It jumps around wildly on 770MHz P25 sites for me, all the time - from low 3-digit numbers to 5 digits. With a 50ohm dummy load connected to the antenna input, it jumps around as much, but seems the numbers are even higher - all 4 or 5 digits, as high as ~40,000. I have yet to see a good description of what the number actually represents.
The above comment about the dummy load is a huge red flag and concern, if true!
Reading comments from others about the noise levels increasing on 700Mhz as the unit heats up, sounds to me like one of the RF/IF/SDR sections is becoming unstable and noisy/spurious at high internal temperatures.
If this is the case, expect another recall because the punters wont be happy about one of the major bands becoming unusable after only a few hours of operation.
We wont really know for sure until someone actually puts an SDS100 on the bench, feeds it with a calibrated signal generator and logs the receive performance over time. Sadly, while I have the P25 test equipment, I don't have an SDS100. After my last (bad) experience with the 436, I doubt I'll be buying one either ;-)
Trying to quantify receive performance from varying off air signals is just a waste of time. Unless you can back it up with meaningful lab results, its unlikely the manufacturers will take you seriously.
Out of interest, compare noise numbers with a dummy load connected, when the unit is cold and again after a few hours.