Rainfall recycling needs to be considered in defining limits to the world's green water resources.

In PNAS, Schyns et al.’s (1) analysis of the limits to the world’s green water resources starts from the commonly made assumption that rainfall is an independent, exogenously determined variable, not influenced by anthropogenic land cover change. An increasing body of research now suggests that this is incorrect (2). It matters where rainfall comes from, and where evapotranspiration (ET) goes to.

While blue and green water as distinct hydrologic concepts advance scientific discussion, setting potential limits to green water use requires an understanding of rainfall recycling across the globe. Portraying water management as a green water versus blue water allocation issue avoids the up- and downwind dynamics of the terrestrial atmospheric moisture contribution to rainfall that depend on ET. The potential to maximize water availability across terrestrial surfaces depends on adequate green water use in strategic locations. The way to increase water availability across terrestrial surfaces—in particular in locations more distant from upwind coasts—is to increase (not reduce) the aggregated green water use.

Data on the global hydrologic cycle and its principal hydrologic flows (3) show that in an average year, ∼40,000 km3 (net) of ocean evaporation enters the terrestrial atmosphere. When equally distributed, this accounts for 268 mm of rainfall. However, average annual terrestrial precipitation of 779 mm requires 116,000 km3 of atmospheric moisture; more than 60% of this is derived from green water use by trees, forests, croplands, other vegetation, wetlands and soils, plus some evaporation of blue water from water bodies or irrigated agriculture. On average, a drop of water entering the atmosphere over land from the ocean falls 2.6 times as rainfall before returning to the ocean in river flow. There is, in fact, no compelling reason that the 2.6 value, and thus the amount of recycled rainfall, cannot increase or decline based on future land use change (via forest landscape restoration or continued deforestation). Location- and timewise, atmospheric moisture derived from blue water use in irrigation areas differs from that of green water use in water-tower forests.

The inevitable and compelling conclusion of an overwhelming number of studies focusing on the concept of precipitation recycling and its implications for regional and continental water availability suggests that terrestrial rainfall is a modifiable quantity (4–7). We have argued elsewhere that attention should be paid to the “full hydrologic cycle,” which recognizes the role of forest–water interactions, both at the basin level and across basins at regional and continental levels (2). Precipitationshed algorithms (7) allow the identification of spatial dependence (teleconnections) for any specified fraction of rainfall over any specific area (e.g., a watershed). Thus, for example, the dependence of Blue Nile rainfall and runoff to the Nile River on White Nile and Congo basin ET deserves attention (8). Precipitationshed governance, as an extension of watershed governance, however, remains in its infancy (9, 10). Without actually knowing what defines a “safe operating space for humanity” with respect to the supply of water, it seems premature to define a safe operating space for the planet’s foliage (and fauna) on the basis of the existing water footprint accounting principles.