Looks to me like the NVIS thing is for people working above 40 m who can't get local communications.
NVIS is for relatively local communications on 160m, 80m, 60m and 40m
the ionosphere is cooperative. Sometimes it is, sometimes it isn't, and to
cover distances of ~100 miles you have to choose the right band for current
conditions. Here in the Pacific NW we haven't seen the critical frequency
reach 40m much if at all for the last 5-6 years, and even now 80m goes
out for most of night. That's not to say you can't make contacts on those
bands, but that they will tend to be further out.
The military NVIS radios that I remember covered 2 to 12 MHz, but rarely
does the Critical Frequency reach the 30m band. (When it does, 10m will
be hopping: a rough rule of thumb is that the Critical Frequency is 4 times
the MUF, so 40m will support NVIS when 10m is open for longer distances.)
Looking at the plots for typical soil types, there is a maximum of about
3 - 4 dB difference among them for the same height. If an antenna works
significantly better in a valley it may be because the trees tend to be
taller in better soil so it is easier to get the antenna up higher.
As long as you are using a horizontal wire dipole on the lower bands,
it probably will have a pattern suitable for NVIS. Then you just need
to be operating below the Critical Frequency for the ionosphere to
support it by reflecting signals straight back down.
Antennas that are NOT particular good for NVIS include verticals and
wires with a "long wire" pattern: for example, an 80m OCFD operated
on 40m has an overhead null (as is the case if it is fed at one end or
as a 1/4 - 3/4 wave dipole), while the same length of wire fed in the
center will have a maximum straight up.