Publications

The atmospheric dynamics of Jupiter are dominated by strong zonal winds engulfing the planet. Since the first gravity measurements taken by Juno at Jupiter, the low-degree gravity harmonics (J ₃–J ₁₀) have been used to determine the depth and structure of the zonal winds observed at the cloud level, limiting inferences on the deep flows to the wide latitudinal structure of these harmonics. Here, using constraints on the dynamical contribution to gravity at high latitude, we present the gravity harmonics up to J _40. We find an excellent correlation between these measurements and the gravity harmonics resulting from the observed cloud-level winds extending inwards cylindrically to depths of ~10⁵ bar (3,000 km). These measurements provide direct evidence that the flows penetrate inwards along the direction of the spin axis, confirming the cylindrical nature of the flow, which has been postulated theoretically since the 1970s. Furthermore, this detailed new gravity spectrum allows us to quantify the contribution of the various jets to the gravity signal, showing the dominance of the strong zonal flows around 20° latitude in both hemispheres.

ESA’s Jupiter Icy Moons Explorer (JUICE) will provide a detailed investigation of the Jovian system in the 2030s, combining a suite of state-of-the-art instruments with an orbital tour tailored to maximise observing opportunities. We review the Jupiter science enabled by the JUICE mission, building on the legacy of discoveries from the Galileo, Cassini, and Juno missions, alongside ground- and space-based observatories. We focus on remote sensing of the climate, meteorology, and chemistry of the atmosphere and auroras from the cloud-forming weather layer, through the upper troposphere, into the stratosphere and ionosphere. The Jupiter orbital tour provides a wealth of opportunities for atmospheric and auroral science: global perspectives with its near-equatorial and inclined phases, sampling all phase angles from dayside to nightside, and investigating phenomena evolving on timescales from minutes to months. The remote sensing payload spans far-UV spectroscopy (50-210 nm), visible imaging (340-1080 nm), visible/near-infrared spectroscopy (0.49-5.56 μm), and sub-millimetre sounding (near 530-625 GHz and 1067-1275 GHz). This is coupled to radio, stellar, and solar occultation opportunities to explore the atmosphere at high vertical resolution; and radio and plasma wave measurements of electric discharges in the Jovian atmosphere and auroras. Cross-disciplinary scientific investigations enable JUICE to explore coupling processes in giant planet atmospheres, to show how the atmosphere is connected to (i) the deep circulation and composition of the hydrogen-dominated interior; and (ii) to the currents and charged particle environments of the external magnetosphere. JUICE will provide a comprehensive characterisation of the atmosphere and auroras of this archetypal giant planet.

The shape of the two gas giants, Jupiter and Saturn, is determined primarily by their rotation rate, and interior density distribution. It is also affected by their zonal winds, causing an anomaly of O(10 km) at low latitudes. However, uncertainties in the observed cloud-level wind and the polar radius, translate to an uncertainty in the shape with the same order of magnitude. The Juno (Jupiter) and Cassini (Saturn) missions gave unprecedented accurate gravity measurements, constraining better the uncertainty in the wind structure. Using an accurate shape calculation, and a joint optimization, given both gravity and radio-occultation measurements, we calculate the possible range of dynamical height for both planets. We find that for Saturn there is an excellent match to the radio-occultation measurements, while at Jupiter such a match is not achieved. This may point to deviations from a barotropic flow above the cloud level, which might be tested with the forthcoming radio-occultation measurements by Juno.

The Juno spacecraft has been collecting data to shed light on the planet’s origin and characterize its interior structure. The onboard gravity science experiment based on X-band and Ka-band dual-frequency Doppler tracking precisely measured Jupiter’s zonal gravitational field. Here, we analyze 22 Juno’s gravity passes to investigate the gravity field. Our analysis provides evidence of new gravity field features, which perturb its otherwise axially symmetric structure with a time-variable component. We show that normal modes of the planet could explain the anomalous signatures present in the Doppler data better than other alternative explanations, such as localized density anomalies and non-axisymmetric components of the static gravity field. We explain Juno data by p-modes having an amplitude spectrum with a peak radial velocity of 10–50 cm/s at 900–1200 μHz (compatible with ground-based observations) and provide upper bounds on lower frequency f-modes (radial velocity smaller than 1 cm/s). The new Juno results could open the possibility of exploring the interior structure of the gas giants through measurements of the time-variable gravity or with onboard instrumentation devoted to the observation of normal modes, which could drive spacecraft operations of future missions.

