(1) For actual location I'd do though experiments on LASER TX to RX based rotated beams to scan the space with possibly a number of LASERS and detectors to get triangulation and improve detectability.
As this is a straight line tx-rx system it needs at least one end to be at or above turbine height. While that's an obvious challenge it gives a system with a much more positive detection signal than many. Cost of say 3 tall thin towers outside WT periphery with cables between to mount txs or rxs as required would be small compared to cost of WT. (Fine if you are the WT owner. Not so fine if you are a researcher).
(2) LIDAR and RADAR with scanning seem good choices. Aperture smaller easily with LASER. Likely less target sensitivity issues with RADAR.
Long ago some people I met were using std Gunn oscillator door alarm units as person detectors and bandpass filtered the doppler returns to extract joint articulation frequencies from the targets knee joints !!! They said this gave them 100% discrimination compared with eg cats, curtains or falling boxes. You may be able to extract wing beat rate, wing beat patterns, possible bat articulation frequencies and just possibly animal heartbeat. More miraculous things have been found to be possible. Sometimes :-).
After some thinking, a scanning LIDAR pointing upwards at an angle from 3 or more ground stations sounds reasonably doable. LASER scans in a half circle probably using a mirror. (Extra points - co-opt existing mirror assembly and LASER from a LASER printer.( More powerful than from a printer is probably in order but beam residence time on any flying creature is small. Arcs of say 3 scanned beams spaced at 120 degrees around circle around WT meet at apex above WT so you have a3 sided pyramid of scanned light. Nothing can pass through a curtain without being illuminated provided that scan rate is fast enough.
Scan period needs to be usefully faster than l/V where l = taget length and V = target max velocity.
Say l = 100mm (~= 4 inches) and Vmax = 10 m/s (~= 22 mph) then scan period needs to be < 0.1/10 = 0.01 second.
Residence period depends on target width along beam arc and height above projector.
Residence time = Period x W/(2.Pi.h)
w = target width. h = height above projector.
eg bat body 20mm wide (big bat?) at 10m above projector is illuminated for
t = 0.01 x 0.02/(2.Pi.10) = 1E-5/Pi ~= 3 microseconds at a 100 Hz scan rate.
Your Bat May Vary.
That sound very short ! At first glance the arithmetic seems OK.
[[[Feel free to find "purposeful error" left for the ellucidation of the student :-). ]]]
Detection: I'm glad you asked. general detectors need to detect and illuminated beastie by reflection during a few microseconds of illumination time in a fairly wide arc. More detectors = lesser arc.
Easy and obvious way of making this MUCH easier is to have detector aperture scan with teh TX beam. This very greatly increases the RX signal level and allows a focused parallel beam detector.
All sounds a little challenging but doable.
Commercial LIDAR may do the job now.
The above gives you a LASER angle at reflection time - two in some cases.
Add a second triangle of LASERS rotated 60 degrees from the first and you get two hits per passage (at least) and more sequential ones for slower speeds. This allows triangulation with increasing size resolution as speed decreases or beastie size goes up. ie at max speed you get one hit from two beams giving a single point.
At 10% of max speed you get about 10 hits from each of 2 detectors. You may get some trajectory, and size and/or speed information. If 3 or more detectors get hits you almost certainly get both size and speed.
If Murphy is feeling generous you may get some secondary modulation of the beam from wing rate. You MAY get difference in reflectivity from different beastie textures or surface 'finishes'. Pigs may fly :-). One can hope.
Given that a Phalanx gun can apparently not only feed depleted uranium rounds to incoming targets at supersonic speeds but also, I'm told, track its own outgoing rounds and make trajectory corrections (Tom Clancy says so so it must be true, right?) RADAR may not be too bad a choice.
One can imagine wide aperture capture system and then a turreted tracking system that uses a tighter RADAR beam to track the turret on the target. May be easier than it sounds. May not.
WW2 technology allowed tracking of fighter aircraft using acoustic rangefinders - arm and two receivers with one rx pivoting to optimise pickup. Something similar may allow passive acoustic (supersonic) tracking of bats. I imagine that active supersonic tracking of bats may be frowned on. (If it flinches or falls out of the sky it is/was a bat)._______
Why bats above birds?
Pay me for a day and I'll dream you up 5 solutions with moderate implementation details, some of which may even work :-).