Modeling terrestrial ecosystems: Biogeophysics & canopy processes

Transcription

1 Modeling terrestrial ecosystems: Biogeophysics & canopy processes Gordon Bonan Na3onal Center for Atmospheric Research Boulder, Colorado, USA CLM Tutorial 2016 Na3onal Center for Atmospheric Research Boulder, Colorado 12 September 2016 NCAR is sponsored by the Na3onal Science Founda3on

2 Role of land surface in Earth system models 2 Provides the biogeophysical boundary condi3ons at the land-atmosphere interface e.g. albedo, longwave radia3on, turbulent fluxes (momentum, sensible heat, latent heat, water vapor) Par33ons available energy (net radia3on) at the surface into sensible and latent heat flux, soil heat storage, and snow melt Par33ons rainfall into runoff, evapotranspira3on, and soil moisture Evapotranspira3on provides surface-atmosphere moisture flux River runoff provides freshwater input to the oceans Provides the carbon fluxes at the surface (photosynthesis, respira3on, fire, land use) Updates state variables which affect surface fluxes e.g. snow cover, soil moisture, soil temperature, vegeta3on cover, leaf area index, vegeta3on and soil carbon and nitrogen pools Other chemical fluxes (CH 4, Nr, BVOCs, dust, wildfire, dry deposi3on) Land surface model cost is actually not that high ( ~10% of fully coupled model)

3 Role of land surface in Earth system models 3 The land surface model solves (at each 3mestep) Surface energy balance (and other energy balances, e.g. in canopy, snow, soil) S + L = S + L + E + H + G S, S are down(up)welling solar radia3on L, L are down(up)welling longwave radia3on is latent heat of vaporiza3on, E is evapotranspira3on H is sensible heat flux G is ground heat flux Surface water balance (and other water balances such as snow and soil water) P = (E S + E T + E C ) + (R Surf + R Sub-Surf ) + SM / t P is rainfall E S is soil evapora3on, E T is transpira3on, E C is canopy evapora3on R Surf is surface runoff, R Sub-Surf is sub-surface runoff SM / t is the change in soil moisture over a 3mestep Carbon balance (and plant and soil carbon pools) NPP = GPP R a = ( C f + C s + C r ) / t NEP = NPP R h NBP = NEP Fire Land Use NPP is net primary produc3on, GPP is gross primary produc3on R a is autotrophic (plant) respira3on, R h is heterotrophic (soil) respira3on C f, C s, C r are foliage, stem, and root carbon pools NEP is net ecosystem produc3on, NBP is net biome produc3on

4 Model design philosophy 4 Coupling with the atmosphere every model 3mestep is a fundamental constraint (< 30 minute 3mestep) So is the need to represent the global land surface, including Antarc3ca, the Tibetan Plateau along with forests, grassland, croplands, tundra, desert scrub vegeta3on, and ci3es Conserva3on of energy and mass is required We strive to develop a process-level understanding across mul3ple ecosystems and at mul3ple 3mescales (instantaneous, seasonal, annual, decadal, centuries) Top-down, empirical modeling Thornthwaite: Monthly poten3al evapotranspira3on driven by air temperature E E p p a 16 L N 10T = I Priestley Taylor equa3on: Daily poten3al evapotranspira3on driven by radia3on s Rn = α s + γ λ Process modeling Penman-Monteith equa3on FvCB photosynthesis model Ball-Berry stomatal conductance model Fick s law of diffusion Darcy s law and Richards equa3on (soil water) Fourier s law (heat conduc3on) Produc3on efficiency model driven by radia3on and empirical scalars GPP = εs f ( T) f ( θ) f ( VPD) Annual NPP driven by temperature and precipita3on 3000, 1+ exp( T) NPP = min exp( P)