The large-scale Hadley circulation is a key element in the global heat and moisture transport. It is traditionally defined as the zonally averaged meridional circulation in the tropics, but was shown to have a strong longitudinal dependence, as seen in a decomposition of the three-dimensional atmospheric flow into spatially dependent meridional and zonal circulations. Recent studies provided a useful analysis of the regional strengthening/weakening of the decomposed circulation but not its patterns. Here, we study the interannual variability of the longitudinally dependent meridional circulation (LMC), with a focus on its spatial patterns. We use hierarchical clustering to objectively determine the four main modes of the LMC interannual variability, and apply a Lagrangian air parcel tracking method to reveal the full circulation patterns. While El Niño and La Niña are found, as in previous studies, to play a role in setting these patterns, we find the patterns are not uniquely characterized by standard El Niño-Southern Oscillation (ENSO) indices (Nino3.4 or Southern Oscillation Index). Instead, ENSO flavors (i.e., East Pacific vs. Central Pacific) have different effects on the LMC. The most prominent interannual variability of the LMC is an east-west shift. Latitudinal shifts, as well as contraction/expansion in both latitude and longitude are also identified. Multiple linear regression analysis shows that while a large fraction of the LMC variance is explained by Sea Surface Temperature, the Madden-Julian Oscillation makes a nonnegligible independent contribution. The clustering patterns are also used to study the remote precipitation and surface air temperature teleconnections.

Jupiter's Great Red Spot (GRS) is the largest atmospheric vortex in the Solar System and has been observed for at least two centuries. It has been unclear how deep the vortex extends beneath its visible cloud tops. We examined the gravity signature of the GRS using data from 12 encounters of the Juno spacecraft with the planet, including two direct overflights of the vortex. Localized density anomalies due to the presence of the GRS caused a shift in the spacecraft line-of-sight velocity. Using two different approaches to infer the GRS depth, which yielded consistent results, we conclude that the GRS is contained within the upper 500 kilometers of Jupiter's atmosphere.

Jupiter’s atmosphere is dominated by multiple jet streams which are strongly tied to its 3D atmospheric circulation. Lacking a rigid bottom boundary, several models exist for how the meridional circulation extends into the planetary interior. Here we show, collecting evidence from multiple instruments of the Juno mission, the existence of mid-latitudinal meridional circulation cells which are driven by turbulence, similar to the Ferrel cells on Earth. Different than Earth, which contains only one such cell in each hemisphere, the larger, faster rotating Jupiter can incorporate multiple cells. The cells form regions of upwelling and downwelling, which we show are clearly evident in Juno’s microwave data between latitudes 60◦S and 60◦N. The existence of these cells is confirmed by reproducing the ammonia observations using a simplistic model. This study solves a long-standing puzzle regarding the nature of Jupiter’s sub-cloud dynamics and provides evidence for 8 cells in each Jovian hemisphere.

During the past few years, both the Cassini mission at Saturn and the Juno mission at Jupiter provided measurements with unprecedented accuracy of the gravity and magnetic fields of the two gas giants. Using the gravity measurements, it was found that the strong zonal flows observed at the cloud level of the gas giants are likely to extend thousands of kilometres deep into the planetary interior. However, the gravity measurements alone, which are by definition an integrative measure of mass, cannot constrain with high certainty the exact vertical structure of the flow. Taking into account the recent Cassini magnetic field measurements of Saturn, and past secular variations of Jupiter's magnetic field, we obtain an additional physical constraint on the vertical decay profile of the observed zonal flows on these planets. Our combined gravity-magnetic analysis reveals that the cloud-level winds on Saturn (Jupiter) extend with very little decay, i.e. barotropically, down to a depth of around 7000 km (2000 km) and then decay rapidly in the semiconducting region, so that within the next 1000 km (600 km) their value reduces to about 1 per cent of that at the cloud level. These results indicate that there is no significant mechanism acting to decay the flow in the outer neutral region, and that the interaction with the magnetic field in the semiconducting region might play a central role in the decay of the flows.

