We have been doing some experiments with Twitter and geospatial data. I just made a 3-minute video this morning that demonstrates the extremely limited availability of geospatial data, even in Tweets collected using the GNIP-enabled Power Track for Twitter (the full fire hose). http://www.screencast.com/t/eVrP2y2UWf This may change over time. I have recently talked to Michael Macy of Cornell, who has made strides identifying techniques for using a combination of features to peg the country of most Tweets. My sense, however, is there is much work to be done.