1. ISCCP-defined weather regimes and air-sea interaction Carol Anne Clayson Woods Hole Oceanographic Institution With Brent Roberts, MSFC ISCCP at 30 CCNY, New York, NY 24 April 2013

2. Motivation and background Use of ISCCP cluster weather states (Jakob and Tselioudis 2003) Tropical convection and MJO (Tromeur and Rossow, 2010; Chen and Del Genio, 2009) Datasets: ISCCP Extratropical Cloud Clusters (35N/S, 2.5°x2.5° 1985-2007, 3-hr) SEAFLUX (1998-2007,0.25°x0.25° 3-hr), LHF/SHF/Surface Variables Product Homogenization: Fluxes regridded and resampled to ISCCP 2.5x2.5 ISCCP 3-hr used to assign a daily class based on the most frequent cluster More convection Less convection

3. Decomposition of surface fluxes by weather state Weather regimes result in distributions of fluxes with different mean and extreme characteristics These are associated with changes in the bulk variables, as should be expected Both wind speed and near- surface humidity gradients are particularly well stratified, though the latent heat flux means are less so Indicates potential compensations LHF SHF Qs-Qa Ts-Ta Wind Speed

4. Compositing methodology Conditionally sample data using weather state classification (WS1-WS8; most convective to least convective) Further sampled based on compositing index to evaluate low-frequency coupled variability Use NOAA Climate Prediction Center (CPC) indices for ENSO and MJO  Examining differences in means can be decomposed as changes in class mean (A), changes in RFO (B), and covariant changes (C) ABC

5. MJO Composites by strength Composite MJO based on index strength not time-lagging All three regions typically show increased evaporation during convective phase and decreased evaporation during suppressed phase The Indo-Pacific region changes  more wind-driven Eastern Pacific changes  more near- surface moisture gradient changes But: EIO more coherent near- surface moisture changes than WP Convective NeutralSuppressed EIO 70E-90E WP 130E-150E EP 130W-110W

6. MJO Composites – Decomposition into Weather states Decompose LHF into weather state means and relative frequency of occurrence (RFO) Systematic variations of both weather state means and RFO with MJO index Both variations contribute to total impact of a given weather state on mean energy exchange associated with MJO evolution

7. MJO Composites – Decomposition of changes The difference between convective, neutral, and suppressed conditions can be quantitatively decomposed into Mean-,RFO-, and covariant- driven change Convective vs. Neutral changes are primarily set by the systematic variation of class properties rather than RFO changes Changes in Western Pacific: wind speed. East Pacific: Qs-Qa. Eastern Indian: both EIOWPEP dX dRFO Covar Total

8. ENSO Composites by strength West Pacific (130E-150E) latent heating anomalies primarily driven by QSQA anomalies For MJO, near-surface wind speed was also anti-correlated but it was stronger than QSQA The East Pacific (130W-110W) LHF acts to damp the existing SST anomalies Unlike on MJO time scales, wind speed and QSQA are positively correlated EIO WP EP El Nino La Nina

9. ENSO Decomposition of changes Surface LHF changes on ENSO time scales show little sensitivity to reorganization of weather states (δRFO) On both MJO and ENSO time scales, near-surface wind speed and vertical humidity anomalies are typically anti-correlated, except in the East Pacific Higher wind speed – more mixing reduces near-surface gradient? EIO WP EP dX dRFO Covar Total EL NINO MJO Convective

10. Summary Cloud-based weather states can be used to provide improved understanding of surface energy flux variability MJO variability is particularly well decomposed using ISCCP weather regimes from convective to neutral and suppressed states Different regions in the tropics show MJO and ENSO variability being driven by different processes Both MJO and ENSO variations in surface latent heat flux are associated mostly with changes in the mean associated with a particular weather state, rather than the change in frequency of occurrence of a weather state

11. Extra Intercomparing products by weather state While there are systematic mean differences in products, the anomalous changes between products (here, SeaFlux & OAFlux) are more closely aligned. The differences here can be related to specific types of weather regimes OAFlux shows a slight increase in the latent heat flux associated with deep convective conditions while SeaFlux shows a slight decrease. In broken stratocumulus conditions, SeaFlux indicates about a 20% change, nearly 2x that of OAFlux, again primarily from differences in near-surface moisture gradients

12. Outline  Motivation & Background Tropical weather states and reasons for clustering  Approach Datasets used; clustering Compositing Methodology  Results How well do the weather states decompose the fluxes? Changes associated with Madden-Julian Oscillation  Summary General Conclusions Future Work