There are three known types: Doppler shifts ( due to motion through space away from the observer); gravitational redshifts ( due to light leaving a strong gravitational field); and cosmological expansion ( where space itself stretches as light travels through it). The way astronomers distinguish between the three depends on the kind of object they are studying. Here's a table of the different kinds of objects and the liklyhood that one of these three is present to create the observed spectral shift.
| Object | Doppler | Gravitational | Cosmological |
| Planets | x | x | |
| Stars | x | ||
| Nebulae | x | ||
| Neutron Stars | x | x | |
| White Dwarfs | x | x | |
| Nearby Galaxies | x | x | |
| Distant Galaxies | x | x | |
| Black Holes | x | x |
Gravitational red shifts are generally very small, and you only get very large ones from the light emitted near neutron stars or black holes...environments you can independently confirm from other observations. Cosmological redshifts are only important and easily distinguishable for rather distant galaxies, but can get mixed up with the Doppler shift from the regular spatial motions of galaxies. With the exception of the sun, no gravitational red shifts have been detected for ordinary stars, but they ought to be present if we had good enough instruments.
Mainly, to distinguish gravitational redshifts from other kinds, you compare the size of the object with its mass to determine how much larger it is than its black hole radius. Objects like nebulae and entire galaxies are trillions of times larger than their BH radius, so the magnitude of the redshift is 1 part in a trillion of the rest frequency. Normal stars are only a few hundred thousand times larger than their BH radius, so light from their surfaces is at the limit of being able to detect, spectroscopically, such a gravitational redshift. Neutron stars and white dwarfs are about 10, and 3000 times larger than their BH size so gravitational redshifts are of the order of 1 part in 10 to 1 part in 1000 of the rest wavelength.
Cosmological redshifts are only seen unambiguously at distances of 100s of megaparsecs. At nearer distances, ordinary Doppler shifts from galaxian motion with respect to a local center of mass ( galaxy cluster) is comparable to the cosmological effect and you have to disentangle the two contributions very carefully. Typical galaxy speeds in a cluster are 300 km/sec, and this equals the cosmological recession at a distance of only 5 megaparsecs or so!
https://einstein.stanford.edu/content/relativity/a11859.html
This doesn't make sense to me. The red shift occurs at the moment the light left the object - what is stretching it out after that time?
As the light is traveling through space, space itself is expanding, thereby stretching the lightwave. This phenomenon is independent of whether the object itself is physically moving towards us or not. Physical motion is reflected in standard Doppler shift. Further, if the light source has a strong gravitational field associated with it, there would be yet another (a 3rd) shifting of the light frequency.
As the piece I posted above notes, there are 3 (known) types of shifting: shifting due to actual motion of a light source, shifting due to a strong gravitational field, and shifting due to universal/cosmic expansion. Cosmic expansion and gravitational shifting is always redshifted, whereas standard Doppler involves an object physically moving EITHER towards OR away from us.
For example, light from a remote high mass/high gravity star which, say, happens to be physically moving towards us, would be shifted in all 3 ways.
There would be blueshifting due to the fact that it (the star) is physically (actually) moving towards us, And separate redshifting due to its strong gravitational field AND cosmic expansion. There actually are ways to sort out which effect is causing each shift.