Multi scenario urban climate projection for Tokyo and Osaka, Japan: An application of dynamical downscaling and urban planning

Tokyo and Osaka are the two largest cities in Japan. With increasing heat-island effect and global warming, along with rapid aging of the population, heat-induced health impact is becoming a major social problem in these cities.

In this study, heat-related health impact indices are projected for the next 80 years using dynamical downscaling (DDS) approach, whereby outputs of coarse-resolution global climate model (GCM) are used as initial and boundary condition for a high-resolution regional climate model. DDS is an effective technique to obtain fine-scale climate information. With urban canopy dynamics implemented, regional climate models allows for proper assessment of urban effect. Projection is made for multiple future decades (namely 2030's, 2050's, 2070's, and 2090's) with multiple GCM's (MIROC, CSIRO, and MRI) to account for projection uncertainties arising from each of them.

Results show that by 2090's, temperature is projected to increase by ~2.5 degrees for Tokyo and Osaka. Yet there exists a considerable uncertainty and sub-decadal variation in the projection. Uncertainty ranges (by different GCM) grow from ~0.5 degrees in 2030's to ~1.5 degrees in 2090's, and are larger for Osaka than for Tokyo. The number of extremely-hot days (days with maximum daily temperature exceeding 35 degree Celsius) is projected to double for Osaka, and increase by four-times more for Tokyo. Based on WBGT, which has been adopted as an official heat-stroke risk index by the Japanese authorities, number of danger-level days (daily maximum WBGT exceeding 31 degree Celsius) are also calculated from the projection. This figure is projected to increase from almost none in the 2000's to more than half of the days in August for Tokyo, and to more than 20 days for Osaka.

In addition, impact of metropolitan-scale urban planning scenarios is evaluated in terms of the potential for urban heat mitigation. We assume three future urban scenarios for Tokyo; status-quo (urban structure remain the same), spread city (population and built-up area spread to suburban areas), and compact city (population and high-rise buildings concentrated in central Tokyo). Additional DDS experiments are conducted with each of the urban scenarios, this time with single GCM (namely MIROC5) and for a single decade (2050's).

Evaluation based on WBGT revealed that urban scenario have relatively small impact (~0.2 degree Celsius) compared to the global warming signal (~2 degree Celsius). However, urban scenarios have statistically significant impact on the local climate. Compared to status-quo scenario, compact city scenario increases WBGT in central Tokyo and reduces in suburban area. Opposite is shown for compact city scenario, with a WBGT reduction in central Tokyo and an increase in suburban areas. However there are some parts in suburban area that showed a decrease in WBGT under compact scenario. This decrease is associated with a substantial reduction in wet-bulb temperature. This result indicates that urbanization generally increases WBGT but may cause reduction depending on moisture response.