The asymmetric gravity field measured by the Juno spacecraft has allowed the estimation of the depth of Jupiter's zonal jets, showing that the winds extend approximately 3,000 km beneath the cloud level. This estimate was based on an analysis using a combination of all measured odd gravity harmonics, J 3, J 5, J 7, and J 9, but the wind profile's dependence on each of them separately has yet to be investigated. Furthermore, these calculations assumed the meridional profile of the cloud-level wind extends to depth. However, it is possible that the interior jet profile varies somewhat from that of the cloud level. Here we analyze in detail the possible meridional and vertical structure of Jupiter's deep jet streams that can match the gravity measurements. We find that each odd gravity harmonic constrains the flow at a different depth, with J 3 the most dominant at depths below 3,000 km, J 5 the most restrictive overall, whereas J 9 does not add any constraint on the flow if the other odd harmonics are considered. Interior flow profiles constructed from perturbations to the cloud-level winds allow a more extensive range of vertical wind profiles, yet when the meridional profiles differ substantially from the cloud level, the ability to match the gravity data significantly diminishes. Overall, we find that while interior wind profiles that do not resemble the cloud level are possible, they are statistically unlikely. Finally, inspired by the Juno microwave radiometer measurements, assuming the brightness temperature is dominated by the ammonia abundance, we find that depth-dependent flow profiles are still compatible with the gravity measurements.

The Hadley circulation is a key element of the climate system. It is traditionally defined as the zonally averaged meridional circulation in the tropics, therefore treated as a zonally symmetric phenomenon. However, differences in temperature between land and sea cause zonal asymmetries on Earth, dramatically affecting the circulation. This longitudinal dependence of the meridional circulation evokes questions about where and when the actual large-scale tropical circulation occurs. Here, we look into the connection between the longitudinally dependent meridional circulation, and the actual large-scale transport of air in the tropics using a coupled Eulerian and Lagrangian approach. Decomposing the velocity field into rotational and divergent components, we identify how each component affects the actual circulation. We propose an alternative definition for the circulation, which describes the actual path of air parcels in the tropics, as a tropical atmospheric conveyor belt.

Jupiter's internal flow structure is still not fully known, but can be now better constrained due to Juno's highprecision measurements. The recently published gravity and magnetic field measurements have led to new information regarding the planet and its internal flows, and future magnetic measurements will help to solve this puzzle. In this study, we propose a new method to better constrain Jupiter's internal flow field using the Juno gravity measurements combined with the expected measurements of magnetic secular variation. Based on a combination of hydrodynamical and magnetic field considerations we show that an optimized vertical profile of the zonal flows that fits both measurements can be obtained. Incorporating the magnetic field effects on the flow better constrains the flow decay profile. This will get us closer to answering the persistent question regarding the depth and nature of the flows on Jupiter.

One of Jupiter's most prominent atmospheric features, the Great Red Spot (GRS), has been observed for more than two centuries, yet little is known about its structure and dynamics below its observed cloud level. While its anticyclonic vortex appearance suggests it might be a shallow weather-layer feature, the very long time span for which it was observed implies it is likely deeply rooted, otherwise it would have been sheared apart by Jupiter's turbulent atmosphere. Determining the GRS depth will shed light not only on the processes governing the GRS, but on the dynamics of Jupiter's atmosphere as a whole. The Juno mission single flyby over the GRS (PJ7) discovered using microwave radiometer measurements that the GRS is at least a couple hundred kilometers deep. The next flybys over the GRS (PJ18 and PJ21), will allow high-precision gravity measurements that can be used to estimate how deep the GRS winds penetrate below the cloud level. Here we propose a novel method to determine the depth of the GRS based on the new gravity measurements and a Slepian function approach that enables an effective representation of the wind-induced spatially confined gravity signal, and an efficient determination of the GRS depth given the limited measurements. We show that with this method the gravity signal of the GRS should be detectable for wind depths deeper than 300 km, with reasonable uncertainties that depend on depth (e.g., +/- 100 km for a GRS depth of 1000 km).

