Analyses of Lagrangian model simulations and atmospheric sonde data reveal a key component of the Madden-Julian Oscillation (MJO): circumnavigating equatorial Kelvinwaves with dynamics that transform between dry and moist, which initiate and dissipate MJO convection. The same compositing procedure is applied to simulated and observed MJOs, which uses a coordinate system that moves with the precipitation centre, and treats developing, mature and dissipating stages separately. MJO structure and evolution are similar in the simulation and the observations. To the west of the developing convection, there is a broad region of low-level (upper-level) perturbation westerlies (easterlies), which is accompanied by a deep negative temperature perturbation. As this feature enters the MJO-formation region, convection intensifies on its eastern edge, and the zonal wind perturbations decrease in zonal extent and propagation speed. This process is shown to be a dynamical consequence of a largely dry, first baroclinic Kelvin wave entering a region where large-scale upward motion is mostly balanced by convective heating. As the MJO matures, a Kelvin wave of opposite sign emerges from its eastern edge, and makes the opposite transition (from moist to dry). The resulting wave, which includes low-level (upper-level) easterlies (westerlies) and a deep positive temperature perturbation, rapidly propagates around the world and dissipates MJO convection. The Kelvin wave that initiates MJO convection is shown to originate from the inactive phase of a previous MJO, so that the complete MJO cycle is characterized by the two kinds of Kelvin waves emerging from active and suppressed phases of MJO convection, circumnavigating the Tropics, and triggering the opposite phase.

Climate models predict increases in drought conditions in many parts of the tropics. Yet the response of tropical forests to drought remains highly uncertain, especially with regards to the factors that generate spatial heterogeneity in drought response across landscapes. In this study, we used Landsat imagery to assess the impacts of a severe drought in 2015 across an ~80,000‐ha landscape in Puerto Rico. Specifically, we asked whether drought effects varied systematically with topography and with forest age, height, and fragmentation. We quantified drought impacts using anomalies of two vegetation indices, the enhanced vegetation index (EVI) and normalized difference water index (NDWI), and fit random forest models of these metrics including slope, aspect, forest age, canopy height, and two indices of fragmentation as predictors. Drought effects were more severe on drier topographic positions, that is, steeper slopes and southwest‐facing aspects, and in second‐growth forests. Shorter and more fragmented forests were also more strongly affected by drought. We also assessed which factors were associated with stronger recovery from drought. Factors associated with more negative drought anomalies were also associated with more positive postdrought anomalies, suggesting that increased light availability as a result of drought led to high rates of recovery in forests more severely affected by drought. In general, recovery from drought was rapid across the landscape, with postdrought anomalies at or above average across the study area. This suggests that forests in Puerto Rico might be resilient to a single‐year drought, though vulnerability to drought varies depending on forest characteristics and landscape position.

No abstract