JAMSTEC > Research Institute for Value-Added-Information Generation (VAiG) > Application Laboratory (APL) > Seminar Schedule

Application Laboratory

Seminar Schedule

59th APL Open Forum

Date & Time:
20 December, Friday, 14:30-16:00
Place:
Meeting Room, 6th floor, IT Building, Yokohama Institute
Speaker 1:
Hideharu Sasaki
Title:
Interannual to decadal variations of winter active submesoscale motions around the Notrh Pacific Subtropical Countercurrent
Abstract:
Seasonal variations of submesoscale motions in the open oceans are active in winter and calm in summer, which was suggested by the high-resolution simulations and then confirmed by the ADCP and satellite observations. The submesoscale motions are enhanced in winter by the baroclinic instability within the mixed layer called mixed layer instability (MLI). However, there are no data to examine the submesoscale variations with the timescales longer than interannual as far as I know.
The North Pacific OFES2 hindcast at the horizontal resolution of 1/30 degree has been conducted from 1990 to 2016. The active submesoscale motions in winter around the North Pacific Subtropical Countercurrent vary at the interannual timescale. The variations are induced by those of the potential energy release defined by the mixed layer depth and horizontal buoyancy gradient, which converts the potential energy to kinetic energy at the submesoscale in the process of the MLI. The composite distributions of surface temperature anomaly in the years of active and calm submesoscale motions in winter are similar to those of the positive and negative PDO respectively. The 5-year running means of submesoscale motions in winter vary synchronously with those of PDO indexes.
Speaker 2:
Yuya Baba
Title:
Improved MJO simulations by three GCMs using a spectral cumulus parameterization
Abstract:
A spectral cumulus parameterization (spectral scheme) was implemented in three GCMs (AFES, ICON-A, and SINTEX-F2) and the validity for simulating MJO was examined. In the case of AFES, a shallow convective closure was additionally examined. The model successfully simulated MJO when the shallow convective closure was adopted. The impacts on the climatology and variability were investigated. The reason for the improved simulation was found to be the enhanced contribution of shallow convection. ICON-A with the spectral scheme outperformed the original model (with Tiedtke scheme) in terms of the MJO. This was found to be due the spectral scheme was able to simulate continuous entrainment profiles during the development of organized convection, whereas the original convection scheme failed to simulate the feature. In the case of SINTEX-F2, the spectral scheme successfully simulated MJO related variability during 20 years simulation. A quantitative analysis for MJO showed that the spectral scheme presented higher performance for MJO. However, further tuning was needed for climatology of SINTEX-F2 with the spectral scheme.