The depth to which Jupiter's observed east-west jet streams extend has been a long-standing question(1,2). Resolving this puzzle has been a primary goal for the Juno spacecraft(3,4), which has been in orbit around the gas giant since July 2016. Juno's gravitational measurements have revealed that Jupiter's gravitational field is north-south asymmetric(5), which is a signature of the planet's atmospheric and interior flows(6). Here we report that the measured odd gravitational harmonics J(3), J(5), J(7) and J(9) indicate that the observed jet streams, as they appear at the cloud level, extend down to depths of thousands of kilometres beneath the cloud level, probably to the region of magnetic dissipation at a depth of about 3,000 kilometres(7,8). By inverting the measured gravity values into a wind field(9), we calculate the most likely vertical profile of the deep atmospheric and interior flow, and the latitudinal dependence of its depth. Furthermore, the even gravity harmonics J(8) and J(10) resulting from this flow profile also match the measurements, when taking into account the contribution of the interior structure(10). These results indicate that the mass of the dynamical atmosphere is about one per cent of Jupiter's total mass.

The gravity harmonics of a fluid, rotating planet can be decomposed into static components arising from solid-body rotation and dynamic components arising from flows. In the absence of internal dynamics, the gravity field is axially and hemispherically symmetric and is dominated by even zonal gravity harmonics J 2n that are approximately proportional to q n, where q is the ratio between centrifugal acceleration and gravity at the planet's equator. Any asymmetry in the gravity field is attributed to differential rotation and deep atmospheric flows. The odd harmonics, J 3, J 5, J 7, J 9 and higher, are a measure of the depth of the winds in the different zones of the atmosphere. Here we report measurements of Jupiter's gravity harmonics (both even and odd) through precise Doppler tracking of the Juno spacecraft in its polar orbit around Jupiter. We find a north-south asymmetry, which is a signature of atmospheric and interior flows. Analysis of the harmonics, described in two accompanying papers, provides the vertical profile of the winds and precise constraints for the depth of Jupiter's dynamical atmosphere.

The Cassini measurements of Saturn's gravity field during its Grand Finale might shed light on a long-standing question regarding the flow on Saturn. While the cloud-level winds are well known, little is known about whether these winds are confined to the outer layers of the planet or penetrate deep into the interior. An additional complexity is added by the uncertainty in the exact rotation period of Saturn, a key factor in determining the cloudlevel winds, with an effect on the north-south symmetric part of the winds. Using Saturn's cloud-level winds we relate the flow to the gravity harmonics. We give a prediction for the odd harmonicsJ(3), J(5), J(7), and J(9) as a function of the flow depth, identifying three ranges of depths. Since the odd harmonics depend solely on the flow, and are not influenced by Saturn's shape and static density distribution, any measured value of the odd harmonics by Cassini can be used to uniquely determine the depth of the flow. We also discuss the flow-induced even harmonics Delta J(2), Delta J(4),...., Delta J(12) that are affected by Saturn's rotation period. While the high-degree even harmonics might also be used to determine the flow depth, the lower-degree even harmonics serve as uncertainties for analysis of the planet's interior structure and composition. Thus, the gravity harmonics measured during the Cassini Grand Finale may be used to get a first-order estimate of the flow structure and to better constrain the planet's density structure and composition.

The close-by orbits of the ongoing Juno mission allow measuring with unprecedented accuracy Jupiter's low-degree even gravity moments J₂, J₄, J₆, and J₈. These can be used to better determine Jupiter's internal density profile and constrain its core mass. Yet the largest unknown on these gravity moments comes from the effect of differential rotation, which gives a degree of freedom unaccounted for by internal structure models. Here considering a wide range of possible internal flow structures and dynamical considerations, we provide upper bounds to the effect of dynamics (differential rotation) on the low-degree gravity moments. In light of the recent Juno gravity measurements and their small uncertainties, this allows differentiating between the various models suggested for Jupiter's internal structure.

