Improve canopy radiative transfer and electron transport modeling

Land surface models typically use broadband RT modeling, potentially resulting in biased PPAR and J calculations. How can we improve this?
LSM
Paper
Author

Yujie Wang

Published

April 7, 2024

Quantifying canopy radiative transfer (RT) is crucial for modeling all the subsequent processes, such as canopy energy balance and leaf photosynthesis. The most widely used canopy RT schemes in current vegetation and land surface models are the two-stream scheme reported by Sellers and the scheme described by Campbell and Norman, both of which dated back to the 1980s (maybe even earlier). These schemes use two broad bands to represent the continuous solar radiation spectra: photosynthetically active radiation (PAR, some also use visible light) and near-infrared radiation (NIR). As plants generally absorb more PAR than NIR (even the green light absorption is higher than NIR), using a constant approximation for PAR seems okay in the old days given the limitation in computation resources and lack of knowledge on how plant traits impact the absorption ratio.

Nevertheless, there are some well-recognized problems in the broadband models, particularly for natural radiation:

Fig. 1 Absorption coefficients of main leaf constitutes and chlorophyll content’s impact on \(f_\text{APAR} \cdot f_\text{APAR}\).

If we consider the photon absorption and electron transport as a series of partitioning problems, we have \[ J_\text{PSII} = \text{PAR} \cdot f_\text{APAR} \cdot f_\text{PPAR} \cdot \frac{1}{2} \cdot \frac{K_\text{pmax}}{K_\text{d} + K_\text{n} + K_\text{pmax} + K_\text{f}}, \] where

In this simplified equation,

However, none of these has been explicitly accounted for in the traditional land surface models, which adopt broadband RT schemes. As we now have more complicated and powerful RT models and an increasing amount of plant trait data, it would be a good opportunity to move forward and use the more accurate hyperspectral canopy RT scheme in future modeling practice. When examining how the use of hyperspectral and broadband RT schemes may impact global simulation results, we found that the globally integrated GPP may differ by 8% while the grid level difference may be > 20%. For a thorough analysis, please refer to the published paper.

Given the substantial impacts, we highly recommend other land surface models use a hyperspectral RT scheme in their future model versions. Or at least some corrections need to be made to account for the plant traits- and light source-related changes in \(f_\text{APAR}\) and \(f_\text{PPAR}\). Please refer to our published paper for the correction methods we proposed in the Journal of Advances in Modeling Earth Systems.