There are advantages to using micro-inverters such as Enphase, like higher efficiency, easier installation, expansion one panel at a time.  BUT these come with the big disadvantage of total non-function during grid down situations.  I believe it should be possible to safely "trick" an enphase system into providing power during a blackout. 
Has anyone done it?
I believe one way this can be done is with a Sunny Island. 
I think there is also a way to do it with a Xantrex XW6048 and another small pure sine inverter, 2 contactors, a small solar panel and a few relays.  Certainly the cheapest blackout backup power would come from a standby generator.  But if one invests in a battery backup system , like the XW6048 and solar panels. must one take the lower performance of processing the solar output into the batteries then 'selling' exess  battery charge to the power company or can micro inverters be somehow used both in grid 'up' and 'down' situations?

See http://www.wind-sun.com/ForumVB/showthread.php?14008-How-can-I-disconnect-from-utility-in-event-of-power-grid-melt-down
and related discussions on same forum.

I've recently learned that the Schnieder electric (used to be Xantrex) XW series of inverters now support "AC coupled" operation. In Ac coupled operation another grid tied system is connected to the OUTPUT of the XW. When the grid is "up" the transfer switch in the XW allows power flow in both directions and the grid tied inverter sees the grid and runs normally. However when the grid is "down" the XW produces grid equivalent AC from the battery. The UL spec for grid tied equipment has tight limits on the grid and a 5 minute turn on delay so after 5 minutes the grid tied inverter will connect and the XW will become a battery charger using power supplied by the grid tied inverter that's not needed by the loads also connected to the output of the XW. Once the XW determines the battery is full the grid tied inverter(s) will be shut down by moving the output frequency above the limit set by UL. The grid tied inverter can be an Enphase system. This operation is very similar to the Sunny Island. Also, Schnieder now has a solar charge controller that can use a 600 volt string and produces 80 amps of charging current to a 48 volt nominal battery. The 600 volt input is just like a lot of grid tied inverters and simplifys installation.

Why not just use a small backup gen to trick the inverters? I have been looking at enphase systems and considered the gen approach since whatever panels I get will not be big enough to do the entire house.

Everyone that has a grid-tie PV system would like it to operate when the grid is down. It is not as easy as connecting up a small generator to fool the PV inverter. The PV system is a current source, not a voltage source, and it needs a big environment to dump the current into. For further discussion, see http://www.wind-sun.com/ForumVB/showthread.php?14008-How-can-I-disconnect-from-utility-in-event-of-power-grid-melt-down

There is a SMA Sunny Island system that can provide some power for emergency loads as described at http://ironedison.com/sma-sunny-island-backup-system. There are significant extra costs involved in providing this capability.

I read the link and don't think it makes any sense. Solar panels and inverters don't need to dump current anywhere, both are happy to simply not draw power from the solar panels.

Most small generators aren't a very steady frequency or voltage source, so that wouldn't be my first choice to simulate the grid being active. A small inverter would be stable, at least until you start applying loads to it. Maybe that's what the issue is; the grid-tie inverter may specifically check for a "grid" capable of fully powering the load. A large inverter running from batteries could supply that and I'd guess that it's for very brief periods (ie, wouldn't actually put much load on the batteries as long as the grid-tie inverters put out the slightly higher voltage that they are designed to).

If a grid-tie inverter is capable of detecting that a large sine-wave inverter is producing power instead of the utility grid, I'd be interested in knowing how.

I don't worry too much about this. "Grid Down" situations are usually in the form of a temporary spike for us. Five minutes later, the wind turbine is running again, and the solar PVs are putting out juice.

I did a little more research and I see that some grid-tie inverters aren't designed to work with reactive loads - they rely on the grid to absorb this power. But a large battery inverter running in parallel with a grid-tie inverter should provide this. Would the battery inverter be unhappy about any of this and shut down - I don't know.

you may want to look into the SMA TL-US. per the manfacturer, it'll supply up to 1500 watts of power should the grid go down.

http://www.sma-america.com/en_US/products/grid-tied-inverters/sunny-boy/sunny-boy-3000tl-us-4000tl-us-5000tl-us.html

Enphase actually have a tech doc on this subject, found at:

href="http://enphase.com/global/files/Enphase_Application-Note_AC-Coupled-Battery-Based-Systems.pdf

It offers a good explanation of the problems of syncing their micro inverters with an external non-grid generated ac waveform and offers several solutions.