Eddy Robertson, Met Office Hadley Centre

Land-use affects the climate by modifying land-atmosphere fluxes of radiation, heat, water, momentum and CO2. UKESM1 includes 2 classes of managed land: pasture and crops. Presently crops cover about 16 million km2 and pasture 8-33 million km2 (definition dependant), that is approximately 11% and 6-23% of the ice free land surface respectively (Figure 1). Representing the carbon cycle and land surface properties over these vast areas is important for climate. It is particularly important to capture transitions between natural and agricultural land, which cause a large land-use change climate forcing. Historically, land-use change has caused around 25% of anthropogenic CO2 emissions (Le Quere et al., 2016) and in HadGEM2-ES the non-CO2 effects of land-use change were the 4th largest source of radiative forcing (Andrews et al., 2016).

Figure 1. Present day crop and pasture distribution. The fraction of land covered by crops (left) and pasture (right) in the LUH2 dataset. Pasture area excludes “rangeland” and so is at the low end of pasture area estimates.

Figure 2. Timeseries of historic global crop and pasture area. Data from the LUH2 dataset. Two estimates of pasture area are shown, one excluding “rangeland” (solid blue line) and one including “rangeland” (dashed blue line).

Physiologically, UKESM1 treats both pasture and crops as natural grasses. The key land-use change signal to capture is the transition between forests and agriculture. Forests store more vegetation and soil carbon than agriculture and they tend to be taller, darker and transpire more. Deforestation releases vegetation carbon to the atmosphere, tropical forests store more carbon than boreal forests, so tropical deforestation releases more CO2. In addition to emitting CO2, boreal deforestation causes a large increase in albedo, which acts to call the climate; albedo increases because grasses are brighter than trees and also because grasses are more easily covered by snow. Tropical deforestation also causes an increase in albedo, but this is out-weighed by a reduction in transpiration that causes surface warming. UKESM1 is forced by time varying distributions of pasture-land and cropland (Figure 2), as either expands the area of natural vegetation is reduced, it is assumed that natural grasses are replaced first, followed by shrubs and then trees. The vegetation carbon removed by expanding agriculture is partitioned between the soil carbon store and 3 wood product pools. Each wood product pool decays at a different rate, releasing CO2 to the atmosphere.

Within the pasture area, the TRIFFID dynamic vegetation scheme is used to determine the fraction covered by the C3 and C4 pasture plant functional types (PFTs), with the residual fraction being covered by bare soil. This produces a PFT distribution that is consistent with the model’s climate. The PFT distribution in the crop areas is calculated in the same way.

A distinction is made between crops and pasture; crops are fertilized and harvested. Crops are “perfectly” fertilized, that is, exactly enough nitrogen is added to the soil to meet crop nitrogen demand, so the crop PFTs are effectively not nitrogen limited. Crop harvest is represented by diverting a fraction of crop PFT litter to the atmosphere instead of to the soil. The main purpose of the crop harvest parametrization is to reduce soil carbon after a transition from natural vegetation to crops, but it also provides a useful measure of crop productivity.


Andrews et al., 2016, Effective radiative forcing from historical land use change, Clim. Dyn., doi:10.1007/s00382-016-3280-7, available at https://link.springer.com/article/10.1007/s00382-016-3280-7.

Le Quere et al., 2016, Global Carbon Budget 2016, Earth syst. Sci. Data, doi:10.5194/essd-8-605-2016, available at http://www.earth-syst-sci-data.net/8/605/2016/.

LUH2, data release v2h (12/14/16), data available at http://luh.umd.edu/data.shtml