AWR1642: question about MmwDemo_interFrameProcessing() in mmwave sdk1.1

Part Number:AWR1642

.

Would someone please explain why the scaling factor is chosen as “obj->log2NumDopplerBins+4”?

For the window gain of 2^4, is it the resulting from Q19 to Q15 format conversion?  

            obj->azimuthStaticHeatMap[binIndex].real = (int16_t) (obj->fftOut2D[0].real >> (obj->log2NumDopplerBins+4)); /* +4 since 2D-FFT window has gain of 2^4 */

            obj->azimuthStaticHeatMap[binIndex++].imag = (int16_t) (obj->fftOut2D[0].imag >> (obj->log2NumDopplerBins+4));

the above code excepts are from mmwave sdk-1.1 dss_data_path.c, line 1729-1730.