Impacts of the MJO in the Indian Ocean and on the Western Australian coast

The large sea surface temperature variations induced by the Madden-Julian Oscillation (MJO) on the northwest shelf of Australia and the remote influence of the MJO on the subtropical Western Australian coast are explored using the POAMA Ensemble Ocean Data Assimilation System reanalyses (PEODAS) for the period 1980–2010. The focus here is during the November–April extended summer season when the impacts of the MJO on and along the west coast of Australia are greatest. The MJO is well known to force equatorial Kelvin and Rossby waves in the Indian Ocean, and these are well depicted in the PEODAS reanalyses. When the downwelling Kelvin waves (forced by the westerly-convective phase of the MJO) reach the Indonesian region at the eastern boundary of the Indian Ocean, a coastally trapped Kelvin wave appears to propagate southeast along the Indonesian coastline. At the same time, the suppressed convection/easterly phase of the MJO arrives in the eastern Indian Ocean, with increased heat flux into the ocean due to reduced latent heat flux and increased insolation. The coastally trapped Kelvin waves do not appear to get onto the Western Australian coast. Rather, the increased heat flux and Ekman-induced downwelling onto the northwest (NW) coast in the suppressed/easterly phase of the MJO drive an increase in sea surface temperature on the NW Australian shelf. The piling up of warm water and associated sea level rise on the NW shelf is then communicated down the Western Australian coast as a coastally trapped wave, resulting in an increase in the Leeuwin current. Thus we conclude that the MJO signal in sea level along the west coast of Australia does not result from transmission of equatorial waves onto the Western Australian coast, but rather a southward-propagating coastal trapped wave that is directly forced on the NW shelf through Ekman-induced vertical advection and surface heat fluxes in the easterly phase of the MJO. Additionally, subtropical coastal sea level variability is reinforced locally via a teleconnection of the MJO to the local meridional wind off the southwest Australian coast. Considering the capability to predict the MJO to about 4 weeks lead time plus the 2 weeks taken for the MJO signal on the NW shelf to influence sea level at Fremantle, the use of MJO forecasts in management of the Western Australian marine environment should be considered for future application.     

Impact of improved assimilation of temperature and salinity for coupled model seasonal forecasts

We assess the impact of improved ocean initial conditions for predicting El Niño-Southern Oscillation (ENSO) and Indian Ocean dipole (IOD) using the Bureau of Meteorology’s Predictive Ocean Atmosphere Model for Australia (POAMA) coupled seasonal prediction model for the period 1982–2006. The new ocean initial conditions are provided by an ensemble-based analysis system that assimilates subsurface temperatures and salinity and which is a clear improvement over the previous optimal interpolation system which used static error covariances and was univariate (temperature only). Hindcasts using the new ocean initial conditions have better skill at predicting sea surface temperature (SST) variations associated with ENSO than do the hindcasts initialized with the old ocean analyses. The improvement derives from better prediction of subsurface temperatures and the largest improvements come during ENSO–IOD neutral years. We show that improved prediction of the Niño3.4 SST index derives from improved initial depiction of the thermocline and halocline in the equatorial Pacific but as lead time increases the improved depiction of the initial salinity field in the western Pacific become more important. Improved ocean initial conditions do not translate into improved skill for predicting the IOD but we do see an improvement in the prediction of subsurface temperatures in the Indian Ocean (IO). This result reflects that the coupling between subsurface and surface temperature variations is weaker in the IO than in the Pacific, but coupled model errors may also be limiting predictive skill in the IO.     

Tropical forcing of Australian extreme low minimum temperatures in September 2019

We explore the causes and predictability of extreme low minimum temperatures (T_min) that occurred across northern and eastern Australia in September 2019. Historically, reduced T_min is related to the occurrence of a positive Indian Ocean Dipole (IOD) and central Pacific El Niño. Positive IOD events tend to locate an anomalous anticyclone over the Great Australian Bight, therefore inducing cold advection across eastern Australia. Positive IOD and central Pacific El Niño also reduce cloud cover over northern and eastern Australia, thus enhancing radiative cooling at night-time. During September 2019, the IOD and central Pacific El Niño were strongly positive, and so the observed T_min anomalies are well reconstructed based on their historical relationships with the IOD and central Pacific El Niño. This implies that September 2019 T_min anomalies should have been predictable at least 1–2 months in advance. However, even at zero lead time the Bureau of Metereorolgy ACCESS-S1 seasonal prediction model failed to predict the anomalous anticyclone in the Bight and the cold anomalies in the east. Analysis of hindcasts for 1990–2012 indicates that the model's teleconnections from the IOD are systematically weaker than the observed, which likely stems from mean state biases in sea surface temperature and rainfall in the tropical Indian and western Pacific Oceans. Together with this weak IOD teleconnection, forecasts for earlier-than-observed onset of the negative Southern Annular Mode following the strong polar stratospheric warming that occurred in late August 2019 may have contributed to the T_min forecast bust over Australia for September 2019.