Observations of the flow on Jupiter exists essentially only for the cloud-level, which is dominated by strong east-west jet-streams. These have been suggested to result from dynamics in a superficial thin weather-layer, or alternatively be a manifestation of deep interior cylindrical flows. However, it is possible that the observed wind is indeed superficial, yet there exists a completely decoupled deep flow. To date, all models linking the wind, via the induced density anomalies, to the gravity field, to be measured by Juno, consider only flow that is a projection of the observed cloud-level wind. Here we explore the possibility of complex wind dynamics that include both the shallow weather-layer wind, and a deep flow that is decoupled from the flow above it. The upper flow is based on the observed cloud-level flow and is set to decay with depth. The deep flow is constructed to produce cylindrical structures with variable width and magnitude, thus allowing for a wide range of possible scenarios for the unknown deep flow. The combined flow is then related to the density anomalies and gravitational moments via a dynamical model. An adjoint inverse model is used for optimizing the parameters controlling the setup of the deep and surface-bound flows, so that these flows can be reconstructed given a gravity field. We show that the model can be used for examination of various scenarios, including cases in which the deep flow is dominating over the surface wind, and discuss the uncertainties associated with the model solution. The flexibility of the adjoint method allows for a wide range of dynamical setups, so that when new observations and physical understanding will arise, these constraints could be easily implemented and used to better decipher Jupiter flow dynamics. (C) 2017 Elsevier Inc. All rights reserved.

The upcoming Juno and Cassini gravity measurements of Jupiter and Saturn, respectively, will allow probing the internal dynamics of these planets through accurate analysis of their gravity spectra. To date, two general approaches have been suggested for relating the flow velocities and gravity fields. In the first, barotropic potential surface models, which naturally take into account the oblateness of the planet, are used to calculate the gravity field. However, barotropicity restricts the flows to be constant along cylinders parallel to the rotation axis. The second approach, calculated in the reference frame of the rotating planet, assumes that due to the large scale and rapid rotation of these planets, the winds are to leading order in geostrophic balance. Therefore, thermal wind balance relates the wind shear to the density gradients. While this approach can take into account any internal flow structure, it is limited to only calculating the dynamical gravity contributions, and has traditionally assumed spherical symmetry. This study comes to relate the two approaches both from a theoretical perspective, showing that they are analytically identical in the barotropic limit, and numerically, through systematically comparing the different model solutions for the gravity harmonics. For the barotropic potential surface models we employ two independent solution methods - the potential-theory and Maclaurin spheroid methods. We find that despite the sphericity assumption, in the barotropic limit the thermal wind solutions match well the barotropic oblate potential-surface solutions.

Jupiter's Great Red Spot (GRS) is the most dominant and long-lived feature in Jupiter's atmosphere. However, whether this is a shallow atmospheric feature or a deeply rooted vortex has remained an open question. In this study, we assess the possibility of inferring the depth of the GRS by the upcoming Juno gravity experiment. This is achieved by an exploration of the possible gravitational signature of the vortex by systematically extending the surface winds into the interior and analyzing the resulting gravity signal. The gravity anomaly is then compared to the expected accuracy in the retrieval of the surface gravity at the GRS location obtained with numerical simulations of the Doppler data inversion based on the expected trajectory of the spacecraft. Starting from observations of the atmospheric velocity at the cloud level, we project the wind using a decay scale height along coaxial cylinders parallel to the spin axis and explore a wide range of decay scale heights in the radial direction. Assuming the large scale vortex dynamics are geostrophic, and therefore thermal wind balance holds, the density anomaly distribution due to Jupiter's winds can be derived from the velocity maps. The novelty of this approach is in the integration of thermal wind relations over a three-dimensional grid, and in the inclusion of the observed meridional velocity as measured during the Cassini flyby of Jupiter. The perturbations in the mean zonal flow give rise to non zero tesseral spherical harmonics in Jupiter's gravitational potential. We provide an estimate of this asymmetric gravity coefficients for different values of the wind decay scale height. We conclude that the mass anomaly associated with the GRS is detectable by the Juno gravity experiment if the vortex is deep, characterized by a vertical height larger than 2,000 km below the cloud level of Jupiter, and that the large mass involved with deep winds does not render much the ability to measure the feature.

