Explicit representation of sunlit and shaded canopy

fraction: fun modeling issues and interesting

WLEF results.

Ian Baker, Joe Berry, C. James Collatz, A. Scott Denning, YingPing Wang, Neil Suits,

Lara Prihodko, Kevin Schaefer, Andrew Philpott

Sunlit/Shaded Scheme in SiB Model Framework

Tcsunlit

e*(Tcsunlit)

Tm

em

Ta

ea

Tg

e*(Tg)

rbsun

rd

ra

rbshade

Tcshaded

e*(Tcshaded)

Replace a single vegetation value with 2 prognostic variables-sunlit and shaded canopy fraction. What changes to the model are required?

Radiation Transfer Submodel

General form of equation•Sunlit fraction= exp(-kL)•Shaded fraction = 1-exp(-kL)

k is a function of solar zenith angle and leaf angle distribution

Light is partitioned between the two canopy elements

Radiative Transfer Submodel (cont.)

0

/ 1 1 expdI dL I I K KL

0

/ 1 1 1 expdI dL I I K KL

Radiation scattered in an upward direction:

Radiation scattered in a downward direction:

These components are convolved with sunlit/shaded canopy fraction based upon Beers’ Law to give full complement of radiative transfer equations. Generally follows Sellers (1985) and Sellers et al (1996).

Canopy Radiative Transfer (cont.)

0 0

_0

T T

T

sunlit beam diffusesunlit sunlit

scattered beam sunlit

L LL L dL L L dL

LL L dL

f fI I I

fI

Sunlit leaves: beam + diffuse + scattered

_0 0

1 1T T

shaded diffuse scattered beamsunlit sunlit

L LL L dL L L dLf fI I I

Shaded leaves: diffuse + scattered

Canopy Nitrogen/Rubisco Velocity Attenuation

Nitrogen decreases with depth in a canopy, in a Beers’ Law relationship similar to LAI.

Multiple ways to represent this, but two popular techniques are:•Normalized: N(L) = N(0)exp(-kL/LT)•Non-Normalized: N(L) = N(0)exp(-KL)

Canopy Nitrogen/Rubisco Velocity Attenuation

(cont.)Does Rubisco Velocity decrease in the canopy 1:1 with Nitrogen (black line)?

Or is Nitrogen re-partitioned with depth in the canopy?•Canopy top: most resources allocated to carboxylation, light capture not as important •Canopy interior: Nitrogen re-allocated to light capture from carboxylation (blue line)

•Beam/diffuse•Saturation at high illumination

Impact of Rubisco

Assumptions on Results

Effect of Rubisco

Treatment on

ResultsWhat happens as more leaf is added to the canopy?

What have we decided to do?

•We like the ‘normalized’ Nitrogen attenuation scheme. It makes sense that the bottom of the canopy has 50% the Nitrogen at the top. Non-normalized schemes can have leaves at the bottom of a dense canopy with 2% Nitrogen compared to top leaves.

•It also makes sense to re-allocate Nitrogen from carboxylation to chlorophyll with depth in the canopy. Not doing so results in excessive photosynthesis in test cases. Caveat: we have not determined the optimal re-allocation scheme for multiple biome types. Also, we are not modifying leaf transmissivity/reflectivity characteristics with canopy depth.

SO WHAT? Or, how can we utilize this new

tool?1. More realistic fluxes of heat, moisture, carbon and momentum

when compared to flux towers2. Higher degree of biophysical

realism: Ability to perform additional botanical/ecological

experiments

But First--Energy Budget

•We know that the eddy covariance fluxes don’t close the energy budget-How should we use this when comparing modeled fluxes to obs?

•What is the diurnal/annual nature of this term?

•Rn = H + LE + G

•Correction factor: C = (Rn – G) / (H + LE)

•But this correction factor has limitations

But First--Energy Budget

Observed

Mod

eled

Limitations to Using ‘Adjusted’ Observations•Can only evaluate model:obs

on a 1:1 plot during restricted periods (i.e. H+LE > 0, Rn>0

•C = (Rn – G) / (H + LE)

•Monthly mean/Diurnal composite?

I need guidance from the observation community for determining a reasonable evaluation strategy for the

models vis-à-vis the energy closure issue!

Improved results

Monthly Mean Values•Summer H decreased•Annual cycle NEE

•Spingtime sign change•Fall return to efflux•Magnitude?

More improved

results

Monthly Mean Diurnal Composite•New code has better shape, when compared to obs•Magnitude?

Taylor Plot

Polar coordinates:

•ANGLE: cos-1(R), where R is the correlation coefficient

•Radius: standard deviation

Taylor Plot

•Correlation coeff of LE, NEE improves•Magnitude of NEE much larger•This plot for all points: how does it break out by month?

Taylor Plot: Sensible Heat

•Amplitude of summertime H decreased•Correlation coeff worse

Taylor Plot: Latent Heat

•Magnitude larger•Correlation coeff better

Taylor Plot: NEE

•Magnitude larger•Correlation coeff better

What else can we do?

Isotopes

Shaded canopy discrimination is around 4‰ smaller that sunlit fraction; this agrees well with observations

Impending projects•Species-specific

•Ewers/Mackay et al; estimate transpiration flux from sap flux data; 4 basic forest types•Model: obtain leaf/species level data from Gutshick, determine model parameters•Initial model results: variable. With new model scheme, can re-address

•Beam-diffuse•‘Global dimming’/aerosol loading (volcanic) change beam/diffuse radiation distribution•Model reproduction/resulting fluxes