     

Assessing the simulation and prediction of rainfall associated with the MJO in the POAMA seasonal forecast system

We assess the ability of the Predictive Ocean Atmosphere Model for Australia (POAMA) to simulate and predict weekly rainfall associated with the MJO using a 27-year hindcast dataset. After an initial 2-week atmospheric adjustment, the POAMA model is shown to simulate well, both in pattern and in intensity, the weekly-mean rainfall variation associated with the evolution of the MJO over the tropical Indo-Pacific. The simulation is most realistic in December?CFebruary (austral summer) and least realistic in March?CMay (austral autumn). Regionally, the most problematic area is the Maritime Continent, which is a common problem area in other models. Coupled with our previous demonstration of the ability of POAMA to predict the evolution of the large-scale structure of the MJO for up to about 3?weeks, this ability to simulate the regional rainfall evolution associated with the MJO translates to enhanced predictability of rainfall regionally throughout much of the tropical Indo-Pacific when the MJO is present in the initial conditions during October?CMarch. We also demonstrate enhanced prediction skill of rainfall at up to 3?weeks lead time over the north-east Pacific and north Atlantic, which are areas of pronounced teleconnections excited by the MJO-modulation of tropical Indo-Pacific rainfall. Failure to simulate and predict the modulation of rainfall in such places as the Maritime Continent and tropical Australia by the MJO indicates, however, there is still much room for improvement of the prediction of the MJO and its teleconnections.     

On the importance of initializing the stochastic part of the atmosphere for forecasting the 1997/1998 El Ni?o

The importance of initializing atmospheric intra-seasonal (stochastic) variations for prediction of the onset of the 1997/1998 El Ni?o is examined using the Australian Bureau of Meteorology coupled seasonal forecast model. A suite of 9-month forecasts was initialized on the 1st December 1996. Observed ocean initial conditions were used together with five different atmospheric initial conditions that sample a range of possible initial states of intra-seasonal (stochastic) variability, especially the Madden-Julian Oscillation (MJO), which is considered the primary stochastic variability of relevance to El Ni?o evolution. The atmospheric initial states were generated from a suite of atmosphere-only integrations forced by observed sea surface temperatures (SST). To the extent that low frequency variability of the tropical atmosphere is forced by slow variations in SST, these atmospheric states should all represent realistic low frequency atmospheric variability that was present in December 1996. However, to the extent that intra-seasonal variability is not constrained by SST, they should capture a range of intra-seasonal states, especially variations in the activity, phase and amplitude of the MJO. For each of these five states, a 20-member ensemble of coupled model forecasts was generated by the addition of small random perturbations to the SST field at the initial time. The ensemble mean from all five sets of forecasts resulted in El Ni?o but three of the sets produced substantially greater warming by months 4?C5 in the NINO3.4 region compared to the other two. The warmer group stemmed from stronger intra-seasonal westerly wind anomalies associated with the MJO that propagated eastward into the central Pacific during the first 1?C2?months of the forecast. These were largely absent in the colder group; the weakest of the colder group developed strong easterly wind anomalies, relative to the grand ensemble mean, that propagated into the central Pacific early in the forecast, thereby generating significantly weaker downwelling Kelvin waves in comparison to the warmer group. The strong reduction in downwelling Kelvin waves in the weakest case acted to limit the warming in the eastern Pacific, resulting in a ??Modoki?? type El Ni?o that is more focused in the central Pacific. Our results suggest that the intra-seasonal stochastic component of the atmospheric initial condition has an important and potentially predictable impact on the forecasts of the initial warming and flavour of the 1997/1998 El Ni?o. However, to the extent that atmospheric intra-seasonal variability is not predictable beyond a month or two, these results imply a limit to the accuracy with which the strength and perhaps the spatial distribution of an El Ni?o can ultimately be predicted. These results do not preclude a predictable role of the MJO and other intra-seasonal stochastic variability at longer lead times if some aspects of the stochastic variability are preconditioned by the evolving state of El Ni?o or other low frequency boundary forcing.     