5 Lack of a common language 5 Flux is propor3onal to the driving force: Flux = propor3onality constant * gradient of driving poten3al Describes heat flow in soil (Fourier s law), water flow in soil (Darcy s law), turbulent fluxes (Fick s law) Dimensionless drag coefficient Dimensionless soil moisture factor kg m 2 s 1 kg m 3 m s 1 kg kg 1 Atmospheric models: Land surface models: Biometeorology: ( ) E = ρc u q q ρ E = r c w s a ( q q ) a s + r s a ( ) p λe = es ea gw γ Pa β mol m 2 s 1 mol mol 1 mol m 2 s 1 resistance (s m 1 ) m s 1 Plant physiology: ( ) E = q q g s a w

6 Model name 6 Model name depends on discipline: Atmospheric sciences land surface model soil vegeta3on atmosphere transfer model Hydrology hydrologic model (SVAT with lateral fluxes) Ecology biogeochemical model dynamic global vegeta3on model ecosystem demography model

7 The Community Land Model 7 Fluxes of energy, water, CO 2, CH 4, BVOCs, and Nr and the processes that control these fluxes in a changing environment Surface energy fluxes Hydrology Biogeochemistry Oleson et al. (2013) NCAR/TN-503+STR (420 pp) Lawrence et al. (2011) J. Adv. Mod. Earth Syst., 3, doi: /2011MS Lawrence et al. (2012) J Climate 25: SpaSal scale 1.25 longitude la3tude ( grid), ~100 km 100 km Temporal scale 30-minute coupling with atmosphere Seasonal-to-interannual (phenology) Decadal-to-century (disturbance, land use, succession) Paleoclimate (biogeography) Landscape dynamics

8 Land surface heterogeneity 8 Sub-grid land cover and plant func3onal types Urban 8.3% Glacier 16.7% Lake 16.7% Crop 8.3% Vegetated 50% 1.25 o in longitude (~100 km) o in la3tude (~100 km) The model simulates a column extending from the soil through the plant canopy to the atmosphere. CLM represents a model grid cell as a mosaic of up to 5 primary land units. Each land unit can have mul3ple columns. Vegetated land is further represented as patches of individual plant func3onal types

9 Surface energy balance and surface temperature 9 Surface energy balance: Soil heat storage: (S -S ) + L σt s 4 H[T s ] E[T s ] = soil heat storage c v T T = κ t z z Flux = Δ concentra3on * conductance H = c ( ) p T T g s a ah ( q* [ Ts] qa) λe = λ 1 1 g ah + g c Sensible heat T a g ah T s Atmospheric forcing S - Solar radia3on (vis, nir; direct, diffuse) L - Longwave radia3on T a - air temperature q a - atmospheric water vapor u - wind speed P - surface pressure Surface properses S - reflected solar radia3on (albedo) - emissivity g ah - aerodynamic conductance (roughness length) g c - surface conductance k - thermal conduc3vity c v - soil heat capacity With atmospheric forcing and surface proper3es specified, solve for temperature T s that balances the energy budget

10 10 Turbulent fluxes logarithmic profiles * 0 () ln ( ) m u z d u z k z ψ ζ = Logarithmic wind profile in atmosphere near surface with z 0 roughness length, d displacement height, and ψ(ζ) corrects for atmospheric stability * 0 * 0 ( ) ln ( ) ( ) ln ( ) s h h s w h z d z k z q z d q z q k z θ θ θ ψ ζ ψ ζ = = Bonan (2016) Ecological Climatology, 3rd ed (Cambridge Univ. Press)

11 11 Turbulent fluxes logarithmic profiles * 0 () ln ( ) m u z d u z k z ψ ζ = with z 0 roughness length, d displacement height, and ψ(ζ) corrects for atmospheric stability * 0 * 0 ( ) ln ( ) ( ) ln ( ) s h h s w h z d z k z q z d q z q k z θ θ θ ψ ζ ψ ζ = = Bonan (2016) Ecological Climatology, 3rd ed (Cambridge Univ. Press)

