I am asked this question regularly by people concerned about cloud seeding and rainfall enhancement. Can country A steal water from neighbour country B by making the clouds rain over country A, so that nothing is left for country B?
Answer: no, even if we were very, very good at artificially making the clouds produce rain. (We are not, by the way.)
I like to explain this by looking at the global perspective: how much water is there in the atmosphere, how fast does this water get depleted and how fast does it get topped up? This turns out to have some surprising answers.
First of all, the total amount of water vapour in the atmosphere varies tremendously with location, time of year, etc. But a global mean amount is about 25kg of water for each square metre. The schematic below shows the numbers in a simple pictorial.
So about 25kg, or 25,000g of water is present in the atmosphere for each square metre. This water is overwhelmingly in the form of water vapour. (By the way, this number is called Total Water Path.)
Of this 25kg, only about 50g is actually in the form of liquid water (or ice) for each square metre. This is again a rounded global average. In tropical clouds this number goes up to above 250g for each square metre. (This number is called Liquid Water Path.)
So the liquid water path, the water and ice in clouds, is on average a factor 500 lower than the total water path, the water vapour available in the atmosphere. Looking up at the skies we normally notice all the clouds, but those clouds only represent about a fifth of a percent of the total water in the atmosphere.
Those clouds produce rain. On a global average this turns out to be approximately 2500g per square metre per day, or about 2.5mm/day.
But this water gets replenished by evaporation (mostly over the oceans), and the rate of evaporation is the same as the rate of precipitation, 2.5kg per square metre per day, so that the total water content of the atmosphere, on a global mean, stays about the same.
So if we were to remove all the water suddenly from the atmosphere, all 25kg for each square metre (this would produce a uniform layer of water of 1 inch deep across the whole surface of the earth), it would take about 10 days for the atmosphere to fill up again by evaporation.
Compare this with the liquid water in clouds: if all the clouds would rain out suddenly (this is a Gedankenexperiment: there is no physically realistic mechanism to suddenly precipitate out all clouds) it would take only about half an hour to replenish those clouds again.
The amount of liquid water stored in clouds is incredibly small considering typical rainfall rates and condensation rates. Clouds are very transient systems: large amounts of water get processed through clouds, but at any time only a small amount of liquid water is contained in the clouds, compared to how much vapour there is available for condensation.
So when a cloud gets removed, that is a very small perturbation on the total water balance: it can locally produce a large amount of rain, but there is lots of water vapour available to replenish the clouds.
(Think of a typical shower in the UK, producing perhaps 5mm of rain, corresponding to 5kg/m2. This is much more than the amount of liquid water that would have been in the clouds before the shower. In other words, while the cloud is producing a rain shower it draws in the majority of that water from other sources, typically humid updraughts where water condenses into the cloud.)
Another way of saying this is that rain production is not limited by water availability. It is in fact limited by the physical processes that can transform water vapour into cloud drops and subsequently into raindrops.
In rainfall enhancement operations (cloud seeding), we seek to enhance this latter process. There is usually no shortage of water, but often the natural set up is such that this water cannot transform efficiently into rain.
One example is presented by the arid regions of the Middle East: the biggest issue with rain formation there is the enormous amounts of dust in the atmosphere: this stimulates the formation of lots and lots of tiny cloud drops and haze drops. But if the liquid water is in the form of many very small drops, this strongly suppresses the formation of rain, which requires coalescence of these small drops into much much bigger drops. Typically, many millions of small drops are needed to produce one rain drop. This turns out to be a highly inefficient process when you start off with many small drops.
So it is then clear that removing a cloud through cloud seeding does not prevent the formation of clouds further downstream. The cloud drops over country A are not the same as the cloud drops over country B even if the same cloud had blown over the border: clouds are transient systems where lots of water gets processed, mainly by condensation and evaporation, but only very little water is actually contained in them.
It takes only a very short while for a cloud to re-establish given the right circumstances.
Theoretically it is still possible for one farmer to steal the cloud from a neighbouring farmer (think of those big American farms) because at those scales we are essentially considering pretty much the same cloud. However cloud seeding operations are not nearly as targeted nor as efficient as to make this a current reality.
All numbers quoted are available from open scientific sources on the internet. A good overview of some of the numbers is in the academic paper by Trenberth et al. (2007): https://doi.org/10.1175/JHM600.1
Reading physics 