Robust Multivariate Bias Correction (RoMBC)
The Robust Multivariate Bias Correction (RoMBC) package is a robust package designed to deal with the distribution and dependence biases in a multivariate time series. The package works around a self-evolving strategy. It starts with a simple single timescale univariate bias correction and depending upon the requirement of the data at hand, grows into a comprehensive multivariate multi time scale bias correction. The package avoids the necessity of choosing the type of bias correction, time levels for nesting and type of cross nesting required. RoMBC is a comprehensive package that offers a wide variety of in-built options by including variants of standard quantile matching and other routinely used bias correction approaches in a time and cross dependence nesting.
In the software, distribution biases at the entry level (daily or monthly) of the time series are corrected using a univariate quantile matching (QM) approach. Auto dependence is represented by LAG1 autocorrelation function while cross-dependence is assumed to be a function of LAG0 and LAG1 correlations. The package looks for the need of bias correction at daily, monthly, seasonal and annual time scales.
Mehrotra and Sharma (2021), A robust alternative for correcting systematic biases in multi-variable climate model simulations, Environ. Model. Software, 139(2021), 105019, https://doi.org/10.1016/j.envsoft.2021.105019
Research publication source: https://www.sciencedirect.com/science/article/pii/S1364815221000621?dgcid=author