I'm a former active-duty Marine Corps officer and I can answer these for you. My answers only apply to the jets in the stack, not the helicopters.
As general principles: (1) Jets, except for the A-10, are average at best, at CAS, (too few, too fast, too high.) (2) Turboprops are much better at CAS (more, slower, lower.) (3) Nothing will ever be better at CAS than the A1-D Skyraider.
What targets are most often engaged? (1) The ones that were too stupid to (or could not) run away (2) who were in close proximity to something that we didn't want blown up with a JDAM. I'm dead serious. Generally speaking, the cue (wait) between asking for air support and getting jet supplied support was a REALLY LONG time. I'm talking STUPID LONG. CAS aircraft (the jets) in Iraq were controlled by two different outfits: the Air Support Operations Center (ASOC) in the army's area of operation and the Direct Air Support Center (DASC) in the Marine Corps area of operations. The airforce and navy operated 'stacks' of jets in the same kill boxes, so before anything happened, your request had to be “de-conflicted” by the Joint Terminal Air Controllers (JTAC) and any target with an IQ over 50 would always run away about 20-minutes into the "de-conflicting process" if they could. So about thirty minutes after your request: (1) a recon ship would fly over to make sure the target wasn't a mosque, (2) and a collateral damage estimate would be made, (3) and then they would locate you (twice), (4) and then all the other friendly units close to you would be located, (5) and then the friendly locations close to you would be located, (6) and then the target was located again to double prevent a friendly fire or collateral damage event (6) and then sometimes an airforce FAC would have to show up and sign off on everything again. So 45 minutes to 1-hour later, you finally heard your guy circling overhead double checking everything again and then you got a call asking, "Where do you want it?"
What kind of ammunition is used? Who cares. It made lots of sparks. It's normally 20mm.
What is the effective range and accuracy? They would pull out from a gun run at about 1000 to 1,500 feet--usually. Gun runs are VERY accurate. The gun sprays bullets in a big circle that gets smaller--there are lots of sparks--and foot thick solid concrete walls are literally vaporized.
Is strafing done from level flight at low altitude or from a dive? Always in a dive. You always want plunging fire to prevent hitting things downrange. I once saw a Cobra shoot a couple of rockets that went way too high (and down range for about a half-mile) and they blew up a house. (Those guys got in trouble.)
At night or during daylight only? Both. Night runs are WAY better because you can illuminate the target with IR beams. It's like shining a Q-beam at house only it's totally invisible. You need night vision to see it. It makes EVERYTHING so much faster and easier to do--everybody paints it and then they shoot it. It takes the error out--they shoot that spot.
How are targets identified? (1) When you got a bunch of guys that are shooting at you, (2) and they can't run away (because you got them trapped,) (3) but it would be a giant pain in the ass to go in there, (4) and you can't drop 500-pound JDAM because something across the street might get damaged--you shoot those guys with 20mm.
Are multiple passes needed to ID target before firing? Hell no. By the time the pilot pulls the trigger, the target has been checked and double checked about 10-times by about 5-layers of command structure.
What is the success rate? Stuff fell down every time. But only total morons ever got killed because it took too long for anything to happen. We never did anything where there was even the remotest chance of REALLY, I MEAN REALLY, needing air support, because we knew to a moral certainty that it would take too long for anyone to show up. It ain't like the movies.
Are most strafing runs only usefull in scaring the target into running? If you had them trapped, they were DEAD. If you didn't, they were gone 30-minutes to an hour before anything happened. And the smart ones always knew what was happening because WE HAD TO PULL BACK a couple hundred yards "for saftey" and we couldn't move out of our safety position during the process. It ain't like the movies.
How often are planes lost to ground fire, MPADS, ground impact? Jets NEVER show up with MPADS around. Are you kidding me. That's hilarious.
The F-35 WILL NEVER do CAS because they are going to cost $140 million each and nobody will want to put that kind of stress and time on the airframes. The Marine Corps needs the F-35 like it needs a division of war elephants. The best idea for the Marine Corps would be to snag all the A-10's that will have to be retired because of the F-35 purchase and then we should develop something that will do CAS even better.
In my opinion, the F-35 is a total FUBAR disaster. The wild weasel mission will be done with UCAVs in the future. Recon will be a UCAV mission in the future. CAS should be a turboprop or a next-gen A-10 mission in the future. Deep strike should be a next-gen bomber mission in the future. Building the F-35 is like building the P-47 in 1955, by the time it ever becomes operational, it will be long since obsolete.
For what it's worth...
Bronc