Computing & Software
My research relies heavily on statistical computing, and I am committed to open and reproducible science. I develop and maintain open-source software packages in R that implement the statistical methods underlying my research, making these tools freely available to the broader scientific community. I also develop interactive web-based tools using R Shiny to make my research accessible to a wider audience. All software and code associated with my publications is publicly available on GitHub.
Software Packages
convoSPAT
The convoSPAT package for R provides computationally efficient tools for fitting convolution-based nonstationary spatial Gaussian process models, in which covariance parameters such as local anisotropy, variance, and nugget are allowed to vary continuously across space using a discrete mixture component representation estimated via local likelihood methods.
BayesNSGP
The BayesNSGP package for R enables fully Bayesian inference for nonstationary spatial Gaussian process models, allowing spatially varying covariance parameters to be specified via regression or stochastic processes, with scalable approximate likelihood methods (NNGP and Vecchia) for large datasets and flexible MCMC sampling implemented via the nimble package.
Interactive Tools
Gridded Seasonal Precipitation Explorer
This app estimates seasonal mean and extreme precipitation over the United States for different combinations of external forcing and modes of climate variability. Users can explore spatial maps of precipitation and partition differences into individual contributions from specific modes of variability.
Impossible Temperatures: Heatwave Attribution
This app provides data-driven upper bounds and event attribution for extreme heatwaves globally. Users can explore factual and counterfactual probabilities for individual heatwave events, providing a rigorous framework for quantifying the influence of climate change on observed temperature extremes.
