Supporting accounting and decision-making in agricultural water management under data scarcity : chances, limits and potential improvements

Alleviating water scarcity and ending hunger are two Sustainable Development Goals that are inextricably linked, as 70% of water is used for agricultural production. However, assessing and identifying improved water management requires costly field trials or knowledge and data to run agro-hydrological models. This thesis addresses this challenge by integrating an agro-hydrological model with global soil, crop and climate datasets into a user-friendly, web-based agricultural water management assessment tool, the Cool Farm Tool Water (CFTW). This dissertation substantially contributes to the development and assessment of this modelling framework as well as identifies levers to enhance the model performance. Model results were compared to published scientific field trials, eddy covariance observations as well as Water Footprint Network data. Finally, the model framework was applied to five cereals across India between 2005 and 2014, representing a high spatial, inter- and intra-annual heterogeneity. The modelling framework explained 96% of the variability in seasonal water footprints and reduced the error of water footprints by 70% compared to state-level averages by the Water Footprint Network. However, CFTW captured only 50% of the variance in seasonal crop water use. CFTW enables the assessment of the daily dynamics of evapotranspiration, can reproduce daily crop water stress but overestimates irrigation requirements by an average of 40%. It helps identify relative improvements in agricultural water management but lacks the accuracy for absolute assessments. Using local input data reduced the bias from 18.6% to 4.3% for evapotranspiration. While results showed site- and season-specific differences, the greatest model improvement is replacing global precipitation data with local observations, reducing e.g. the bias by 70.6% for daily water use. Water footprints and water use of India revealed the importance of considering crop-, season-, year- and location-specific differences. Total Rabi (dry season) water footprints were between 33.4% and 45.0% lower than Kharif (monsoon) water footprints. However, the Rabi blue water footprints accounted for up to 78.3%. Overall, India increased cereal production by 26.4% without utilising additional water resources, mainly through increases in yield, partly by shifting production to the higher-yielding Rabi season. CFTW supports addressing water scarcity and food security at seasonal timescales and, to some extent, at daily timescales. Global datasets combined with an uncalibrated agro-hydrological model can inform improved agricultural water management by providing water footprints, water use and water stress. However, model outcomes also revealed considerable offsets. These results highlight the difficulties of overcoming data scarcity in decision support tools for agricultural water management. The rapid development of global datasets may help to overcome identified limitations in the future.