yield and ET during the growing period—differs significantly among regions due to different climatic and management conditions (Liu et al. 2007; Fader et al. 2010).
This presentation provides an estimation of GW and BW availabilities for countries (with outlooks on river basins) and compares these to the country-specific water requirements for producing a diet of 3,000 kcal cap–1 d–1 (with 80% vegetal products) calculated from local CWP. The resulting new water scarcity indicator is applied for the present situation and for a number of climate change scenarios (from 17 General Circulation Models (GCMs), including direct CO2 ef-fects on plants; A2(r) emissions and population). All calculations were done at high spatial resolution (0.5° global grid) and high temporal resolution (daily, up to 2099), while results are presented as 30-yr country averages (blending results for rainfed and irrigated land) for the present (1971–2000) and for a future time slice (2070–2099 or “2080s”). Most material presented here is taken from the study by Gerten et al. (2011submitted).
Methods and data
We used the global ecohydrological model LPJmL (Bondeau et al.
2007; Rost et al. 2008), which simulates the growth, production and phenology of 9 “plant functional types” (representing natural veg-etation at the level of biomes), grazing land, and 11 “crop functional types” (CFTs, representing the world’s major food crops; see Bondeau et al. 2007 for detailed description). The fractional coverage of grid cells with CFTs was prescribed using a dataset of present cropland distribution combined with a dataset of maximum monthly irrigated and rainfed harvested areas of individual crops (see Fader et al. 2010 for data and further details).
Water requirements and water consumption—and thereby the CWP—of irrigated and rainfed CFTs are distinguished, also on ir-rigated land. We assumed that the irrigation water requirements of the CFTs as controlled by their water limitation and by country-wide irrigation efficiencies can always be fulfilled (details in Rost et al. 2008). Seasonal phenology of CFTs was simulated based on 20-yr average climate, allowing for adaptation of varieties and growing periods to climate change (Bondeau et al. 2007). Management/yield was calibrated using statistical data for the period around 2000 by sequentially varying parameters for cropping density and other management features (Fader et al. 2010). We assumed that neither changes in management nor in the extent of agricultural land will occur in the future.
The BW resource and the GW resource (both in m3 yr–1) were com-puted at the grid cell level and then summed up for the respective country, presuming that the food produced with this water is distrib-uted evenly within a country rather than produced and consumed within individual grid cells or within river basins. For determining the BW resource, we computed the runoff generated in a river basin and redistributed it across all cells within the basin in proportion to their accumulated discharge. This way, cells with a high share of discharge relative to the basin’s total discharge were assigned a relatively high BW resource. The BW resource per country was given by the sum of this runoff for all grid cells. We also assumed that just 40% of this resource are actually available for food production, e.g. to account for environmental flow requirements (further details in Gerten et al., 2011submitted).
The GW resource was computed as the precipitation water that evapotranspires from crop and pasture areas within a country. GW contribution from grazing land was constrained either by total grass-land ET or by the global average water requirement of 251 m3 cap–1
yr–1 from grazing land. The GW resource also reflects the extent of agricultural area, meaning that countries with a large agricultural area may show a relatively high GW resource. The total green and blue water resource GWBW (m3 yr–1) was calculated as the sum of the GW and BW resources in a country.
GWBW availability (in m3 cap–1 yr–1) was calculated by relating the an-nual GWBW resource to the number of people inhabiting a country, assuming that they benefit uniformly across the country from its total water resource. Subsequently, GWBW scarcity was computed for each country as the ratio between the GWBW availability and the water requirement for food production (see the following).
Water requirements for producing a healthy diet of 3,000 kcal cap–1 d–1 with 20% calories from animal products were estimated from the water needs for producing vegetal calories on a country’s present cropland and from a hypothetical livestock sector. Details can be found in Gerten et al. (2011submitted) and only a short summary is provided here. Water needs for the vegetal part were estimated by relating the total amount of calories produced on cropland (inferred from simulated yields using calorie conversion factors from the FAO’s Food Balance Sheets) to the total amount of GWBW consumed on cropland during the growing period, yielding to a global require-ment of 0.409 m3 1,000 kcal–1. Following Rockström et al. (2007), the eightfold amount of water is required to produce an equivalent amount of animal calories. This results in a global average of 1,075 m3 of water per capita and year required for the above specified diet (358 m3 cap–1 yr–1 for the vegetal share plus 716 m3 cap–1 yr–1 for the animal share). Water needs to produce the animal share were attributed to cropland and grazing land assuming a mixed livestock system with a non-grazing and a partly grazing sub-system, each consuming 50% of the water (as in Rockström et al. 2007; see also Gerten et al. 2011submitted). As a result, 840 m3 cap–1 yr–1 are
re-quired to produce food and feed on cropland, and, respectively, 251 m3 cap–1 yr–1 to produce grazed biomass. For the water requirements from grazing land we used for each country the global average of 251 m3 cap–1 yr–1, assuming that grassland management and grazing intensity are not related to cropland productivity.
LPJmL was forced for the period 1901–2000 by monthly values of air temperature, precipitation amounts, number of wet days and cloud cover, taken from the CRU TS 3.0 climate database (Mitchell and Jones 2005; http://badc.nerc.ac.uk/data/cru/). Climate projec-tions for the transient simulaprojec-tions up to 2099 were derived from 17 GCMs under forcing from the SRES A2 emissions scenario (overview of models in Gerten et al., submitted), as follows. First, the individual GCMs’ monthly mean temperatures, precipitation sums and mean cloudiness were brought to 0.5° resolution and smoothed using a 30-yr running mean. Second, anomalies relative to the 1971–2000 mean were calculated for each month of 2001–2099 and applied to the observed 1971–2000 baseline (additive for temperature, and using mixed additive-multiplicative approaches for precipitation and cloudiness, see Gerten et al. (2011). Third, annual CO2 concentrations until 2099 were taken from the Bern CC carbon cycle model. To quan-tify the net physiological and structural effects of atmospheric CO2 concentration and thus on water scarcity, we ran additional LPJmL simulations in which CO2 was held constant at the year 2000 level.
Finally, we used population projections consistent with the emis-sions and climate projections to determine future per capita water availabilities. In the case of A2 we chose the revised “A2r” scenario (Grübler et al. 2007).
Figure 1. Availability of green plus blue water, illustrated at the country level (values in m3 cap–1 yr–1 and averaged over 1971–2000).
Modified after Gerten et al. (2011).
Figure 2. Country-level GWBW requirements (m3 cap–1 yr–1, 1971–2000 average) for producing 3,000 kcal cap–1 d–1 with 80% vegetal and 20% animal products (top), and water scarcity defined as the percentage ratio between GWBW availability (cf. Figure 1) and these requirements (bottom). © American Meteorological Society – used with permission.