Interannual variability of crop water footprint

The crop water footprint, CWF, is a useful tool to investigate the water-food nexus, since it measures the water requirement for crop production. Heterogeneous spatial patterns of climatic conditions and agricultural practices have inspired a flourishing literature on the geographic assessment of CWF, mostly referred to a fixed (time-averaged) period. However, given that both climatic conditions and crop yield may vary substantially over time, also the CWF temporal dynamics need to be addressed. As other studies have done, we base the CWF variability on yield, while keeping the crop evapotranspiration constant over time. As a new contribution, we prove the feasibility of this approach by comparing these CWF estimates with the results obtained with a full model considering variations of crop evapotranspiration: overall, the estimates compare well showing high coefficients of determination that read 0.98 for wheat, 0.97 for rice, 0.97 for maize, and 0.91 for soybean. From this comparison, we derive also the precision of the method, which is around ±10% that is higher than the precision of the model used to evaluate the crop evapotranspiration (i.e., ±30%). Over the period between 1961 and 2013, the CWF of the most cultivated grains has sharply decreased on a global basis (i.e., -68% for wheat, -62% for rice, -66% for maize, and -52% for soybean), mainly driven by enhanced yield values. The higher water use efficiency in crop production implies a reduced virtual displacement of embedded water per ton of traded crop and as a result, the temporal variability of virtual water trade is different if considering constant or time-varying CWF. The proposed yield-based approach to estimate the CWF variability implies low computational costs and requires limited input data, thus, it represents a promising tool for time-dependent water footprint assessments.