One of the advantages of a grid is that it can be used to transfer electrical energy around very efficiently and cost-effectively. There must be an optimally efficient way of storing energy for a district, region or for a whole country, taking into account the trade-offs between the size of the individual storage facilities and the distances travelled by the energy that is stored there and retrieved later. It is very unlikely that this mathematical optimal pattern, which one might call a 'network of temporary storage nodes', would consist of many small storage facilities at a myriad of wind turbine sites. It's much more likely that it would mirror something like the transport logistics networks for the major road, rail and sea freight businesses, (eg one such model is hub-and-spoke), which operate a relatively small number of large 'temporary storage facilities' (ie warehouses) in strategically placed locations with good access and which minimise typical distances to the places where the contents of the warehouses eventually get delivered to. If this sort of model was deployed for the energy network, the storage facilities could be designed (and located) with optimal efficiency of the whole delivery network in mind. It's quite possisble that a by-product of this sort of design would be a resolution of the intermittency challenge of some renewable energy sources.
As a slight digression on this, there are many ways of designing an optimal network, but one intriguing possibility is to learn from the way nature does it. An example is the use of slime mould, as described in this article from 2010.