Just using electric locomotives, you're looking at cutting fuel costs to about a third of their previous (methodology: NTD says ~10.5 million vehicle revenue [miles] per year, 2.3 gallons per mile (Amtrak's average) and $3.25 per gallon is ~$78.5 million. A BR Class 90 hauling 500t [six Guardian cars is 450 short tons] gets 22.62 kWh/mile, LA Metro pays 12 cents per kWh, comes out to about $28.5 million).Now that may be somewhat optimistic, that figure is for hauling freight rather than commuter travel. An IC225, which is an electric Class 91 locomotive hauling a cab car and coaches with a total of 554 seats (and more room to them than Metrolink's commuter coaches have), runs .038 kWh per seat-kilometer in a simulated run (pages 22-23). That would bump things up to 33.8 kWh per train-mile, though with a longer and faster consist than Metrolink uses, and raise energy costs to the realm of ~42.5 million dollars, still a significant decrease. The use of smaller consists or multiple units, especially on routes that are not terribly well patronized, would drop it back down again however.
A 45-66% decrease in fuel costs is nothing to sneeze at of course, especially since Metrolink is raising fares to cover a $13 million funding gap, four million of which is due to rising fuel costs. However, it could actually be reduced down even further, potentially even to no effective energy costs at all.
Electrifying all 388 route-miles of Metrolink would be a fairly expensive endeavor, costing about two to three billion dollars or more if there needs to be significant engineering work in order to build it such as widening the I-10 median and tunneling through San Clemente and San Juan Capistrano. It would also need tremendous political capital, not simply to build it, but also to swing it past BNSF and Union Pacific, who own much of the track in question. Solar, despite its general political acceptability, is unfortunately far too expensive. On a net metering basis, even with low end electricity consumption figures, the cost of installing sufficient solar capacity runs to about $900 million dollars, representing a major increase in the budget.
Nuclear, on the other hand, is rather more affordable, the reason lying in the far greater capacity factor of nuclear compared to solar. Given the timeline of an electrification project, a small modular nuclear reactor should be available and a 30MWe capability would suffice to power the system at current levels of usage, again on a net metered basis. The costs are also rather more reasonable, perhaps $200 million for a 40MWe plant capable of dealing with future growth and generating excess energy in the meanwhile to offset its operational costs (which would amount to just under $4.8 million per year at a 90% capacity with non-fuel O&M; with fuel it would be $7 million). Even without additional energy sales to off-set O&M costs, it would pay for itself in only a few short years and over a reactor's life would result in a substantial cost-savings to the program, to the tune of $445 million with a thirty year life-span or rather more with a longer life.
Unfortunately, despite what is actually a rather good safety record, many environmentalist groups have set themselves firmly against the use of nuclear power; though one might imagine that having sufficient political capital to electrify in the first place would allow them to use a small, safe nuclear reactor, odds are that it would lead to an enormous backlash. In its place, natural gas is most likely (though geothermal can't be ruled out depending on local suitability).
Without carbon sequestration, an advanced natural gas combined cycle plant would cost a thousand dollars per nameplate capacity kilowatt. If operated at an 85% capacity factor and built for the same handling of 33% more future power demand, a 42MWe plant would be required, adding 42 million dollars to the budget, truly a pittance (note, however, that in California, such plants operate at only 50% capacity factor; it might be better therefore to build with the intention of providing sufficient capacity for peak power and run the entire system off such a plant rather than run it with a net metering contract). The EIA quotes a price of $2.185 per million Btu (MMBtu) for the week ended June 13th on NYMEX, and the previous link about Californian plants quotes a heat rate of 7,176 Btu per kWh. Should such a 42MWe plant be operated at 85% capacity factor over the course of a year, fuel costs would amount to $4.9 million dollars plus an additional amount for maintenance (the first link in this paragraph would suggest $1.5 million annually; similarly, if we use its numbers for nuclear, a 40MWe plant would cost $213.4 million to build and cost $4.2 million per year to operate). However, natural gas prices are currently rather low due to a glut, and should climb quite appreciably in the future (as this page shows, it's dropped tremendously from a high of $15.4 per MMBtu in December of 2005).
So, what does all this amount to? Well, at the end of the day, electrification, in and of itself, does do a good job of reducing the fuel costs which are currently squeezing commuter agencies. The main benefit, of course, is increased acceleration and reduced trip times, but over a thirty year period of time, it does look as though it could pay for itself, if one factors in additional revenue and a higher rate of increase in the price of oil than the rate of increase in the price of electricity. Without such factorings, it still recovers much, if not all of the cost. However, if a rail agency is permitted to construct, own, and operate their own power plant to defray the costs of power, especially if they are allowed to sell excess power at market rates, electrification should pay for for itself without any additional factors taken into account.
Update: Much of this is based on a mistaken assumption. The vehicle miles actually appear to refer to car miles rather than train-miles as indicated by the train-miles in the 2012-2013 Metrolink budget.