12 Plant canopies 12 Bonan (2016) Ecological Climatology, 3rd ed (Cambridge Univ. Press)

13 CLM perspecsve of the land surface 13 Flux to atmosphere uses MOST T a H, λe, E, τ x, τ y L, albedo T rad Surface is an imaginary height (where wind speed extrapolates to zero) T 2m T s T g T v Sunlit Shaded Big-leaf canopy without ver3cal structure Deardorff (1978) JGR 83C: Dickinson et al. (1986) NCAR/TN-275+STR Dickinson et al. (1993) NCAR/TN-387+STR

14 RadiaSve transfer 14 Bonan (2016) Ecological Climatology, 3rd ed (Cambridge Univ. Press)

15 Two-stream radiasve transfer 15 CLM uses the two-stream approxima3on (Dickinson, Sellers) di dx di dx ( β) ωl βωl β0 ωl, b = 1 1 KI d KI d KI b sky be ( ) ( ) b = 1 1 β ωl KI d + βωlki d + 1 β0 ωlki b sky, be Kx Kx

16 How do we scale from leaf to canopy? 16

17 Leaf temperature and fluxes 17 Leaf energy balance: T c Q T c ( T T ) g λ q ( T ) q g t l 4 L = a 2εσ l l + 2 p l a bh + * l a l Atmospheric forcing Q a - radia3ve forcing (solar and longwave) T a - air temperature q a - water vapor (mole frac3on) u - wind speed P - surface pressure Leaf properses l - emissivity g bh - leaf boundary layer resistance g l - leaf resistance to water vapor c L - heat capacity With atmospheric forcing and leaf proper3es specified, solve for temperature T l that balances the energy budget

18 Leaf boundary layer 18 Bonan (2016) Ecological Climatology, 3rd ed (Cambridge Univ. Press)

19 Stomatal gas exchange 19 Leaf Cuticle Guard Cell CO 2 H 2 O Guard Cell Photosynthetically Active Radiation Chloroplast light CO H 2 O CH 2 O + O 2 + H 2 O Low CO 2 Moist Air Stomata Open: High Light Levels Moist Leaf Warm Temperature Moist Air Moderate CO 2 High Leaf Nitrogen Photosynthesis Transpiration Bonan (2016) Ecological Climatology, 3rd ed (Cambridge Univ. Press)

20 Stomatal conductance 20 Scale bar 50 µm Ball-Berry stomatal conductance model g = g + g Ah c s 0 1 n s s Empirical rela3onship between stomatal conductance and photosynthesis and is applied separately to sunlit canopy and shaded canopy Op3miza3on theory Stomata op3mize photosynthe3c carbon gain per unit transpira3on water loss while preven3ng leaf desicca3on ψ > ψ ΔAn ι DsΔg and s l lmin Williams et al. (1996) Plant Cell Environ. 19: Bonan et al. (2014) Geosci. Model Dev. 7:

21 Leaf photosynthesis 21 Farquhar, von Caemmerer, Berry photosynthesis model A = min( A, A ) R n c j d w c is the rubisco-limited rate of photosynthesis, w j is light-limited rate allowed by RuBP regenera3on rubisco-limited rate is cmax( i A = V c Γ* ) c c i + K c(1 + o i/ K o) RuBP regenera3on-limited rate is ( i A = Jc Γ* ) j 4( ci+γ 2 *) Bonan (2016) Ecological Climatology, 3rd ed (Cambridge Univ. Press)

22 Leaf physiological parameters 22

23 Canopy conductance gradients of PAR 23

24 Sunlit and shaded canopy 24 SUNLIT Depth in Canopy SHADED

25 Nitrogen profile 25 Decline in foliage N (per unit area) with depth in canopy yields decline in photosynthe3c capacity (Vcmax, Jmax) sun b ( ) = f x e ( ) = ( ) V x V e 0 Kx n cmax 25 cmax 25 Kx L ( ) ( ) V (sun) = V x f x dx cmax 25 cmax 25 sun 0 L ( ) ( ) V (sha) = V x 1 f x dx cmax 25 cmax 25 sun 0 Bonan et al. (2011) JGR, doi: /2010jg Note: CLM5 has a more complex canopy op3miza3on

