Journal cover Journal topic
Geoscientific Model Development An interactive open-access journal of the European Geosciences Union
doi:10.5194/gmd-2017-3
© Author(s) 2017. This work is distributed
under the Creative Commons Attribution 3.0 License.
Development and technical paper
03 Feb 2017
Review status
A revision of this discussion paper is under review for the journal Geoscientific Model Development (GMD).
Tiling soil textures for terrestrial ecosystem modelling via clustering analysis: a case study with CLASS-CTEM (version 2.1)
Joe R. Melton1, Reinel Sospedra-Alfonso2, and Kelly E. McCusker2,3 1Climate Research Division, Environment and Climate Change Canada, Victoria, B.C., Canada
2Canadian Centre for Climate Modelling and Analysis, Climate Research Division, Environment and Climate Change Canada, Victoria, B.C., Canada
3School of Earth and Ocean Sciences, University of Victoria, Victoria, B.C., Canada. Now at University of Washington, Department of Atmospheric Sciences
Abstract. We investigate the application of clustering algorithms to represent sub-grid scale variability in soil texture for use in a global-scale terrestrial ecosystem model. Our model, the coupled Canadian Land Surface Scheme – Canadian Terrestrial Ecosystem Model (CLASS-CTEM), is typically implemented at a coarse spatial resolution (ca. 2.8° × 2.8°) due to its use as the land surface component of the Canadian Earth System Model (CanESM). CLASS-CTEM can, however, be run with tiling of the land surface as a means to represent sub-grid heterogeneity. We first determined that the model was sensitive to tiling of the soil textures via an idealized test case before attempting to cluster soil textures globally. To cluster a high-resolution soil texture dataset onto our coarse model grid, we use two linked algorithms (OPTICS (Ankerst et al., 1999; Daszykowski et al., 2002) and Sander et al. (2003)) to provide tiles of representative soil textures for use as CLASS-CTEM inputs. The clustering process results in, on average, about three tiles per CLASS-CTEM grid cell with most cells having four or less tiles. Results from CLASS-CTEM simulations conducted with the tiled inputs (Cluster) versus those using a simple grid-mean soil texture (Gridmean) show CLASS-CTEM, at least on a global scale, is relatively insensitive to the tiled soil textures, however differences can be large in arid or peatland regions. The Cluster simulation has generally lower soil moisture and lower overall vegetation productivity than the Gridmean simulation except in arid regions where plant productivity increases. In these dry regions, the influence of the tiling is stronger due to the general state of vegetation moisture stress which allows a single tile, whose soil texture retains more plant available water, to yield much higher productivity. Although the use of clustering analysis appears promising as a means to represent sub-grid heterogeneity, soil textures appear to be reasonably represented for global scale simulations using a simple grid-mean value.

Citation: Melton, J. R., Sospedra-Alfonso, R., and McCusker, K. E.: Tiling soil textures for terrestrial ecosystem modelling via clustering analysis: a case study with CLASS-CTEM (version 2.1), Geosci. Model Dev. Discuss., doi:10.5194/gmd-2017-3, in review, 2017.
Joe R. Melton et al.
Joe R. Melton et al.
Joe R. Melton et al.

Viewed

Total article views: 232 (including HTML, PDF, and XML)

HTML PDF XML Total BibTeX EndNote
157 61 14 232 6 14

Views and downloads (calculated since 03 Feb 2017)

Cumulative views and downloads (calculated since 03 Feb 2017)

Viewed (geographical distribution)

Total article views: 232 (including HTML, PDF, and XML)

Thereof 232 with geography defined and 0 with unknown origin.

Country # Views %
  • 1

Saved

Discussed

Latest update: 27 Apr 2017
Publications Copernicus
Download
Short summary
Climate models have large grid cells due to the computational cost of running these complex models. Within grid cells like these, the land surface can vary dramatically impacting the exchange of water, carbon, and energy between the atmosphere and land. We use a technique to determine natural clusters of high-resolution soil texture within large grid cells and use them as inputs to our model. We find relatively low sensitivity to soil texture changes except in very dry regions and peatlands.
Climate models have large grid cells due to the computational cost of running these complex...
Share