Assessments of surface latent heat flux associated with the Madden–Julian Oscillation in reanalyses

Abstract

To understand the accuracy and uncertainty of surface latent heat flux (LHF) associated with the Madden–Julian Oscillation (MJO), the LHF from each of the six global reanalysis datasets is compared with LHF based on in situ data and the objectively analyzed air–sea flux (OAFlux), in terms of tropical intraseasonal variability. The reanalysis products used in this study include the European Centre for Medium-Range Weather Forecasts Interim Reanalysis (ERA-I), the Modern-Era Retrospective Analysis for Research and Applications (MERRA), three generations of reanalysis from the National Center for Environmental Prediction (NCEP R1, R2 and CFSR), and the twentieth century reanalysis (20CR). We find that the intraseasonal LHF of the reanalysis products agrees well with the OAFlux over the tropical oceans in terms of patterns, but there is a significant spread in amplitude among the reanalysis products. Both ERA-I and MERRA show smaller biases in the power spectral analysis, while the other reanalysis products (NCEP R1, NCEP R2, CFSR, and 20CR) tend to overestimate the intraseasonal LHF when compared with the TAO buoy products and OAFlux. The role of anomalous LHF in supporting the MJO convection identified by previous TAO buoy data studies is confirmed by the long-term global reanalyses. The feature of increasing LHF accompanied by growing MJO observed in the recent MJO field campaign in the central Indian Ocean (DYNAMO/CINDY2011) is also well captured by the reanalysis products. Among the reanalysis datasets, MERRA has the smallest bias in temporal variability of LHF during the DYNAMO/CINDY2011 period.