Bias correction method with skewed distribution for projection of cardiovascular diseases mortality rate based on extreme temperature

• Main Author: Mohd Esa, Aina Izzati
• Format: Thesis
• Language: English; English
• Published: 2022
• Subjects: Climatic changes; Cartography
• Online Access: http://psasir.upm.edu.my/id/eprint/112165/1/FS%202022%2054%20-%20IR.pdf; http://psasir.upm.edu.my/id/eprint/112165/
• Institution: Universiti Putra Malaysia

Bias correction method is useful in reducing the statistically downscaled biases of global climate models’ outputs and preserving statistical moments of the hydrological series. However, bias correction method is less efficient under changed future conditions due to the stationary assumption and perform poorly for removing bias at extremes thereby causing unreliable bias-corrected data. Thus, the existing bias correction method with normal distribution needs to be improved by incorporating skewed distributions into the model with linear covariate to account for non-stationarity. This study develops bias correction method with skewed distribution using quantile mapping technique to reduce biases in the extreme temperatures data of peninsular Malaysia. The network input is the MIROC5 model output gridded data for the period 1976-2005, and the model target used for bias correcting the input data is the observed extreme temperatures sourced by the Malaysian Department of Irrigation and Drainage for the same period. Results indicate that the proposed model obtains more accurate estimates of future mortality rates based on model diagnostics and precision analysis. Bias correction method with skewed distribution is used for bias correction of MIROC5 modeled projected extreme temperatures for 2006-2100 corresponding to the representative concentration pathways emission scenarios and it can correct the biases of future data, assuming skewed distribution of future extreme temperatures data for emission scenarios. Lognormal and Gumbel with linear covariate are the most appropriate distributions to model the annual extreme temperatures. Simulation study was conducted to validate the results. It was found that Gumbel with covariate is the best fitted distribution for extreme temperature series than other distributions. Higher projection of extreme temperatures is more pronounced under RCP8.5 with precise estimates ranging between 33- 42◦C compared with that under RCP4.5 with precise estimates ranging 30-32◦C. Finally, the projection of extreme temperatures is used to calculate the mortality rate of cardiovascular diseases across all regions in peninsular Malaysia which coincide with high extreme temperatures ranging between 0.002 to 0.014.