I have found systematically negative in California. Below is an example histogram of the calculated during a monte carlo simulation for a single core. These values are obviously unphysical.

Below is a screenshot of the HI spectrum at (ra, dec) ~ (4:34:00, 36:30:00) with the Planck Av 4 and 8 mag contours. The median velocity range from the monte carlo simulation is ~ -8 to 4 km/s.

To double check the column densities, N(H2) is given by

For the spectrum below, we will integrate from -10 to 5, i.e. across 15 km/s with an average of . So . A typical DGR = mag. = 3.7 mag at this pixel.

This seems quite low. The high DGR value is the culprit.

To calculate and I first calculate using equation (1), compute the surface densities individually by

is then the sum of and ._

Likely the source of this problem is the residuals during the masking procedure are not trivial for California. See below. I’m interpreting the large negative residuals to mean that there is excess HI which is not traced by any dust.