A Comparison of Two Methods for Initiating Air Mass Back Trajectories

Wednesday, 17 December 2014
Annie Putman, Eric S Posmentier, Anthony M Faiia, Leslie J Sonder and Xiahong Feng, Dartmouth College, Earth Sciences, Hanover, NH, United States
Lagrangian air mass tracking programs in back cast mode are a powerful tool for estimating the water vapor source of precipitation events. The altitudes above the precipitation site where particle’s back trajectories begin influences the source estimation. We assume that precipitation comes from water vapor in condensing regions of the air column, so particles are placed in proportion to an estimated condensation profile. We compare two methods for estimating where condensation occurs and the resulting evaporation sites for 63 events at Barrow, AK. The first method (M1) uses measurements from a 35 GHz vertically resolved cloud radar (MMCR), and algorithms developed by Zhao and Garrett1 to calculate precipitation rate. The second method (M2) uses the Global Data Assimilation System reanalysis data in a lofting model. We assess how accurately M2, developed for global coverage, will perform in absence of direct cloud observations. Results from the two methods are statistically similar. The mean particle height estimated by M2 is, on average, 695 m (s.d. = 1800 m) higher than M1. The corresponding average vapor source estimated by M2 is 1.5⁰ (s.d. = 5.4⁰) south of M1. In addition, vapor sources for M2 relative to M1 have ocean surface temperatures averaging 1.1⁰C (s.d. = 3.5⁰C) warmer, and reported ocean surface relative humidities 0.31% (s.d. = 6.1%) drier. All biases except the latter are statistically significant (p = 0.02 for each). Results were skewed by events where M2 estimated very high altitudes of condensation. When M2 produced an average particle height less than 5000 m (89% of events), M2 estimated mean particle heights 76 m (s.d. = 741 m) higher than M1, corresponding to a vapor source 0.54⁰ (s.d. = 4.2⁰) south of M1. The ocean surface at the vapor source was an average of 0.35⁰C (s.d. = 2.35⁰C) warmer and ocean surface relative humidities were 0.02% (s.d. = 5.5%) wetter. None of the biases was statistically significant. If the vapor source meteorology estimated by M2 is used to determine vapor isotopic properties it would produce results similar to M1 in all cases except the occasional very high cloud. The methods strive to balance a sufficient number of tracked air masses for meaningful vapor source estimation with minimal computational time.
  1. Zhao, C and Garrett, T.J. 2008, J. Geophys. Res