Weather Modelling of Urban Heat Island in Megacities Using Distributed Urban Parameters Derived from Global Datasets

Ⅰ. Introduction

The urban heat island (UHI) phenomenon is characterized by warmer air temperatures in urban areas compared to the surrounding countryside. According to the United States Environmental Protection Agency, the direct impacts of UHI includes increased energy consumption, elevated emissions of air pollutants and greenhouse gases, compromised human health and comfort, and impaired water quality. Modifications to climate such as local thunderstorms are also linked with the occurrence of UHI¹. Weather models such as Weather Research and Forecasting model coupled with single-layer Urban Canopy model (WRF-UCM)²are powerful tools to investigate UHI. The model requires urban parameters as inputs.

Traditionally, urban parameters are uniformly assumed in the WRF-UCM but this leads to unrealistic urban representation. Distributed urban parameters calculated from real building distribution can also be incorporated into the model to consider urban effects together with improvement in their parameterization^3,4. The problem is that distributed building information is not readily available for most megacities such as those belonging to developing countries. Furthermore, detailed study of UHI in developing megacities still remains a challenge due to the lack in reliable meteorological observation data. In this study, we address the issue of urban representation by using a simplified method derived from globally available data.

Ⅱ. Methodology

1. Estimation of urban parameters using a global dataset

Necessary urban parameters⁵ consist of average building height (H_ave), frontal area index (λ_f) plain area index (λ_p), roughness length for momentum (z₀) and displacement height (d). H_ave, λ_f and λ_p are building morphological parameters. z₀ and d are classified as aerodynamic parameters influencing the wind distribution. Definition of λ_f and λ_p are the ratio of the total frontal area of buildings to the total surface area of the grid considered and the total area occupied by buildings to the total surface area of the grid considered. z₀ is defined as the height above the displacement plane at which mean wind becomes zero when extrapolating logarithmic wind-speed profile downwards. d is the height above the ground at which zero wind speed is achieved under neutral conditions. Empirical equation of H_ave can calculate standard deviation of building height (σ_H) and maximum building height (H_max). These parameters can be calculated from the correlation between σ_H, H_max and H_ave⁴.

To be able to estimate urban parameters in any city around the world, a relationship was examined between global datasets and each of the building morphological parameters. Global datasets utilized are high-resolution population dataset (LandScan by Oak Ridge National Laboratory), nighttime lights (NOAA), and MODIS-based maximum green vegetation fraction (NASA). Realistic urban morphological parameters were calculated from real 3D building database or other urban parameter datasets (NUDAPT) of Tokyo, New York city, Melbourne, Istanbul, Jakarta and so on. Regression equations were then derived from the block-averaging of spatial pairs of building morphological parameters and global datasets to estimate the morphological parameters (see discussion). Finally, the distributed aerodynamic parameters are calculated by using the new aerodynamic parameterization by Kanda et al⁴.

2. Model improving and simulation setting

In this study, the Advance Weather Research and Forecasting Model-ARW version 3.3.1 (WRF) modified³ to incorporate the distribution of roughness parameters mentioned in the previous sub-section. WRF⁶is a fully compressible, nonhydrostatic and limited model with terrain-following mass coordinate system.

The National Center for Environment Prediction (NCEP) final (FNL) global analysis datasets was used for lateral and boundary conditions. Simulation was conducted on April and May in 2012. Five megacities (Dhaka, Jakarta, Karachi, Mumbai and New Delhi) were focused in the finest domain with grid spacing of 2 kilometers (see Fig. 1). Selected megacities belong to different climate zones defined from the Koppen climate classification. Distribution of anthropogenic heat emission⁷was also released at the first atmosphere level of the model. To investigate UHI, 2 simulations were conducted; one is with the distributed urban parameters, another is where urban cover was replaced with grass.

 Ⅲ. Discussion

Results showed that H_ave can be a function of population adjusted by nightlights data (see Fig. 2). Regression curves of H_ave between Tokyo and New York increased linearly with population adjusted by nightlights, however Istanbul and Jakarta showed lower H_ave values and varying less with increasing population. Melbourne was an exception. In spite of low population, Melbourne has high building heights but similarly showed an almost steep linear relationship. λ_f on the other hand shows large dependency with λ_p. Furthermore, it behaved similarly with H_ave. Dependence to population adjusted the nightlight was not found for λ_pbut showed inverse relationship with vegetation fraction in all cities, as expected for urban areas (see Fig. 3).

Finally, the simplified but precise approximation of urban parameters through readily-available, high-resolution global datasets and our derived regressions was utilized to estimate a global distribution of urban parameters which was later used in the WRF model. Furthermore, simulation showed that each megacity had different UHI intensity. UHI difference is related to each city’s local and regional condition, which includes distance from the sea and representative, climate classification. In the future, we plan to explain the investigation to all megacities around the world numerically (Global Urban Climatology).

Ⅳ. Acknowledgment

This research was supported by the Environment Research and Technology Development Fund (S-14) of the Ministry of the Environment, Japan.

Ⅴ. References

1. Li, D., Bou-Zeid, E., Baeck, M. L., Jessup, S., & Smith, J. A., Journal of Hydrometeorology, 14(4), (2013) 1098-1118.
2. Chen, F., Kusaka, H et al, Int. J. Climatol., 31: 273-288, 2011.
3. Varquez, A. C. G., Nakayoshi, M. and Kanda, M., Boundary-Layer Meteorology. Springer, 154(3), pp. 449–469. doi: 10.1007/s10546-014-9985-4, 2015.
4. Kanda, M., Inagaki, A., Miyamoto, H., Gryschka, M. and Raasch, S. Boundary-Layer Meteorol, 148(2), 357-377, 2013.
5. Grimmond, C. S. B., and Timothy R. Oke. Journal of applied meteorology 38.9 (1999): 1262-1292.
6. Skamarock, W. C., et al. National Center for Atmospheric Research, Boulder, Colorado, USA (2008).
7. Dong, Y., Varquez, A., Kanda, M. Atmospheric Environment,  (submitted)