Typically, the lower the freq the farther the groundwave will go, all else being the same. Also the angle of radiation off the transmitter and receiver antennae come into play. And the height of the ionospheric layer the signal reradiates from comes into play at times. As you've seen, the freqs that are under the muf and above luf can be employed via nvis, wich is variable on a hourly to seasonal basis. As an example of stable hf line of sight work, two 10/11m beams pointed at each other can do 50 miles or more of very stable contact. A friend used to talk to his wife via 11m ssb from Council Bluffs Ia to Sioux City Ia (about 94 miles) via a 11m vertical atop his tower fed by his Ic706, her car had a Ic706 also and some mobile cb antenna, 100w out at each end.
To pull off the same thing at 2.5Mc in the daytime would take the typical broadcasting tower at each end - and keep in mind that short yet efficient antennas at 2Mc and 27Mc are very different from each other as you can easily understand. Also one must consider vertical radiation is most desireable for MW work as the D layer absorbs vertical polarity least at MW/ambc freqs - explains why most every ambc antenna is a vertical that is wavelength calculated to cover the locality as well as possible.
As an example of MW daytime groundwave abilities, an ambc station in Omaha Ne has coverage out to Des Moines Ia to the east, Kansas City to the south, and into Sioux Falls SD to the north, from a vertical fed with 1kw.
http://radio-locator.com/info/KCRO-AMhttp://radio-locator.com/cgi-bin/patg?id=KCRO-AM&h=DIt doesn't hurt that some of the most conductive soil in the US is under their tower.