Long-Term Changes in Soil Water Tables over thePast 4500 Years: Relationships with Climate and NorthAtlantic Atmospheric Circulation and Sea Surface Temperature

A record of mean annual water table changes in Englandover the past 4500 years is derived from a transferfunction applied to protozoan assemblages in peatprofiles. A 100-year running mean value based on threeprofiles shows short-term changes which are closelyrelated to instrumental records of mean annualtemperature changes (past 300 yr) and documentaryrecords of summer wetness and winter severity (past900 yr). Long-term (>1000 yr) changes in thiscomposite record are positively correlated with theGISP2 ice accumulation record, suggesting moisturebudgets were in phase across the North Atlantic regionover at least the last 2000 years and probably changeddue to north-south movements of major pressurecentres. This hypothesis is further supported bycomparisons with indicators of sea surface temperatureand ocean circulation. Existing regional climatepredictions may underestimate the impact of futurewarming on soil moisture status, with significant implications for agriculture and water supply in northwestern Europe.     

Seasonal prediction of the Leeuwin Current using the POAMA dynamical seasonal forecast model

The potential for predicting interannual variations of the Leeuwin Current along the west coast of Australia is addressed. The Leeuwin Current flows poleward against the prevailing winds and transports warm-fresh tropical water southward along the coast, which has a great impact on local climate and ecosystems. Variations of the current are tightly tied to El Niño/La Niña (weak during El Niño and strong during La Niña). Skilful seasonal prediction of the Leeuwin Current to 9-month lead time is achieved by empirical downscaling of dynamical coupled model forecasts of El Niño and the associated upper ocean heat content anomalies off the north west coast of Australia from the Australian Bureau of Meteorology Predictive Ocean Atmosphere Model for Australia (POAMA) seasonal forecast system. Prediction of the Leeuwin Current is possible because the heat content fluctuations off the north west coast are the primary driver of interannual annual variations of the current and these heat content variations are tightly tied to the occurrence of El Niño/La Niña. POAMA can skilfully predict both the occurrence of El Niño/La Niña and the subsequent transmission of the heat content anomalies from the Pacific onto the north west coast.     

Simulation and prediction of blocking in the Australian region and its influence on intra-seasonal rainfall in POAMA-2

We assess the depiction and prediction of blocking at 140°E and its impact on Australian intra-seasonal climate variability in the Bureau of Meteorology’s dynamical intra-seasonal/seasonal forecast model Predictive Ocean Atmosphere Model for Australia version 2 (POAMA-2). The model simulates well the strong seasonality of blocking but underestimates its strength and frequency increasingly with lead time, particularly after the first fortnight of the hindcast, in connection with the model’s drifting basic state. POAMA-2 reproduces well the large-scale structure of weekly-mean blocking anomalies and associated rainfall anomalies over Australia; the depiction of total blocking in POAMA-2 may be improved with the reduction of biases in the distribution of Indian Ocean rainfall via a tropical-extratropical wave teleconnection linking blocking activity at 140°E with tropical variability near Indonesia. POAMA-2 demonstrates the ability to skilfully predict the daily blocking index out to 16 days lead time for the ensemble mean hindcast, surpassing the average predictive skill of the individual hindcast members (5 days), the skill obtained from persistence of observed (2 days), and the decorrelation timescale of blocking (3 days). This skilful prediction of the blocking index, together with effective simulation of blocking rainfall anomalies, translates into higher skill in forecasting rainfall in weeks 2 and 3 over much of Australia when blocking is high at the initial time of the hindcast, compared to when the blocking index is small. POAMA-2 is thus capable of providing forecast skill for blocking rainfall on the intra-seasonal timescale to meet the needs of Australian farming communities, whose management practises often rely upon decisions being made a few weeks ahead.     

Impact of including surface currents on simulation of Indian Ocean variability with the POAMA coupled model

Impacts on the coupled variability of the Indo-Pacific by including the effects of surface currents on surface stress are explored in four extended integrations of an experimental version of the Bureau of Meteorology??s coupled seasonal forecast model POAMA. The first pair of simulations differs only in their treatment of momentum coupling: one version includes the effects of surface currents on the surface stress computation and the other does not. The version that includes the effect of surface currents has less mean-state bias in the equatorial Pacific cold tongue but produces relatively weak coupled variability in the Tropics, especially that related to the Indian Ocean dipole (IOD) and El Ni?o/Southern Oscillation (ENSO). The version without the effects of surface currents has greater bias in the Pacific cold tongue but stronger IOD and ENSO variability. In order to diagnose the role of changes in local coupling from changes in remote forcing by ENSO for causing changes in IOD variability, a second set of simulations is conducted where effects of surface currents are included only in the Indian Ocean and only in the Pacific Ocean. IOD variability is found to be equally reduced by inclusion of the local effects of surface currents in the Indian Ocean and by the reduction of ENSO variability as a result of including effects of surface currents in the Pacific. Some implications of these results for predictability of the IOD and its dependence on ENSO, and for ocean subsurface data assimilation are discussed.