58th APL Open Forum

Date & Time:
29 November, Friday, 14:30-16:00
Place:
Meeting Room, 6th floor, IT Building, Yokohama Institute
Speaker 1:
Masami Nonaka
Title:
Predictability of interannual variability in eddy activity in the Kuroshio Extension
Abstract:
Eddies in the Kuroshio Extension have been shown to affect the atmosphere aloft and also oceanic ecosystem. It is therefore important to improve our understanding of the mechanisms and predictability of variability in their activity. For that purpose, we have analyzed a ten-member ensemble, of an eddy-resolving OGCM. Focusing on interannual-to-decadal variability by the 13-month running mean, eddy activity in the downstream Kuroshio Extension region (32-38N, 153-165E) shows rather limited ensemble spread and ensemble mean has significant correlation with the observation (correlation coefficient r=0.58). Also, the eddy activity variability highly correlates with the local current speed variability that propagates westward from the central North Pacific, inducing predictability of the eddy activity with several years lead time. Similar predictability is also found in the observed data, although it is not statistically robust due to limited data period. In contrast, in the upstream Kuroshio Extension region (32-38N, 141-153E), the spread has amplitude similar to that of interannual variability. While possibly model dependent, thus result implies that the observed eddy activity in the upstream Kuroshio Extension may have a significant intrinsic component on the interannual time scale, and its predictability is limited.
Speaker 2:
Ingo Richter
Title:
A first look at the performance of CMIP6 models in the tropical Atlantic
Abstract:
Output from the Coupled Model Intercomparison Project Phase 6 (CMIP6) is increasingly becoming available. While the archive is still far from complete it now contains output from more than 30 general circulation models (GCMs). In the present study I evaluate the pre-industrial control simulations (piControl) from 34 models in terms of their ability to reproduce mean state and variability of the tropical Atlantic.The severe SST biases along the equatorial Atlantic remain a serious problem in CMIP6, with the average over the model ensemble showing little improvement over the previous generation (CMIP5). There are, however, two GCMs that have only small SST biases in the equatorial Atlantic, one from the Hadley Centre in the UK, the other from the Institut Pierre Simon Laplace in France. Analysis of the origin of the SST biases yields a similar picture as for previous model generations: an erroneous southward shift of the inter-tropical convergence zone (ITCZ) in boreal spring is accompanied by a westerly wind bias over the equator. This deepens the thermocline in the east and decreases upwelling efficiency in the following months. Despite their severe mean state biases, some models produce quite realistic variability patterns. Conversely, some models with a rather realistic mean state, display poor phase locking and amplitude of equatorial variability. The results indicate that model biases in the tropical Atlantic will persist for the foreseeable future. Despite these biases, interannual variability can be reasonably well represented. It remains an open question, however, whether mean state biases significantly affect global change projections for the region.

57th APL Open Forum

Date & Time:
25 October, Friday, 13:30-15:00
Place:
Meeting Room, 5th floor, IT Building, Yokohama Institute
Speaker 1:
Miho Ishizu
Title:
Mechanisms of seasonal variability of carbon cycle in the North Western Pacific: a biogeochemical and carbon modeling study coupled with an operational ocean model product
Abstract:
To understand mechanisms of seasonal carbon cycle in the North Western Pacific, we examined a high-resolution biological and carbon model coupled with an operational ocean model outputs. The model showed that air-sea CO2 exchange process were mainly balanced with the vertical mixing at the surface throughout the latitudes, but the detailed feature of the balances seasonally changed. The surface patterns in the subarctic region were contrast to that in the subtropical region, and biogeochemical processes were evident at the surface in the subarctic region. The active layer of the biological processes and their resulting DIC consumption/production deepened to the south with the euphotic layer. The situation influences the long term tendency of the carbon cycle and could influence the regional differences of ocean acidification in the North Western Pacific.
Speaker 2:
Tsutomu Hihara
Title:
Constructing the JCOPE-T DA and JCOPE-T 1km model
Abstract:
We have run operationally the data assimilation product name as "JCOPE-T DA" from Nov. 2018. The temporal and spatial resolution of JCOPE-T DA is hourly and 3km. And, this product uses the tide considering model. We assimilate directory the satellite sea surface height data, the GTS in situ data, and the satellite sea surface temperature data (Himawari-8, AMSR2, GMI, and WindSAT) provided by JAXA. The outputs of JCOPE-T DA is opened to the public on the "JAXA Himawari Monitor" website.Now, we are constructing the higher spatial resolution (1/120˚ grid) model (JCOPE-T 1km model) which is the same domain to JCOPE-T DA. In this model, we are going to use near-real-time hourly river discharge data calculated by the CaMa-Flood model instead of monthly climate data. We have provided the JCOPE-T DA and 1km outputs to the JAPAN SAILING FEDERATION(JSAF) for the sailing games in the Tokyo Olympics which will be held in the east-north area of Sagami bay. So, we provided test outputs (about 1 month) of JCOPE-T 1km for the pre-competition in Aug. 2019. We compared the outputs with HF reader observation data provided by the Japan Coast Guard.

