We investigate the linear orthogonal modes associated with monthly precipitation in the northeastern United States (N.E. U.S.), using CESM1 LENS and two reanalysis datasets (ERA5 and NOAA-CIRES-DOE 20CRv3). Using precipitation anomaly time series and anomalies for relevant 2D atmospheric fields, a novel linear orthogonal decomposition (LOD) method is applied to sub-sample the time series of all atmospheric predictor fields at all grid points, with the ultimate goal of maximizing the linear predictability (using multiple linear regression, MLR) of N.E. U.S. precipitation. LOD is complementary to other methods for decomposing the atmospheric fields associated with precipitation, such as k-means clustering and self-organizing maps, with the added advantage that the linear orthogonal modes are easily combined within a single linear model.

By projecting these linear orthogonal modes onto the full set of 2D atmospheric fields, we associate the first mode in each month with vapor transport along the Atlantic seaboard, the second mode with westward vapor transport from extratropical cyclones south of the N.E. U.S., and the third mode with vapor transport from the Gulf of Mexico during the fall, winter, and spring, and mesoscale convective systems in the summer.

This presentation will show that the methodology applies reasonably well to all three datasets, and MLR produces R-squared values of 0.52 – 0.65 for CESM1 LENS, and values of around 0.60 – 0.88 for the reanalysis products.

Our study provides a framework by which to examine regional-scale precipitation around the world, and the large-scale drivers that can influence the precipitation. Such insight can be a valuable resource for future regional water management practices. As such, with time permitting, we will apply this methodology to future time periods in order to look for any climatological trends in the precipitation and its associated large-scale drivers.