A small idea for how to reduce the cost of integrating wind through the addition of complexity:
Run irrigation equipment when the wind allows it. Less succinctly: use real-time market-based electricity pricing to influence the short term power consumption of large electric loads to manage the fluctuations of output from large concentrations of must-run power generation (a replacement for running reserves).
I've previously talked about some of the challenges to integrating large amounts of wind energy into a mature power system and mentioned how real time pricing of power for plug in hybrid electric vehicles could be greatly beneficial. I still think it is a good idea, but one with significant barriers to success. One of these barriers being that there are virtually no plug in hybrid electric vehicles. Another being the general principle that the more people are involved in a process the more likely it is to fail. And this process would require that everybody be involved.
So, the same idea applied to a smaller number of larger loads that actually exist today is called for. Simply stated: when there is excess generation in a local portion of the electric system (when the wind is blowing), the price of electricity would be lowered for large customers and they would choose to run their equipment (the irrigation runs). When there is a shortage of generation in a local portion of the electric system (when the wind isn't blowing), the price for large consumers would be raised and they would choose to not run some of their equipment (the irrigation stops running).
This is a good example because the precise timing of irrigation isn't terribly important. Irrigation represents a large fraction of load in rural areas, so if you can influence irrigation scheduling you may be able to make a measurable impact on your ability to cheaply integrate wind power. Most wind power is located in rural areas near the irrigation loads. A potential problem may be that in many places, the highly variable wind inputs may happen at times of the year when there aren't crops to water (winter in the Dakotas, for example, probably isn't a big watering season).
What does this have to do with pumped storage (seen in the title)? The only practical way to store large amounts of energy today is by pumping water up behind a reservoir when you have excess power, then running it through turbines to generate power when you need it. Using irrigation controls as discussed here is basically the same idea, except smarter and more likely to succeed because it doesn't involve building any new infrastructure.