Please use this identifier to cite or link to this item:
|Title:||Implementation of a regional climate model to improve downscaling and modeling of precipitation over western maritime continent||Authors:||Singh, Saurabh Kumar||Keywords:||DRNTU::Engineering::Civil engineering||Issue Date:||2018||Source:||Singh, S. K. (2018). Implementation of a regional climate model to improve downscaling and modeling of precipitation over western maritime continent. Doctoral thesis, Nanyang Technological University, Singapore.||Abstract:||The Western Maritime Continent (WMC) is recognized to have major weather and climate importance on both global and local scales. However, because of excess convection and complex topography, the region faces severe climate and weather modeling challenges. Further, the region has scarce station data, and Global Climate Models (GCMs) such as the Hadley Centre Global Atmospheric Model (HadGAM1) have been systematically found to incorrectly model precipitation. Thus, it is essential that WMC is properly represented by Regional Climate Models (RCMs) to capture and reproduce regional precipitation correctly. The discrepancy in GCM to accurately modeling rainfall over the WMC region results in errors that propagate to both the tropics and extra-tropics precipitation output as well. The RCMs have been observed to provide "added information" over the GCM data as they consider local topography and climate. The "added information" by RCM was used in three studies conducted in this thesis. In the first study, an integrated downscaling approach, i.e., combination of statistical and dynamical downscaling was applied over the WMC. RCM downscaled predictors having better spatial resolution capture the local information more appropriately and were used as input to a statistical downscaling model. The RCM used was the Weather Research and Forecasting (WRF) model. Integrated downscaling approach would thus fill data gap over data sparse regions of the WMC with a good quality downscaled dataset, thus, enabling a better analysis of precipitation for impact analysis. The most noticeable improvements are for Singapore location where using WRF predictors resulted in an RMSE value of 1.37 (0.81) for mean monthly rainfall for validation (calibration) period while R2 predictors had a much larger RMSE of 3.41 (3.86). In the second study, one of the outputs of CMIP5 GCM model output data, i.e., CESM1 bias corrected using ECMWF by NCAR was provided to WRF to downscale precipitation over the year 2030 to 2060 for RCP 4.5 and RCP 8.5 scenarios. Furthermore, the WRF precipitation was extracted and was used to generate future IDF curves over three locations in the WMC. In the final study, the Wards method was used to find clusters across 43 locations in WMC. The spatial-temporal groups formed were explained using CHIRPS (for historical) and WRF CMIP5 (for future). This is among the few studies done over WMC that utilizes outputs from two different RCMs to model future changes in extreme rainfall intensity. The output from the study would be significant in planning drainages and accessing future flooding considering the climate change. A region-specific parametrization scheme validated with observation data to capture convective precipitation was found for WRF and used for the three different studies. An integrated approach capable of producing better mean and extreme rainfall statistics than via conventional statistical downscaling from GCM was verified in the first study over all the four locations. In the second study, bias-corrected IDF curves were constructed for the two RCPs 4.5 and 8.5 scenarios. Significant increments in precipitation were observed over the three locations in WMC. The WRF-IDF was further compared to CORDEX-IDF, and it was noted that WRF predicted a greater increment in Intensity as compared to CORDEX. Both WRF and CORDEX outputs are considered as different realizations of the two-member ensemble. Hence these two realizations can be used to access the probable uncertainty in future IDF curves. The predicted increment in extreme rainfall intensity by RCP 8.5 was larger compared to RCP 4.5 and historical for almost all the locations and return periods. For Singapore (100 year RP and RCP 8.5) average intensity increase was 36.1% for WRF-IDF at RCP 8.5 while it was 16.8% for CORDEX-IDF. Over Jakarta (100 year RP) the RCP 8.5 WRF-IDF showed 52.5% average increment while the WRF-IDF showed 74% increase in average intensity. Lastly, the cluster analysis revealed some interesting new grouping in both spatial and temporal dimensions from the third study. This is the first study over WMC that examined the changes in spatial and temporal monthly precipitation using future RCP downscaled CMIP5 GCM data. It was observed that Spatial Groups (SG) formed for WRF-CMIP5 (RCP 4.5 and RCP 8.5) were very different from historical. The difference between SGs would signify changes in precipitation over the next 25 years. While comparing the SGs for RCP 4.5 and RCP 8.5, it was observed that of all the landmasses, Borneo would see the most significant redistribution and would become more homogenous in its rainfall pattern.||URI:||http://hdl.handle.net/10356/73371||Fulltext Permission:||open||Fulltext Availability:||With Fulltext|
|Appears in Collections:||CEE Theses|
Page view(s) 50153
checked on Oct 25, 2020
checked on Oct 25, 2020
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.