Error message

Deprecated function: The each() function is deprecated. This message will be suppressed on further calls in _menu_load_objects() (line 579 of /home/cnrcwp/cnrcwp_drupal/includes/

Evaluation of geographical distribution of vegetation over North America

Our terrestrial ecosystem plays a vital role in regulating climate and weather through land-atmosphere exchange and contributes in mitigating climate change. Until now the projected sink of atmospheric CO2 is uncertain due to disagreements among the Earth system models (ESMs) due to differing responses of their terrestrial ecosystem modules to future changes in atmospheric CO2. This uncertainty arises primarily because of the differences in the strength of the CO2 fertilization effect on the land carbon cycle components but also because of differences in the response of vegetation. Models differ in how the spatial distribution of vegetation, and its composition, changes in response to changing climate and increasing CO2. These differences are also resolution dependent. For example, models with coarse grid resolutions cannot explicitly resolve climatic niches, which in turn potentially contribute to biases in simulated vegetation distribution.

In this study, we evaluate the competition module of the CLASS-CTEM modelling framework at the regional scale over the North American domain at 1° spatial resolution. This resolution is much finer than the resolution (3.75°) used in the previous studies. The model performance is assessed over North America by comparing the simulated geographical distribution of plant functional types (PFTs) with two observation-based estimates. The model successfully reproduces the broad geographical distribution of trees, grasses and bare ground. In particular, compared to the two observation-based estimates, the simulated fractional vegetation coverage is lower in the arid south-west North American region and higher in the Arctic region. The lower than observed simulated vegetation coverage in the south-west region is attributed to lack of representation of shrubs, limited number of PFTs in the model and plausible errors in the observation-based data sets. The observation-based data indicates vegetation fractional coverage of more than 60% in this arid region, despite only 200-300 mm of precipitation that the region receives annually and observation-based leaf area index (LAI) in the region lower than one. The higher than observed vegetation fractional coverage in the Arctic is due to the lack of representation of moss and lichen PFTs and also because of inadequate permafrost soil layers in the model as a result of which the C3 grass PFT performs overly well in the region. The model generally reproduces the broad spatial distribution and the total area covered by the two primary tree PFTs (needleleaf evergreen and broadleaf cold deciduous trees) reasonably well. The model shows increasing trend of simulated LAI and fractional coverages of tree PFTs after 1960s which indicates the response of warming due to anthropogenic activities. However, differences between observed and simulated PFT coverages highlight limitations in the model and provide insight into physical and structural processes that need improvement.

Figure: Spatial distribution of total vegetated coverage across North America. Simulated, observation-based, and differences are presented in the left, middle and right columns, respectively. The differences column includes model biases with respect to WANG06 (Wang et. al, 2006; top panel) and MODIS (Moderate Resolution Imaging Spectroradiometer; middle panel), and the difference between the two observation-based estimates (bottom panel). Root mean square difference (rmsd) and coefficient of determination (r2) are also shown in each case.

Add new comment