(Update 12/14/2012) I asked Ryuichi Kino (who's now allowed back in TEPCO press conferences) about these maps. As far as he knows, these maps still ignores geographical information, and treat the land as flat. He says he will ask about it in the next press conference of the Nuclear Regulatory Authority/Agency.
The dispersion simulation maps first announced by the Nuclear Regulatory Authority right were full of errors and had to be redone. NRA announced the corrected maps on December 13, but according to the article by Tokyo Shinbun, the real story is not the maps but how the maps were created.
According to the Tokyo Shinbun article as quoted at "Zamamiyagarei" blog, the history is as follows:
Nuclear and Industrial Safety Agency under the Ministry of Economy, Trade and Industry, before it was shut down, decided to have the simulation maps created for each nuclear power plant in Japan. It contracted Japan Nuclear Energy Safety Organization (JNES) to do the job. [I've read elsewhere that the job was for mere 10 million yen, or US$119,000).]
JNES had one person in charge of the job, and he gave it to a long-term subcontractor called CSA Japan in Tokyo, as usual.
CSA Japan had one person doing the job of creating the simulation maps for the entire nuclear power plants in Japan, who used his pocket calculator to calculate the dispersion models. He apparently made numerous errors, resulting in the inaccurate simulation maps that NRA announced, to their embarrassment.
No one at JNES or NRA checked the work.
It's not that using a pocket calculator is bad. An Indian "computor" (a person who computes) calculated the height of Peak XV in Himalaya by hand calculation. It is all in the brain, after all.
CSA Japan was incorporated in 1999 with the capital of 10 million yen. Its main customers are government agencies and large corporations in nuclear industry. JNES comes near the top of the customer list. TEPCO also seems to be their regular customer:
Japan Atomic Energy Agency
Japan Nuclear Energy Safety Organization
Tokyo Electric Power Company
The Japan Atomic Power Company
TEPCO Systems Corporation
Institute of Nuclear Safety System ,Incorporated
Genden Information System Co.
Mitsubishi Heavy Industries, Ltd.
Nuclear Fuel Industries ,Ltd.
Mitsui Engineering & Shipbuilding Co., Ltd.
MHI Nuclear Engineering Company, Limited
Itochu Techno-Solutions Corporation
TOiNX (Tohoku Information Systems Company, Incorporated)
Why did NISA need to create simulation maps by contracting them out cheaply, which ended up on the lap of one unhappy employee at a subcontractor who used a pocket calculator? They have SPEEDI.
My guess is that NISA figured it was cheaper to contract out anew than to run SPEEDI.
"Cheap, cheap", sang a NISA bird. So sings just about every government agencies in Japan, large and small. Money before safety. That's what two "lost decades" have done to the country.
By the way, the first set of maps not only got the wind directions wrong, but they didn't consider the topography at all, treating the land in Japan as flat. I don't know if that has been corrected in the new version. With a pocket calculator, it may have been a bit too complicated.