I tried a (aT + bP + c)^d type of exponential empirical function
fitting but so far not very exact.

In any case I revised the koch chart into more details with both F and
C temp units and with take off / climb "factors" from SL - easier to
use. Everybody is welcome to the chart at
http://www.abri.com/sq2000/Koch-Chart.gif

My experimental with my engine/prop combo does not have POH values for
takeoff-climb values with altitude and I have flown some over
mountains. Thats why the chart is important to me.