You could add the Get Raster Properties tool to our model to get the mean and standard deviation of your raster, naming those model elements raster mean and raster stdev. Once you've done that, use the Raster Calculator to extract your top zone using Con, and wrap that with the Zonal Centroid tool to get your centroid cell. The results (a grid that has one data cell at the zone's centroid) could then be passed to Raster To Point to get a point feature.
(I'm assuming you have access to Spatial Analyst since you were using the Reclassify tool.)
Isn't map algebra awesome?
ZonalGeometry(
Con("raster" > (%raster mean% + (2 * %raster stdev%)), 1),
"VALUE", "CENTROID")
Side note: I noticed that your model is piping your raster to the Summary Statistics tool to get some raster stats. Depending what you are calculating, the Get Raster Properties tool would be a better choice to get raster statistics, as the Summary Statistics tool calculates stats off the raster table, not the raster.