『Summary
　Crop production requires large amounts of green and blue water. We develop the new global crop water model GCWM to compute consumptive water use (evapotranspiration) and virtual water content (evapotranspiration per harvested biomass) of crops at a spatial resolution of 5' by 5', distinguishing 26 crop classes, and blue versus green water. GCWM is based on the global land use data set MIRCA2000 that provides monthly growing areas for 26 crop classes under rainfed and irrigated conditions for the period 1998-2002 and represents multi-cropping. By computing daily soil water balances, GCWM determines evapotranspiration of blue and green water for each crop and grid cell. Cell-specific crop production under both rainfed and irrigated conditions is computed by downscaling average crop yields reported for 402 national and sub-national statistical units, relating rainfed and irrigated crop yields reported in census statistics to simulated ratios of actual to potential crop evapotranspiration for rainfed crops. By restricting water use of irrigated crops to green water only, the potential production loss without any irrigation was computed. For the period 1998-2002, the global value of total crop water use was 6685 km³ yr^-1, of which blue water use was 1180 km³ yr^-1, green water use of irrigated crops was 919 km³ yr^-1 and green water use of rainfed crops was 4586 km³ yr^-1. Total crop water use was largest for rice (941 km³ yr^-1), wheat (858 km³ yr^-1) and maize (722 km³ yr^-1). The largest amounts of blue water were used for rice (307 km³ yr^-1) and wheat (208 km³ yr^-1). Blue water use as percentage of total crop water use was highest for date palms (85％), cotton (39％), citrus fruits (33％), rice (33％) and sugar beets (32％), while for cassava, oil palm and cocoa, almost no blue water was used. Average crop yield of irrigated cereals was 442 Mg km^-2 while average yield of rainfed cereals was only 266 Mg km^-2. Average virtual water content of cereal crops was 1109 m³ Mg^-1 of green water and 291 m³ Mg^-1 of blue water, while average crop water productivity of cereal crops was 714 g m^-3. If currently irrigated crops were not irrigated, global production of dates, rice, cotton, citrus and sugar cane would decrease by 60％, 39％, 38％, 32％ and 31％, respectively. Forty-three per cent of cereal production was on irrigated land, and without irrigation, cereal production on irrigated land would decrease by 47％, corresponding to a 20％ loss of total cereal production. The largest cereal production losses would occur in Northern Africa (66％) and Southern Asia (45％) while losses would be very low for Northern Europe (0.001％), Western Europe (1.2％), Eastern Europe (1.5％) and Middle Africa (1.6％). Uncertainties and limitations are discussed in the manuscript, and a comparison of GCWM results to statistics or results of other studies shows good agreement at the regional scale, but larger differences for specific countries.

Keywords: Crop water requirement; Global crop water model; Virtual water content; Irrigation water use; Crop production; Crop yield』

Introduction
Data and methods
　Growing areas and cropping seasons
　Climate and soil data
　Modeling of crop-specific blue and green consumptive water uses
　Modeling of irrigated and rainfed crop productions and potential production losses without irrigation
　Virtual water content and crop water productivity
　Simulation protocol
Results
　Crop water use of blue and green water
　Crop yields, crop production and potential production losses without irrigation
　Virtual water content and crop water productivity
Discussion
　Limitations and uncertainties
　　Input data
　　Method to compute reference crop evapotranspiration
　　Grouping of crops
　　Rice
　　Rooting depth
　　Optimal versus actual water use in irrigated agriculture
　　Method to compute potential production losses without irrigation
　Model validation
　　Comparison to independent estimates of irrigation water withdrawals and consumptive use
　　Comparison to estimates included in the water footprint of nations (Arjen Hoekstra and colleagues)
　　Comparison to irrigated and rainfeld yields in developing countries (FAO)
　　Comparison to independent data on crop water productivity
　Future improvements of GCWM
Conclusions
References