18th Conference on Weather and Forecasting, 14th Conference on Numerical Weather Prediction, and Ninth Conference on Mesoscale Processes

Tuesday, 31 July 2001
Revisiting the utility of Newtonian nudging for four dimensional data assimilation in high latitude mesoscale forecasts
Jeffrey S. Tilley, Univ. of Alaska, Fairbanks, AK; and X. Fan
Poster PDF (410.4 kB)
Although there have been previous studies regarding four dimensional data assimilation (FDDA) with regards to high latitude mesoscale, it remains relatively unexplored compared to the vast array of work pertaining to mid-latitude mesoscale modeling or global modeling applications. In previous work (Tilley et. al 1996) we briefly explored theutility of a one-dimensional variational approach (VA) for moisture variables in the PSU/NCAR MM5 model, with and without Newtonian nudging (NN) of the wind fields. We found that the results depended heavily on the degree of agreement between the MM5 background and the satellite retrieved profiles used in the VA.

The VA approach was computationally expensive to implement compared to the NN method already available in MM5. As such, for applications with other large computational constraints over the high latitudes (expensive parameterizations, ensemble forecasts, etc.), NNM may still be able, even with limited conventional data, to provide acceptable results if an optimal formulation can be developed for the nudging itself

In this paper we revisit the NN method for an extended heavy rain event (totals of 1-2" reported) in Interior Alaska during August 2000. This type of event is relatively rare from a climatological standpoint and should provide an extreme test of the NN method for a relatively data sparse region. We conduct a series of MM5 simulations in which the variables nudged, the nudging approach (nudging to observations only, analyses only or both), and the nudging coefficients are modified. The results are compared with each other, analyses and observations including verification skill scores. We also construct an ensemble of these forecasts to determine if there is potentially any added value in such an approach. These simulations also form the basis of comparison for experiments, presented in a companion paper, using an intermittent data assimilation approach.

Supplementary URL: