Premium
Characterizing noise and spurious convection in convective data assimilation
Author(s) -
Lange Heiner,
Craig George C.,
Janjić Tijana
Publication year - 2017
Publication title -
quarterly journal of the royal meteorological society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.744
H-Index - 143
eISSN - 1477-870X
pISSN - 0035-9009
DOI - 10.1002/qj.3162
Subject(s) - ensemble kalman filter , spurious relationship , data assimilation , convection , radar , advection , meteorology , kalman filter , noise (video) , environmental science , statistical physics , extended kalman filter , mathematics , computer science , physics , statistics , artificial intelligence , telecommunications , image (mathematics) , thermodynamics
The occurrence of noise and spurious convective cells in analyses of convective‐scale data assimilation (DA) is a problem that is not quantified sufficiently to develop reliable set‐ups for the DA of radar observations in an ensemble Kalman filter. This article presents two approaches to quantifying these phenomena using a local ensemble transform Kalman filter testbed, where simulated radar observations are assimilated with varying spatial and temporal DA parameters. Firstly, the dynamical coupling of cold pools and emerging cells of organized deep convection is quantified in terms of spatio‐temporal correlations, so that spurious perturbations can be identified. Secondly, the abundance of gravity‐wave noise is characterized by the variance of the vertical velocity field during DA cycling and the ensemble forecasts. This evaluation is performed separately for subsets of model points inside observed storms, in their vicinity and in the surrounding clear‐air regions. Varying horizontal localization lengths, observation averaging scales and the length of the assimilation cycling interval are compared, with respect to the two diagnostic approaches. It is concluded that the cold‐pool coupling method and the partitioned variance of vertical velocity fields are viable measures to quantify the influence of DA parameters and algorithms on the balance of analysis states that are used to initiate forecasts of convection.