JAMSTEC-IPRC Collaboration Research (JICoRe) Seminars
hosted by APL, DCOP and CEMA

Part I

Date & Time:
1 October, Tuesday, 10:00-12:00
Place:
Room 1+2, 2nd floor, Conference Building, Yokohama Institute
Speaker 1:
Prof. Kelvin Richards (International Pacific Research Center and Department of Oceanography, University of Hawai'i at Manoa, Honolulu, HI, USA)
Title:
Enhanced turbulent mixing in the equatorial thermocline
Abstract:
Enhanced mixing caused by small vertical scale features (SVSs) in the equatorial thermocline is known to impact the state of the ocean and its interaction with the atmosphere, in particular the sea temperature of the Pacific cold tongue and ENSO variability. The SVSs are produced by wind variability and instabilities. The good news is that with enough resolution these features can be captured in both observations and models. From observations we show that the vertical distribution of turbulent activity in the thermocline is very dependent on the turbulent length scale. From models we show that inertial and parametric subharmonic instability play a role and that wind driven inertia-gravity waves lead to an enhancement in mixing by a combination of three factors: a stronger superinertial component of the wind forcing close to the equator, wave action convergence at turning latitudes for equatorially trapped waves, and nonlinear wave-wave interactions between equatorially trapped waves. Using a combination of ideal models and an OGCM we investigate the properties of SVS activity and its impact on mixing. Of particular interest is the dependency on stratification, the spatial and temporal variability of wind forcing and model resolution (both vertical and horizontal). The impact of the spatially and temporally varying mixing on the seasonal and interannual variability of the Pacific with be discussed. Such knowledge is invaluable in the planning of future observational studies and the design of the next generation climate models.
Speaker 2:
Prof. Niklas Schneider (International Pacific Research Center and Department of Oceanography, University of Hawai'i at Manoa, Honolulu, HI, USA)
Title:
Scale and Rossby number dependences of observed wind responses to ocean-mesoscale sea surface temperatures
Abstract:
The horizontal-scale dependences of in-phase and lagged imprints of ocean-mesoscale sea surface temperatures on surface winds are investigated using daily AMSR-E radiometer and QuikSCAT scatterometer observations in the Southern Ocean. Spectral transfer functions separate underlying processes dependent on large-scale winds, horizontal wave-numbers and corresponding Rossby numbers. For Rossby numbers smaller than one, winds reflect modulations of the Ekman layer by sea surface temperature induced changes of hydrostatic pressure. Rossby numbers large compared to one suggest a balance of advection and modulations of vertical mixing. Corresponding impulse response functions reveal Doppler-shifted near-inertial lee waves excited by enhanced vertical mixing over warm sea surface temperature perturbations that interact with the downstream wake of warm air and low atmospheric pressure. Inertial turning of winds and hydrostatic pressure gradient forces strengthen the downwind convergence relative to the upwind divergence, and enhance the anticyclonic wind curl through near resonance on the right side of the wake. Time averaging smooths the response, and degrades the lee-wave.

Part II

Date & Time:
4 October, Friday, 10:30-11:30
Place:
Room 1+2, 2nd floor, Conference Building, Yokohama Institute
Speaker 3:
Prof. H. Annamalai (International Pacific Research Center and Department of Oceanography, University of Hawai'i at Manoa, Honolulu, HI, USA)
Title:
Atmospheric rivers during boreal summer seasonn