There are many sources of very low frequency (VLF3-30kHz) and extremely low frequency (ELF3-3000Hz) radiation in the Earth-ionosphere waveguide (e.g., lightning and ELF/VLF communication transmitters). At distances of thousands of kilometers from these sources, the vertical component of the ELF/VLF AC magnetic fields is expected to be very weak and several orders of magnitude lower than the horizontal magnetic components. However, measurements in Israel show a relatively strong vertical magnetic component in both the ELF and VLF bands, at the same order of magnitude as the horizontal components. Our measurements suggest that the real Earth-ionosphere waveguide might often be very different from the theoretical waveguide used in model calculations. In addition, our results imply that using only the horizontal components for direction finding or the absolute magnetic field strength may result in errors, since often a significant fraction of the magnetic field energy hides in the vertical component.

Estimates of global thunderstorm activity have been made predominately by direct measurements of lightning discharges around the globe, either by optical measurements from satellites, or using ground-based radio antennas. In this paper we propose a new methodology in which thunderstorm clusters are constructed based on the lightning strokes detected by the World Wide Lightning Location Network (WWLLN) in the very low frequency range. We find that even with low lightning detection efficiency on a global scale, the spatial and temporal distribution of global thunderstorm cells is well reproduced. This is validated by comparing the global diurnal variations of the thunderstorm cells, and the currents produced by these storms, with the well-known Carnegie Curve, which represents the mean diurnal variability of the global atmospheric electric circuit, driven by thunderstorm activity. While the Carnegie Curve agrees well with our diurnal thunderstorm cluster variations, there is little agreement between the Carnegie Curve and the diurnal variation in the number of lightning strokes detected by the WWLLN. When multiplying the number of clusters we detect by the mean thunderstorm conduction current for land and ocean thunderstorms (Mach et al 2011 J. Geophys. Res. 116 D05201) we get a total average current of about 760 A. Our results show that thunderstorms alone explain more than 90% in the variability of the global electric circuit. However, while it has been previously shown that 90% of the global lightning occurs over continental landmasses, we show that around 50% of the thunderstorms are over the oceans, and from 00-09UTC there are more thunderstorm cells globally over the oceans than over the continents. Since the detection efficiency of the WWLLN system has increased over time, we estimate that the lower bound of the mean number of global thunderstorm cells in 2012 was around 1050 per hour, varying from around 840 at 03UTC to 1150 storms at 19UTC.

One of the costliest natural hazards around the globe is flash floods, resulting from localized intense convective precipitation over short periods of time. Since intense convective rainfall (especially over the continents) is well correlated with lightning activity in these storms, a European Union FP6 FLASH project was realized from 2006 to 2010, focusing on using lightning observations to better understand and predict convective storms that result in flash floods. As part of the project, 23 case studies of flash floods in the Mediterranean region were examined. For the analysis of these storms, lightning data were used together with rainfall estimates in order to understand the storms' development and electrification processes. In addition, these case studies were simulated using mesoscale meteorological models to better understand the local and synoptic conditions leading to such intense and damaging storms. As part of this project, tools for short-term predictions (nowcasts) of intense convection across the Mediterranean and Europe, and long-term forecasts (a few days) of the likelihood of intense convection, were developed and employed. The project also focused on educational outreach through a special Web site http://flashproject.org supplying real-time lightning observations, real-time experimental nowcasts, medium-range weather forecasts and educational materials. While flash floods and intense thunderstorms cannot be prevented, long-range regional lightning networks can supply valuable data, in real time, for warning the public, end-users and stakeholders of imminent intense rainfall and possible flash floods.