26 Plant canopy as a big leaf 26 Stomata Leaf Boundary Layer 2g b Canopy 2g b *L Atmosphere Surface Layer g a T s T s T ac T a T a e i =e*(t s ) g s g b e i =e*(t s ) g s *L g b *L g a e s e a e s e ac e a g s /1.6 g b /1.4 g s /1.6 *L g b /1.4 *L c i c i c s c a c s c a Most models use two-leaves (sunlit and shaded)

27 Flux towers & model validason s: single site comparison, short intensive observing period Boreal Ecosystem Atmosphere Study (BOREAS) Observa3ons Model Bonan et al. (1997) JGR 102D:

28 Flux towers & model validason 2000s: annual cycle, mul3-site comparison (boreal to tropical) Morgan Monroe State Forest, Indiana Stöckli et al. (2008) JGR, 113, doi: /2007JG CLM3.0 dry soil, low latent heat flux, high sensible heat flux CLM3.5 we er soil and higher latent heat flux 28

29 Flux towers & model validason Pg C yr Pg C yr -1 Control 155 Pg C yr Pg C yr -1 Radia3ve transfer for sunlit and shaded canopy Radia3ve transfer and photosynthesis FLUXNET-MTE data from Mar3n Jung and Markus Reichstein (MPI-BGC, Jena) CLM4 overessmates GPP. Model revisions improve GPP. Similar improvements are seen in evapotranspirason Bonan et al. (2011) JGR, doi: /2010jg001593

30 Improved annual latent heat flux x 10 3 km 3 yr x 10 3 km 3 yr x 10 3 km 3 yr -1 Model improvements reduce ET biases, especially in tropics, and improve monthly fluxes Bonan et al. (2011) JGR, doi: /2010jg001593

31 31 Modeling across scales Tower Large basin Global Morgan Monroe State Forest Mississippi basin Water storage (mm) Annual latent heat flux CLM3.5 GRACE CLM3 CLM4 CLM3 dry soil, low latent heat flux Stöckli et al. (2008) JGR, 113, doi: /2007JG D. Lawrence et al. (2012) J Climate 25: Bonan et al. (2011) JGR, doi: /2010jg001593

32 Research areas 32 Surface fluxes Roughness sublayer, mul3layer canopies RadiaSve transfer 3D structure, canopy gaps Photosynthesis Temperature acclima3on, CO 2 response, product-limited rate, C4 plants Stomatal conductance Soil moisture stress, WUE op3miza3on, CO 2 response Canopy scaling Op3mal distribu3on of nitrogen

33 Canopy turbulence and the roughness sublayer 33 Profiles from the CSIRO flux sta3on near Tumbarumba CLM (and most other models) use MOST, which fails above and within plant canopies MOST Obs RSL MOST RSL Harman & Finnigan (2007) Boundary-Layer Meteorol. 123: Harman & Finnigan (2008) Boundary-Layer Meteorol. 129:

34 Two ways to model plant canopies 34 Photographs of Morgan Monroe State Forest tower site illustrate two different representa3ons of a plant canopy: as a big leaf (below) or with ver3cal structure (right) Depth in Canopy SUNLIT SHADED Big-leaf canopy Two big-leaves (sunlit, shaded) Radia3ve transfer integrated over LAI (two-stream approxima3on) Photosynthesis calculated for sunlit and shaded big-leaves Depth in Canopy SUNLIT SHADED MulSlayer canopy Explicitly resolves sunlit and shaded leaves at each layer in the canopy Light, temperature, humidity, wind speed, H, E, A n, g s, ψ L New opportuni3es to model stomatal conductance from plant hydraulics (g s, ψ L )

35 FricSon velocity (momentum flux) 35 US-Ha1, July 2001 US-Var, March 2006 CLM4.5 Obs Model Mid-day Midday CLMml

36 US-Ha1, July 2001 (DBF) 36 CLM4.5 Obs Model CLMml