Tuesday 17 Nov 2015Spatial Modelling of Extreme Precipitation in Urban Areas

Chandra Rupa - Indian Institute of Science, Bangalore, India

Harrison 170 14:30-15:30


Climatology in urban areas is different when compared to other areas, both spatially and temporally. In an urban setting, the spatial variation of precipitation can be high; the precipitation amounts and patterns often vary within short distances of less than 5 km. Therefore it is crucial to study the uncertainties in the spatial variation of precipitation in urban areas A Bayesian hierarchical model is used to obtain the return levels maps over Bangalore city and its surrounding, where a return level is a common measure of extreme precipitation events for a specified return period. The block maxima of rainfall is considered to follow Generalized Extreme Value (GEV) distribution with the location and scale parameters representing the underlying variables of hierarchical model’s process layer, and considering the shape parameter to be constant over the study region. The location and scale parameters are modelled as a Gaussian process with mean as linear regression over spatial covariates, geological and climatological, and standard deviation as exponential covariance model. This latent process model allows obtaining estimates of return levels at gauged as well as at ungauged locations. A Markov Chain Monte Carlo (MCMC) algorithm is used for solving the hierarchical model



 



 



Climatology in urban areas is different when compared to other areas, both spatially and temporally. In an urban setting, the spatial variation of precipitation can be high; the precipitation amounts and patterns often vary within short distances of less than 5 km. Therefore it is crucial to study the uncertainties in the spatial variation of precipitation in urban areas A Bayesian hierarchical model is used to obtain the return levels maps over Bangalore city and its surrounding, where a return level is a common measure of extreme precipitation events for a specified return period. The block maxima of rainfall is considered to follow Generalized Extreme Value (GEV) distribution with the location and scale parameters representing the underlying variables of hierarchical model’s process layer, and considering the shape parameter to be constant over the study region. The location and scale parameters are modelled as a Gaussian process with mean as linear regression over spatial covariates, geological and climatological, and standard deviation as exponential covariance model. This latent process model allows obtaining estimates of return levels at gauged as well as at ungauged locations. A Markov Chain Monte Carlo (MCMC) algorithm is used for solving the hierarchical model



 



About Chandra:



Chandra Rupa is a research student with Prof. P P Mujumdar in Indian Institute of Science, Bangalore, India. Her research is mainly focused on estimating uncertainties in extreme precipitation in urban areas, quantification of uncertainty in Intensity Duration Frequency (IDF) relationships, spatio-temporal modelling of IDF relationships for urban areas. She is having a work experience of about 7 years in designing hydro-power plants, irrigation structures and bridges.


Visit website

Add to calendar

Add to calendar (.ics)