Thunderstorms are often the cause of severe and disastrous flash floods. Lightning activity within these storms can be detected and monitored continuously from thousands of kilometers away and can therefore be very useful in improving forecasts and now-casts of severe thunderstorms. An improvement in the now-casting of such storms can assist in reducing damages and saving lives.Using the ZEUS ground-based VLF lightning detection network and the Warning Decision Support System-Integrated Information (WDSS-II) software, we performed now-casting simulations using 1 year of lightning data over the Mediterranean area. Thousands of thunderstorms were observed and now-cast 30, 60, 90 and 120 min ahead. Statistical analysis was then done by calculating the hit, miss and false-alarm rates, as well as the POD, FAR and CSI scores in order to determine the success of the now-casting.The results show that the algorithm is overall successful in now-casting the location of the lightning clusters, especially when applied to strong and consistent lightning events (it is these events which also have the strongest connection to flash floods). The probability of detection (POD) values range between 0.46 for 30 minute now-casts and 0.25 for 120 minute now-casts. The critical success index (CSI) values are quite similar, but slightly lower. The now-casting has a low false-alarm rate, 0.03 for 30 minute now-casts, which is also beneficial for operational use.This method has been implemented for the use in real-time now-casting and is used in the EU FLASH project to seek and track areas of thunderstorm risk according to lightning intensity. The experimental now-casts appear on the project website (www.flashproject.org/). (C) 2010 Elsevier B.V. All rights reserved.

An experimental ENSO prediction system is presented, based on an ocean general circulation model (GCM) coupled to a statistical atmosphere and the adjoint method of 4D variational data assimilation. The adjoint method is used to initialize the coupled model, and predictions are performed for the period 1980-99. The coupled model is also initialized using two simpler assimilation techniques: forcing the ocean model with observed sea surface temperature and surface fluxes, and a 3D variational data assimilation (3DVAR) method, similar to that used by the National Centers for Environmental Prediction (NCEP) for operational ENSO prediction. The prediction skill of the coupled model initialized by the three assimilation methods is then analyzed and compared. The effect of the assimilation period used in the adjoint method is studied by using 3-,6-, and 9-month assimilation periods. Finally, the possibility of assimilating only the anomalies with respect to observed climatology in order to circumvent systematic model biases is examined. It is found that the adjoint method does seem to have the potential for improving over simpler assimilation schemes. The improved skill is mainly at prediction intervals of more than 6 months, where the coupled model dynamics start to influence the model solution. At shorter prediction time intervals, the initialization using the forced ocean model or the 3DVAR may result in a better prediction skill. The assimilation of anomalies did not have a substantial effect on the prediction skill of the coupled model. This seems to indicate that in this model the climatology bias, which is compensated for by the anomaly assimilation, is less significant for the predictive skill than the bias in the model variability, which cannot be eliminated using the anomaly assimilation. Changing the optimization period from 6 to 3 to 9 months showed that the period of 6 months seems to be a near-optimal choice for this model.

One of the major factors determining the strength and extent of ENSO events is the instability state of the equatorial Pacific coupled ocean-atmosphere system and its seasonal variations. This study analyzes the coupled instability in a hybrid coupled model of the Indo-Pacific region, using the adjoint method for sensitivity studies. It is found that the seasonal changes in the ocean-atmosphere instability strength in the model used here are related to the outcropping of the thermocline in the east equatorial Pacific. From July to December, when the thermocline outcrops over a wide area in the east Pacific, there is a strong surface-thermocline connection and anomalies that arrive as Kelvin waves from the west along the thermocline can reach the surface and affect the SST and thus the coupled system. Conversely, from February to June, when the thermocline outcropping is minimal, the surface decouples from the thermocline and temperature anomalies in the thermocline depth range do not affect the surface and dissipate within the thermocline. The role of vertical mixing rather than upwelling in linking vertical thermocline movements to SST change is emphasized. It is therefore suggested that the seasonal ocean-atmosphere instability strength in the equatorial Pacific is strongly influenced by the thermocline outcropping and its seasonal modulation, a physical mechanism that is often neglected in intermediate coupled models and that can be represented properly only in models that employ the full dynamics of the mixed layer.