Use of Energy Preserving Empirical Mode Decomposition to Determine Trends in Simulated Precipitation and Temperature from a High Resolution Ensemble of Climate Models over the Contiguous U.S.
Abstract
Several methods exist to quantify trends in hydroclimatic parameters including the widely used non-parametric Mann-Kendall monotonic trend test and Sen slope estimator. Energy preserving empirical mode decomposition (EPEMD) can be used on non-stationary data and can decompose a given signal into a finite number of intrinsic mode functions (IMFs) including a trend component, which is not necessarily monotonic. Here we apply a suite of statistical trend analyses to determine historical and projected changes in the hydroclimate variables across the contiguous U.S. The analysis is based on an 11-member ensemble of high-resolution (18 km) regional climate model output that includes a 1966-2005 baseline period and 2011-2050 future period under Representative Concentration Pathway 8.5. Changes to annual, seasonal and monthly precipitation, maximum temperatures, and minimum temperatures are evaluated at spatial scales corresponding to 2-digit hydrologic unit codes (HUC2). For precipitation in some regions, it was found that the Mann-Kendall test does not detect significant trends (alpha = 0.05) in the raw time series, while trends are diagnosed in the low frequency IMFs obtained from the EPEMD algorithm. In contrast, for both minimum and maximum temperatures the Mann-Kendall is able to detect trends in the raw time series, most likely since projected temperature changes are much more monotonic. The discrepancy between these methods highlights the need for more thorough and careful choice of trend analysis methods to detect robust deviations in observed and simulated climate. The use of EPEMD to analyze trends allows for a more exhaustive comparison of inter-model performance amongst the climate model ensemble members and determination of statistically significant IMF's from baseline to future climate scenarios.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2016
- Bibcode:
- 2016AGUFMGC11B1138P
- Keywords:
-
- 1616 Climate variability;
- GLOBAL CHANGEDE: 1626 Global climate models;
- GLOBAL CHANGEDE: 1655 Water cycles;
- GLOBAL CHANGE