Comparison of SST diurnal variation models over the tropical warm pool region

Four sea surface temperature (SST) diurnal variation (DV) models have been compared against Multifunctional Transport Satellite-1R (MTSAT-1R) SST measurements over the Tropical Warm Pool (TWP) region (908E-1708E, 258S-158N) for 4 months from January to April 2010. The four models include one empirical model formulated by Chelle Gentemann (hereafter CG03), one physical model proposed by Zeng and Beljaars in 2005 (ZB05) and its updated version (ZB1T), and one air-sea coupled model (the Met Office Unified Model Global Coupled configuration 2, GC2) with ZB05 warm layer scheme added on top of the standard configuration. The sensitivity of the v3 MTSAT-1R data to the ‘‘true’’ changes in SST is first investigated using drifting buoys and is estimated to be 0.6060.05. This being significantly different from 1, the models are validated against MTSAT-1R data and the same data scaled by the inverse of the sensitivity (representing an estimate of the true variability). Results indicate that all models are able to capture the general DV patterns but with differing accuracies and features. Specifically, CG03 and ZB1T underestimate strong (>2 K) DV events’ amplitudes especially if we assume that sensitivity-scaled MTSAT-1R variability is most realistic. ZB05 can effectively capture the DV cycles under most DV and wind conditions, as well as the DV spatial distribution. GC2 tends to overestimate small-moderate (< 2 K) DV events but can reasonably predict large DV events. One to three hour lags in warming start and peak times are found in GC2.