Rates of worldwide irrigation totalling 3000-4000 km³ have increased 1-2% annually from around 1960 as needed to feed the increasing global population. While this new consumption is only 3% of the global hydrological cycle our regional model shows that evapotranspiration of about 60% of this additional water is dynamically focussed on dry environments where its greenhouse effect is maximised. Furthermore, water is shown in the model to cause more than 80% of the natural atmospheric global warming of 33 ^oC, charging air with heat by (i) absorption of shortwave solar radiation, (ii) latent heat of evapotranspiration and (iii) greenhouse interception of surface longwave radiation resisting heat flow to space. An estimated increase of 1% in atmospheric water vapour from this extra irrigation since 1960 corresponds to a global increase of about 0.25 ^oC, comparable to the warming estimated from increasing CO₂ since 1960. The 4-6 year El Nino-La Nina cycles result in a global mean temperature range approaching 1 ^oC, with atmospheric water vapour varying 4%. This increased water vapour results from accelerated evaporation of seawater from a warmer Pacific Ocean surface, shown to elevate and cool the maritime atmosphere reducing the outgoing longwave radiation (OLR) to space by up to one-third. By contrast, global warming of only 1 ^oC per century is attributed to the trend of increasing greenhouse gases (GHG) in the atmosphere; in climate models, water is assigned a secondary though important amplifying role, but as a positive feedback from an atmosphere previously heated by other GHGs. Fortunately, this hypothesis of regional warming can be tested in the model, using, for example, the satellite data on regional OLR acquired since the 1970s, relating this to the major flooding of Lake Eyre in 1974 and local trends of increasing irrigation in the Ord River basin in Australia.