56th APL Open Forum

Date & Time:
6 September, Friday, 14:30-16:00
Place:
Meeting Room 1+2, 2nd floor, Conference Building, Yokohama Institut
Speaker 1:
J. Venkata Ratnam
Title:
Statistical prediction of Indian Ocean Dipole index
Abstract:
In this study an attempt is made to predict the Indian Ocean Dipole (IOD) index using statistical techniques such as linear regression and artificial neural networks. The predictors are identified from the sea surface temperature, geopotential height, zonal wind and meridional wind fields variables for the period spanning from 1949 to 2018. The results indicate the statistical models to be capable of forecasting the IOD index in advance with high skill.
Speaker 2:
Tomomichi Ogata
Title:
Mechanisms of long-term variability and recent trend of salinity along 137°E
Abstract:
To investigate mechanisms for observed features along 137°E such as freshening trend since the 1990s and decadal variability, spatial pattern of salinity anomaly at 25.4σ isopycnal surface and its variability were investigated based on OGCM outputs. While sea surface salinity (SSS) is restored to its climatology in the OGCM, the model captures recent salinity trend and meridionally coherent interannual to decadal variability at 137°E. The trend signal seems to be traced back over the eastern North Pacific (around 30°N, 160°W). In spite of the limited SSS variability, meridional shift of outcrop line caused by sea surface temperature variation is found to determine the decadal spiciness variability subducting on the isopycnal surface at 25.4σ, that is, warm-saltier anomalies with southward shift, and cold-fresh anomalies with northward shift as colder-fresher water distributes to the north at the sea surface. Furthermore, we conducted tracer experiments with/without mesoscale eddies to examine possible roles of mesoscale eddies in propagation of the spiciness anomalies. Tracer diffuses across mean stream line by mesoscale eddies, and spreads from 10°N to 25°N. However, in tracer experiment with steady mean flow (i.e. without mesoscale eddies), center of tracer propagates along mean stream line and does not diffuse across mean stream line. This suggests that mesoscale eddies are important for the synchronized salinity trend and variability found in 10°N to 30°N.

55th APL Open Forum

Date & Time:
2 August, Friday, 13:30-15:00
Place:
Meeting Room 1+2, 2nd floor, Conference Building, Yokohama Institute
Speaker 1:
Ayako Yamamoto
Title:
Key role of western boundary currents in wintertime Euro-Atlantic blocking
Abstract:
Atmospheric blocking events are well known for their crucial role in modulating the mid-latitude subseasonal atmospheric variability. The detailed physical and dynamical mechanisms responsible for the formation and maintenance of blocking events, however, are yet to be fully understood, and state-of-the-art climate models still have deficiency in accurately simulating their observed occurrence. Conventionally, atmospheric blocking was considered in the dry dynamics framework. Recent studies, nonetheless, attributed some of the modelled blocking frequency deficit to biases in the modelled ocean, implying the importance of the air-sea exchange of heat and moisture in the formation of the blocking events. Furthermore, a crucial role of condensational warming in the blocking formation and maintenance has recently been highlighted. The source of heat and moisture associated in this process, however, is yet to be identified. In this study, we aim to identify the sources of the atmospheric heat and moisture involved in wintertime Euro-Atlantic blocking events using a Lagrangian approach driven with CFSR. Using an atmospheric dispersion model, we track atmospheric particles from the blocking centres backward in time for 10 days, and estimate the associated turbulent heat fluxes whenever the particles fall within the planetary boundary layer over the ocean. Our results indicate that approximately half of the particles released from the blocking centres receive moisture and heat from the western boundary currents and their extensions, which then undergo latent warming while ascending, consistent with the previous studies. We found that the potential vorticity along these particles which undergo diabatic processes is more strongly anticyclonic compared to the adiabatic particles, implying the central role of the diabatic processes in blocking events.
Speaker 2:
Takeshi Doi
Title:
Westerly Wind Burst/Easterly Wind Surge-like stochastic forcing and its potential roles to remedy the over-confident problem of ENSO prediction ~ Lesson from failure of 2014 El Niño prediction ~
Abstract:
Over-confident problem of ENSO prediction: the ensemble spread is too small relative to the root mean square error of the ensemble mean prediction, is common among almost all dynamical prediction systems based on GCMs. That problem was clearly exposed in 2014; most of the dynamical prediction systems had alerted high chance of El Niño occurrence in the boreal summer, however the prediction failed. In this study, it is found that the high-frequency zonal wind variability over the western tropical Pacific, including the strong easterly wind surges in June-July 2014, is responsible for failure of 2014 El Niño prediction by the SINTEX-F2 seasonal prediction system. It is almost impossible to predict individual high-frequency wind events beyond the weather prediction limitation, because the occurrences are not determined by the sea surface temperature-state. However, by adding easterly wind surge-like stochastic forcing into the prediction system, we have successfully fallen the 2014 El Niño prediction within the expected uncertainty. It is also shown that the additional stochasticity of westerly wind burst and/or easterly wind surge could effectively remedy the over-confident of ENSO prediction on the condition that the Pacific Warm Pool extended further eastward. The presented results are useful to reduce false alarm of ENSO prediction, and thus its related possible human and economic losses.

