Near my friend's home, there was a gully that apparently amplified the wind and funneled the fire up into the housing tract.

Fire behaviour "usually" follows predictable rules. In wooded vegetation of the sort on SE Oz (not dissimilar to California, I'm told) it's the preheating effect of radiated heat that can be most influential. On a windless day in a perfectly flat area, radiation is probably the same around the whole periphery of a fire. Any wind will cause the flames to lean downwind, increasing the radiant heat load of any fuel downwind. This means the main flame front will "run before the wind" in perfectly flat terrain. In Oz vegetation, the rule of thumb in windless conditions is that the speed of flame front movement will double for every 10 degrees of slope, as the fuel upslope is preheated more rapidly as the angle between the flames and the ground decreases. A narrow gully intensifies this as the bottom of the gully will usually have a decent slope to it and its sides will restrict the radiant heat to the immediately adjacent fuel even in windless conditions, Any wind will usually funnel up the gully, intensifying the whole effect.

If this wasn't bad enough, a fire racing upslope can occasionally create a sudden vortex that is horizontal and fully flame laden for up to 100m (300') across the slope at right angles to its main path. As far as I know, there has (yet) been no way that this behaviour can be predicted. Not a good place to be caught out in the open.

The very worst place to build a house is at the head of gully that faces the prevailing winds during a fire season, but it's often the location that gives the best views.

Sigh!

The fuel most vulnerable to becoming preheated by radiation in bushfires is vegetation with a diameter less than 6mm or 1/4"; grass, twigs, bark all carry very fast-moving fires. Some species may be fire retardant because of their chemical makeup but others accelarate flame rates for the same reason. Minimising the quantity and arrangement of fine fuels downslope and/or upwind of one's house for a minimum of 30 metres (100') is a necessary (but sometimes insufficient) precaution; fires that are crowning (moving through the tree tops in front of flames at ground level) will have a high radiation output at least four times further away ("horizontally)") than the trees' height, increasing the clearance distance even further.

All the best!

Cheers, Rowan