Resolution is not the most important parameter to look at . It all depends on how your crystal is ordered , where the modification is and how many modifications are there. 

Assuming your sample is 100% modified you should be able to see the modification clearly as low as 3-4 A resolution, in case of phosphorylation maybe at even lower resolution. 5 or 6 maybe? That would depend on the data quality, too. With very low occupancy you might have to know the site in order to find it, even at atomic resolution.

Other factors besides low occupancy that could work against you are high B-factors of the modified region. If your modification is in a highly disordered region you won't see it at all, regardless of resolution.

I have solved and published a phosphoryalated protein at 1.9 A resulution.

To distinguish the blanched O atoms in crystal structure, I think at least 2.5 A resolution is necessary.

But at the same time I think it is very difficult to determine the modification only from the crystal structure. Actually I had to verify the site direct phosphorylation with mass mass spectrometry to publish the paper. 

If you already have crystal, I suggest you to check the modification with another way. This will strongly will help you to make reviewer believe the presence of modification even the resolution is not so high.

Also depends on the quality of the detector, quality of your x-rays, using the latest software to do the data reduction and scaling etc. 

As Stefan said, the local order is also very important. So it really depends on how nice your density looks. And mass spec is cheap, does it not show up in a mass spec of the protein? SDS-PAGE, band excision, stick it in an eppendorf and ship it off to the mass spec people. 

Thank you.