Personally, when I computed those numbers above for Los Angeles, I was thinking that 20,000-30,000 fatalities was really not that bad [of course - it's not great if you're one of them!]. It still means that 99% of the residents of L.A. would survive the event.

The fires spread rapidly due to high winds from a nearby typhoon off the coast of Noto Peninsula in Ishikawa Prefecture, and some developed into firestorms which swept across cities. As a result many people died when their feet got stuck in melting tarmac; however, the single greatest loss of life occurred when approximately 38,000 people packed into an open space at the Rikugun Honjo Hifukusho (Former Army Clothing Depot) in downtown Tokyo were incinerated by a firestorm-induced fire whirl. As the earthquake had caused water mains to break, putting out the fires took nearly two full days until late in the morning of September 3. The fires were the biggest causes of death.

These casualty and damage figures could be low; Mark Selden wrote in Japan Focus:
The figure of roughly 100,000 deaths, provided by Japanese and American authorities, both of whom may have had reasons of their own for minimizing the death toll, seems to me arguably low in light of population density, wind conditions, and survivors' accounts. With an average of 103,000 inhabitants per square mile (396 people per hectare) and peak levels as high as 135,000 per square mile (521 people per hectare), the highest density of any industrial city in the world, and with firefighting measures ludicrously inadequate to the task, 15.8 square miles (41 km2) of Tokyo were destroyed on a night when fierce winds whipped the flames and walls of fire blocked tens of thousands fleeing for their lives. An estimated 1.5 million people lived in the burned out areas.[7]