I'm not using the ideal estimate of the integral length since I'm doing
the integral with the 1D spectrum, but I think this should be close
enough.