54th APL Open Forum

Date & Time:
28 June, Friday, 14:30-16:00
Place:
Meeting Room, 5th floor, IT Building, Yokohama Institute
Speaker 1:
Kunihiro Aoki
Title:
Application of Machine learning for Probabilistic Forecast of Kuroshio Large Meander
Abstract:
This study proposes a method of the probabilistic forecast, in which the solutions of an ensemble forecasts with 80 members are clustered by the mixture Gaussian distribution model (MGM), an unsupervised machine learning technique. Targeting the period before the development of the Kuroshio large meander in 2017, we conduct the ensemble forecast in a long-range initialized by the assimilation based on the Local Ensemble Kalman Filter algorithm applied for the forecasts spun up from the different initial conditions. All the members show the non-large meander path almost along the coast in the initial stage, but finally some members show the large meander path at the end date. Applying MGM for a traditional one-dimensional Kuroshio large meander index successfully identifies the bistates with correponding uncertainties and occurrence probabilities. The statistical significance of the bimodality can be evaluated by information criteria. With the help of the empirical orthogonal function (EOF) analysis, the application of this clustering method can be extended to the sea surface height field, which has the dimension as large as the number of horizontal model grids. We found that the bimodal nature appears only on the axis of the first EOF mode in the phase space even for such a large dimensional data. From both a theoretical argument and data analysis, we also show that the first EOF mode is almost equivalent to the first singular vector associated with the tangential operator of the equation for the sea surface height. This suggests that the bistates found here are sets of non-linearly stable solutions evolved through a linear growth at an initial stage.
Speaker 2:
Nobumasa Komori
Title:
Experimental seasonal climate prediction using CFES: Comparison with the SINTEX-F systems
Abstract:
An experimental seasonal climate prediction system has been developed based on the Coupled atmosphere-ocean general circulation model for the Earth Simulator (CFES). Following the well-established system based on the SINTEX-F model, initial conditions for seasonal climate prediction are constructed by strongly nudging sea surface temperature (SST) to observed one.
At this stage, 12-member ensemble 6-month predictions from the 1st day of each month have been conducted from 1983 through 2018, after 32-years of coupled spin-up integration with SST-nudging to the observed climatology. The experimental system exhibits skill in predicting variability of seasonal-mean 2-m air temperature over the tropical Pacific, and contributes to the improvement of multi-model ensemble prediction with the SINTEX-F systems.

APL Guest Seminar

Date & Time:
3 June, Monday, 14:30-15:30
Place:
Meeting Room 1+2, 2nd floor, Conference Building, Yokohama Institute
Speaker:
Dr. Dmitri Kondrashov (University of California at Los Angeles, USA)
Title:
Data-driven climate modeling and real-time prediction
Abstract:
Climate dynamics is fundamentally nonlinear, high-dimensional and extremely complex. General circulation climate models (GCMs) based on first principles, are typically expensive to run and are subject to various biases due to imperfect physical parameterizations of unresolved physical processes. These shortcomings in turn are affecting accuracy of seasonal-to-intraseasonal prediction of key climate phenomena with global social-economic impacts, such as El-Nino–Southern Oscillation (ENSO) and summertime Arctic Sea Ice Extent (SIE).
This talk will highlight recent advances in alternative data-driven methodologies to help analyze and model high-dimensional observational datasets. Real-time predictions in multi-model ensembles demonstrate that data-driven forecasts of ENSO and Arctic SIE currently perform just as well as those performed with GCMs but at a fraction of costs.

