Objective Array Design: Application to the Tropical Indian Ocean

Abstract

A simple, versatile, computationally efficient ensemble-based method for objectively designing an observation array is described. The method seeks to compute the observation array that minimizes the analysis error variance, according to Kalman filter theory. While most elements of the method have been described elsewhere, this paper attempts to present a simple, yet comprehensive, recipe for array design based on an ensemble of anomalies that represents the background error covariance. The versatility of the method is demonstrated through a series of applications to the tropical Indian Ocean (TIO). The first application uses model-generated fields of high-pass-filtered mixed layer depth to design an array to monitor intraseasonal variability. The second uses gridded observations of sea level anomaly to design an array to monitor intraseasonal-to-interannual variability. For both applications, the objectively designed arrays are compared to an array that will soon be implemented under the auspices of the Climate Variability and Predictability?Global Ocean Observing System (CLIVAR?GOOS) Indian Ocean Panel (CG-IOP). The authors conclude that the CG-IOP array produces results that compare well to the objectively designed arrays for intraseasonal variability, and observations to the east and northeast of the TIO and south of India are most important for resolving intraseasonal variability. The authors also find that observations near 9°S, where seasonal Rossby waves dominate, are important for observing seasonal-to-interannual variability. The described method for objective array design can be applied to a wide range of geophysical applications where time series of gridded modeled or observed fields are available.