A data driven approach to identifying crop rotations in the Global Land Model


* What are we trying to do?

* What are crop rotations? Why are they important?

* How have crop rotations been estimated before?

Crop rotations (the practice of growing crops on the same land in sequential seasons) reside at the core of agronomic management as they can influence key ecosystem services such as crop yields, carbon and nutrient cycling, soil erosion, water quality, pest and disease control.

As the dominant land-use type on Earth, agro-ecosystems cover more than a third of ice-free land surface (Ramankutty et al., 2008). They have a profound impact on the environment which is manifested through global fluxes of greenhouse gases (McCarl and Schneider, 2001), soil carbon dynamics (Lal, 2004), increased surface temperature and drought conditions (Hertel et al., 2010), and provision of ecosystem services (Foley et al., 2011). Human management of these agro-ecosystems, based on economic realities and ecological conditions, can influence both the magnitude and the nature of impact on ecosystem services (Robertson at al., 2000).

A key management activity performed by farmers is the development of crop rotation plans based on economic opportunities and adapted to environmental conditions. Crop rotations have been practiced for thousands of years but crop rotations practiced today are much simpler than those practiced in the past (Bullock, 1992 and Plourde et al., 2013). Compared to a monoculture cropping system supplied with optimum nutrient levels, the practice of crop rotations usually leads to higher yields, which are mainly attributed to improved soil fertility and tilth (Hesterman et al., 1987 and Pierce and Rice, 1988), as well as enhanced pest, disease and weed control (Liebman and Dyck, 1993 and Tilman et al., 2002).

When practiced together with a low-intensity tillage regime, crop rotations can potentially reduce the global warming potential of agro-ecosystems (West and Post, 2002). Conversely, the simplification of agro-ecosystems, through expansion of agricultural land supporting a single crop type is an important cause behind the decline in farmland biodiversity (Bianchi et al., 2006). Consequently, ecosystem services associated with diversified crop rotations, like nutrient recycling, addition of organic matter and microclimate regulation have also deteriorated. Beyond their ecological importance, these ecosystem services provide other tangible benefits. For instance, the suppression of pest populations in crops by natural enemies can reduce yield loss and the need for increased use of pesticides (Landis et al., 2008, Gardiner et al., 2009 and Meehan et al., 2011), although there is uncertainty about the linkage between landscape simplification and pesticide use (Larsen, 2013).

Introduction from Sahajpal et al. 2014



* Should be able to handle land use change from non-crop to crop and vice-versa: extensification and abandonment and fallowing

* Should be linked with fertilizer usage data

* Should show changes in crop rotations: intensification


Our approach is based on the following concepts:

1. Hub crop: Each crop rotation is organized around a hub crop which dominates the rotation, and provides the farmer a competitive edge in the market. E.g. corn is the hub crop in most of the US Midwest.

2. Substituting time by space: Crop rotations exist as a time-series of crops grown on the same parcel of land. At a coarser resolution (say 0.25 degree), we can substitute the temporal dimension of a crop rotation by looking at the relative acreage of different crops in the region. Greater the acreage, greater the share of that crop in the rotation.

3. Using fertilizer data to infer monoculture versus rotation: Crop rotations reduce the amount of fertilizer needed to produce subsequent crops. Using information on fertilizer usage, we can deduce whether a crop is rotating or a monoculture. We will also correct for fertilizer prices, if available.

4. Local knowledge: Information on historic and present crop rotations practiced in a region will be used to correct and inform the algorithm.


Algorithm to identify crop rotations is as follows:

1. Compute moving (5-year) averages of crop acreage. Use these averages to compute the percentage of a crop rotation occupied by a specific crop.

1.1 If the area occupied by the crops in a region/grid cell is not the same, use the least area crop to determine the crop rotations. The crop with the greatest area will necessarily have some monoculture.

2. Compute decadal average of fertilizer usage by crop.

3. Use the decadal average as a proxy for crop rotation intensity. Correct by price of fertilizer to avoid case of extra fertilizer usage when price is low.