Cropping at the sensor at time of capture to 15 mpixel is not the same as resampling the 24 mpixel image to 15 mpixesl (binning) or simply collecting a 15 mpixel image in the first place from a full-size DX sensor (with larger, more efficient, light-gathering photocells). By cropping at the sensor, you are basically using your D7100 like a 4/3 format camera, with all the disadvantages of smaller individual photocells with no frame-wide resolution compensation.
The whole point of the 24 Mpixel sensor is to achieve higher spatial resolution in the final image. If you don't need the spatial resolution, then you are best served by downsampling the image (binning) to increase signal to noise. This is done routinely in scientific imaging where we might use 2x2 binning to increase the effective size of a pixel--in this case each "pixel" is a combination of the intensities from each of 4 pixels. It's a little more complicated for a color camera sensor than a single-color sensor, but the principle of binning is approximately the same.