The main reason is, it's a lot of effort and complexity for very little gain.
The minimum velocity required to keep a spacecraft in low earth orbit is about 8km/s (28800km/h). Your 700km/h is less than 2.5% of that. It is easier just to add a bit more propellant to the launcher. In practice delta-V nearer 10km/s is needed to get to orbit due to air resistance, and to fight gravity at the lower speeds. Rockets actually have to throttle down shortly after liftoff due to the effect of air resistance (the stage of launch known as max q.)
Building a 5km vertical railgun is a major project, and adds another thing that can go wrong at launch. Additionally, the frequency of launches is quite low and doesn't warrant the cost of such hardware. According to http://spaceflight101.com/2017-space-launch-statistics/ there were 18 Falcon 9 launches and 15 Soyuz launches in 2017 (there were also 16 Long March launches, but this is a far more diverse family.) So typically a single rocket model only very rarely has more than two launches a month. This is compounded by the fact that the most frequently used rocket models are launched from more than one site, so 10 launches a year for a given model and site is a high number.
An alternative to the railgun approach is launch from under the wing of an aircraft. This achieves close to 700km/h while lifting the rocket through the thickest part of the atmosphere. It also has the advantage that the aircraft can be flown to any desired launch site. Several systems are being developed, one operational https://en.wikipedia.org/wiki/Air_launch_to_orbit
That said, Rocket sleds are quite similar to what you propose. They have been tried (but never put into operation.) https://en.wikipedia.org/wiki/Rocket_sled_launch and https://en.wikipedia.org/wiki/Rocket_sled indicate that ground speeds of over 10000km/h have been achieved at research scale, and about half this speed for 60000kg masses. Rather than dig down, the sled track can be built up a mountain, to get the rocket into thinner air.
I would caution that there is a misleading statement in one of these rocket sled articles that the space shuttle used 1/3 of its propellant to reach 1000mph - the solid rocket boosters were by design far less mass efficient (had a lower specific impulse) than the liquid propellant system, and were jettisoned early in flight, so this figure is heavily skewed by the mixture of propellant systems on the shuttle.