As we’ve seen last time, there’s sort of a Mexican standoff in space: Spacecraft are vulnerable to ground fire, but planetary defences can be killed fairly easily by orbital bombardment. This, by the way, fits with the “if you can see it, you can kill it” which seems to have been the military trend of the last few decades.
This implies that spacecraft can hang back (such that they can dodge incoming shots) and slowly search for any suspicious ground craft which they then bombard (using the fact that those railgun tanks cannot dodge as well as spacecraft). But this cannot kill any submarines or tanks lying low (a fuel cell or MHD turbine tank that’s idling will be almost invisible).
So - are the spacecraft screwed? I don’t think so.
The disadvantages of planetary defence
As we’ve seen above, Planetary Defense Railgun Platforms (which I’ll abbreviate as PDRPs, pronounced “P-Derps”) can be killed once seen, and can be seen once they are firing.
Accordingly, they can be presented with two bad alternatives by introducing drop pods. A drop pod is, essentially, a simple and small target which contains people or supplies and which attempts an unpowered descent. An example is the Bolide Drop Pod (SS4:16), which is a ten-ton pod (SM+4) carrying seven people and half a ton of supplies. Shifting the armour systems and using advanced metallic laminate) would grant it 10d/1d/2d+2 of armour.
Even an SM+5 PDRP’s main battery would do 45d, and is able to penetrate the pod. But that’s exactly what we want: If they fire at the pod (which may well be empty or contain munitions to seek out whoever destroyed it), we know their position and retaliating railgun fire is soon to follow. If they don’t, then a seven-people spec-ops team might be on planet and work on finding the PDRPs. This situation, in fact, might be compared to the “Scud Hunt” during the First Gulf War. As at that time, and despite this high-level reasoning we just heard, most targets might not actually be destroyed but - hopefully at least - deterred from operating.
Some pods might also just contain explosives (they have roughly the same explosives capacity as a MOAB or Daisy Cutter) to prepare the landing site by clearing it from mines and foliage.
Once these defences are suppressed over a sufficiently large area (or you are willing to accept losses), further operations can occur.
The creation of a spacehead (following traditional naming of the bridgehead, beachhead, and airhead) is probably the second part of any operation. Where beforehand, operations focused on the identification and suppression of PDRPs, it is now time to establish some (hopefully secure) position on the planet. The first wave, probably, is again drop pods except dropped in greater numbers and possibly larger pods. This risks higher casualties (a PDRP might kill multiple pods before being blown up) but can both overwhelm PDRP defences (which can be expected to get off at most 20 shots during landing) and land sufficient people on the planet to withstand first counter attacks.
The main goal of the people landing is to establish a few dozen kilometres of safe area in which supplies can be stockpiled, people can recover, and maintenance can be conducted. Oh, and new supplies can land. By pushing the parameter farther outwards, they can also produce areas free of the PDRP threat.
Once they have expanded to 500 kilometres, airspace is safe up to at least 20 kilometres. That’s with distance-to-horizon (i.e. shooting through the thickest part of the atmosphere); I’ll just take 500 kilometres as the distance needed to keep your shuttles safe while descending from atmosphere. That’s quite large - roughly the area of France, Texas, Paraguay (north-to-south; it’s a bit larger), Egypt, or about double the Koreas.
To do something like that will take people and time. Time to advance, at the rough estimate of current-day military speed, we’ll call 250km per day against light resistance. A circle with a radius of 500 kilometres is about 3000 kilometres of circumference, which would suggest about 75 divisions or about 1.5 million people (at WW2 numbers), advancing over two or three days; more without roads. Just to secure the spacehead!
A more reasonable assumption is reducing this. 2003’s invasion of Iraq saw 180,000 soldiers deployed during the initial phase, which attacked over an initial front of about 150 kilometres and reaching up to more than 600 kilometres. That’s only 300 people per kilometre, compared to 500 per kilometre from above. Of course, that’s during the advance; while defending, there’ll be fewer people deployed. Call it 100 per kilometre, for 15 divisions. That’s much more manageable (but still a large number).
In truth, we can probably reduce this somewhat: Intelligent choice of landing site will result in far less area to defend. Any island, for example, significantly reduces the area we need to secure, as does a peninsula. Profligate use of orbital bombardment might increase the effective frontage of military formations.
Additionally, dropping people and supplies not in the centre but further out will also decrease time to secure a spacehead. This might end up forming the difference between the setting’s US Army and Marines (not that you’d see them, since they’ll pretty much stay on Earth): The latter’s task is to establish and secure the spacehead, which the former then use for landing.
At the same time as this happens, the defenders try and mask people to crush the landing size. That’s going to be difficult - by definition, they have lost any orbital support, there probably are going to be multiple spaceheads established (it’s a tradeoff between individual strength and redundancy), and there should be even more decoy landings.
Now, you might have noticed that I never actually talked about landing in an urban or strongly defended area. That’s because it’s, arguably, similar to landing first parachutists, then helicopter-borne troops, then transport planes in (a) a modern city, or (b) the middle of an enemy force. As seen during Market Garden (for an older example; nobody appears to even contemplate doing these things against “real” defended locations); I’m assuming that will hold for space landings.
But what, you might argue, if you try to invade one of those whole-city planets (Trantor, Coruscant)? Well, there are two issues with that: One is that you’ll then essentially fight a whole war in a mega-city, so any casualties taken during landing are nothing compared to the casualties you’ll take during the combat itself. And second is that there will always be acceptable landing sites: Islands, deserts, non-settled areas.
In summary: Orbital surveillance and drop-inserted special forces hunt for enemy PDRPs. Once sufficiently attrited and suppressed (or once you decide to accept the casualties), you secure a landing site first with drop pods, followed by shuttles. After the first few days, you should have succeeded in securing at least one spacehead against light resistance (since your enemy will have issues bringing up their own forces), and combat then devolves to a “normal” ground campaign.