Update to the Global Climate Data package: analysis of empirical bias correction methods in the context of producing very high resolution climate projections

While global climate models (GCMs) are useful for simulating climatic responses to perturbations in the Earth's climate system, there are many instances where higher spatial resolution information is necessary. In all instances, interpretation of interpolated or downscaled GCMs must be done cautiously because each method has its own set of assumptions and potential disadvantages. Here, we present an update to the Global Climate Data (GCD) package, which enables the package to efficiently bias correct and interpolate precipitation and air temperature output from GCM simulations to very high spatial resolutions using the delta change method. While the delta change method is relatively simple, it has previously been shown to enhance the physical representation of interpolated climate time-series compared to directly interpolating the gridded climate time-series to a higher spatial resolution. The bias correction methods programmed into the GCD package are univariate empirical quantile mapping (QM) and bivariate empirical joint bias correction (JBC). The skill of QM and JBC for improving GCM simulations processed with the delta change method is evaluated through comparing the cumulative distribution functions (CDFs) of the interpolated GCM simulations to the CDFs of Global Historical Climatology Network (GHCN) station observations for three test regions: Oregon (in the USA), the Alps (spanning several countries in Europe), and the Ganges Delta (in India and Bangladesh). We also assess the representation of precipitation and mean temperature joint probability distributions relative to those present in GHCN station observations. Overall, GCM simulations that are bias corrected with QM prior to being input to the delta change method perform best under our analysis.