat fluxes, CO2 flux, net radiation rain, température, humidity, incoming radiation, wind, CO2 SECHIBA energy & water cycle photosynthesis ORCHIDEE Dt = 1 hour LAI, roughness, albedo soil water, surface temperature, GPP LPJ spatial distribution of vegetation (competition, fire,…) NPP, biomass, litterfall STOMATE vegetation & soil carbon cycle (phénologie, allocation,…) vegetation types vegetation types Dt = 1 year Dt = 1 day prescribed vegetation Global biospheric model ORCHIDEE
ORCHIDEE model simulations • Spin up (1000 y) using 1901 climate data, and 1850 CO2 concentration • Simulate from 1850 to 1900 using 1901-1910 climate data, and corresponding every year CO2 concentration. • Simulate from 1901 to 2002, using corresponding every year climate data and CO2 concentration. Save every day C flux from 1980 to 2002.
Comparison of spring (AM) LAI Latitude (degree)
Comparison of autumn (SO) LAI Latitude (degree)
Satellite sensor change Interannual Variability in LAI Spring SDORCHIDEE = 0.04 SDGIMMS = 0.06 SDPAL = 0.19 Autumn SDORCHIDEE = 0.02 SDGIMMS = 0.07 SDPAL = 0.13
D B A C Define growing season and carbon uptake periods CUP GSL Growing Season From rate of change of LAI Carbon Uptake From NEP zero-crossing dates C = net carbon uptake start D = net carbon uptake end CD = Carbon Uptake Period (CUP) A = growing season start B = growing season end AB = growing season length (GSL)
Carbon Uptake Growing season (phenology) Mapping the growing season and carbon uptake timing Onset date increases with increasing latitude CUP start occurs later than GS start (because of spring respiration) Start (day) The distribution of End date in autumn is less uniform than in spring, (reflects vegetation type, as well as water / temperature limitations on plant growth). End (day) Shortest GSL = Central Siberia near the Arctic coast (4 months). Shortest CUP = Northern Eurasian forests and water limited steppes - also show the shortest GS length. Duration (days) Derived from ORCHIDEE simulation
Trends GSL and CUP during 1980-2002 Spring dGSLstart/dt = 0.16 days/yr dCUPstart/dt = 0.19 days/yr Same response of CUPstart and GSLstart to warming trend RGSLstart-temp =-0.91 P<0.001 RCUPstart-temp = -0.62 P=0.002 Autumn dGSLend/dt = 0.14 days/yr dCUPend/dt = -0.07 days/yr Opposite response of CUPend and GSLend to warming trend ! RGSLend-temp = 0.71 P<0.001 RCUPend-temp = -0.51 P=0.01 ORCHIDEE > 25°N
Carbon Uptake Trends Growing season Trends Mapping the trends More than 70% of the study region exhibits an advancing trend in the GSL start, especially in Eurasia. In North America, large regions show delayed trends in the CUP start Beginning GSL: most of northern North America shows a trend towards later GSL end, BUT there is a trend to earlier GSL end in temperate Western Eurasia (Europe). CUP: 70% of the study region display a trend towards an earlier CUP end. End GSL length : Trends to increasing GSL over high latitude regions, usually as a result of earlier beginning of growing season in Eurasia and later end of growing season in North America CUP length : North America shows a trend to shorter CUP length, Eurasia has the opposite behaviour Length Derived from ORCHIDEE
Comparison with satellite observation Spring Autumn (1) Period from 1980-2002; (2) Period from 1982-1998; (3) Period from 1988-2000
Atmosphere CO2 measurements • Although Keeling et al. (1996) found that there were no significant long-term changes in the upward zero crossing time at site Mauna Loa from mid-1970s to 1994, pronounced advancement at a rate of 0.77 days yr-1 (R=-0.65, P=0.001) is observed in the period of 1980-2002.
Temperature vs. Carbon Uptake Period Spring RBRW = -0.85, P<0.001 RMLO= -0.40, P=0.056 Autumn RBRW = -0.60, P=0.003 RMLO= -0.59, P=0.005 (excluding 1992, 1993)
Differential response of gross C Fluxes to the warming trend in Northern Hemisphere (>25°N) Spring: Warm temperatures accelerate growth more than soil decomposition. The annual relationship of NEP to temperature is positive => Warming enhances carbon uptake Autumn: Warm autumn accelerate growth less than soil decomposition. The annual relationship of flux to temperature is negative. => Warming reduces carbon uptake Derived from ORCHIDEE
Autumn (SON) temperature vs. C Flux Derived from ORCHIDEE
Conclusion • Most of the study region exhibited extending of GSL, usually as a result of earlier vegetation green-up in Eurasia and later vegetation senescence in North America, which strongly supports a lengthening of growing season and greening trend at northern hemispheric observed in the past two decades. • Due to parallel stimulating soil carbon decomposition, increase in GSL does not necessarily lead to increase in CUP and eventually result in higher C net uptake. • Autumn warming does not benefit terrestrial net C uptake through postponing vegetation growing season end in the northern mid and high latitudes.
Relevance to IGCO • Need for in situ phenological data • Need of long flux time series to confirm processe • Need for snow cover / frozen status of soil data • Long term satellite biophysical products (large differences between sensors & data processing) • New CO2 column satellite obseravtions may allow an unprecedented quantification of the spatial distribution in the CO2 seasonal cycle -> regional trends detection • Integration of surface with atmospheric information