One of Jay's most recent b-v diagrams :

Here are a few spectra from the GBT. This is a deep integration on our field C4, which is at l=219, b=0. The total integration time is about 2.5 hours. The data are frequency switched at 4 MHz, with a BW of 10 MHz and 1024 channels per IF. Here we see two linear polarizations, but just one half of the freq switch cycle. At the top is the raw data, on the bottom is the data after some flagging and baseline removal.

What we were searching for is the faint wing on the red (low freq)
side of the line. It is pretty repeatable, and widespread over the area
which we mapped.

The message here is : The GBT can make precise, repeatable measurements
of Galactic HI at the 10 mK level. Maybe better, the noise is still integrating
down as 1/sqrt(time).

Our proposal is here
!

Some very nice spectra of the region of 4C73... (center plus offsets) from Jay:

And here's a model for different scale-lengths using an exponential drop in HI density :

It looks like it's pretty hard to have a scale length longer than about 0.75 Rsun. This model includes a linear flare outside the solar circle. There is no widening of the velocity width due to random motions or HV gas. These effects will only reduce the scale length.