53th APL Open Forum

Date & Time:
24 May, Friday, 14:30-16:00
Place:
Meeting Room, 5th floor, IT Building, Yokohama Institute
Speaker 1:
Yu-Lin (Eda) Chang
Title:
Tropical and subtropical tropical cyclone formation in the western Pacific and the prediction based on seasonal forecasting model
Abstract:
ENSO is considered to be the leading factor in affecting tropical cyclone (TC) activity in the Pacific, and it has been widely used in the statistical model in predicting TC activity. The recent study noticed the importance of the local low-level wind curl, which revealed a higher correlation to TC formation in comparison to ENSO indices. This study explores the air-sea coupled response and its link to subtropical and tropical TC formation in the western Pacific. The results suggest that the formation of subtropical and tropical TC formation was affected by the local low-level wind curl, which was trigger by the eastern and central Pacific warming/cooling, respectively. The dynamic will then be applied using the seasonal forecasting model results from SINTEX-F to examine the retrospective and future prediction.
Speaker 2:
Yushi Morioka
Title:
Role of sea-ice initialization in climate predictability over the Weddell Sea
Abstract:
Potential influence of sea-ice cover initialization on the interannual climate predictability over the Weddell Sea is identified using a coupled general circulation model (SINTEX-F2). Climate variability in the Weddell Sea is generally believed to have association with remote forcing such as El Niño-Southern Oscillation and the Southern Annual Mode. However, sea-ice variability in the Weddell Sea has recently been suggested to play additional roles in modulating local atmospheric variability through changes in surface air temperature and near-surface baroclinicity. When both the model’s sea-surface temperature (SST) and sea-ice concentration (SIC) are initialized with observations using nudging schemes, reforecast experiments from September 1st of 1982-2016 show improvements in predicting the observed SIC anomalies in the Weddell Sea up to four months ahead, compared to the other experiments with only SST initialization. During austral spring (Oct-Dec) of lower-than-normal sea-ice years in the Weddell Sea, reforecast experiments with the SST and SIC initializations reasonably predict high surface air temperature anomalies in the Weddell Sea and high sea-level pressure anomalies over the Atlantic sector of the Southern Ocean. These results suggest that accurate initialization of sea-ice conditions during austral winter is necessary for skillful prediction of climate variability over the Weddell Sea during austral spring.

APL Guest Seminar

Date & Time:
8 May, Wednesday, 14:00-15:00
Place:
Meeting Room 1+2, 2nd floor, Conference Building, Yokohama Institute
Speaker:
Dr. Andréa Sardinha Taschetto (Climate Change Research Centre, University of New South Wales, Australia)
Title:
The contribution of ocean versus atmospheric variability to droughts in the tropics
Abstract:
Variations in ocean temperature, such as the El Niño – Southern Oscillation (ENSO), drive fluctuations in rainfall extremes, and are generally associated with droughts and wet periods over land. Droughts however can occur due to causes unrelated to ocean temperature, for example, as a result of intrinsic atmospheric and land variability. Disentangling the contribution of each climate component to drought frequency, intensity and persistence is a complicated task using observations. We assess the contribution of sea surface temperature variability to droughts over land within a state-of-the-art climate model. The National Center for Atmospheric Research (NCAR) Community Earth System Model (CESM) is used to perform multi-century numerical experiments where ocean variability is eliminated from simulation. Results show that interannual variations of sea surface temperature play an important role in increasing rainfall variability everywhere in the globe. However, multiyear drought events are also simulated by internal atmosphere variability. This suggests that despite making droughts worse, ocean variability does not add predictive skill for duration of long-term drought events. The results for rainfall mean and extremes with and without ocean variability will be discussed in this talk.

52th APL Open Forum

Date & Time:
26 April, Friday, 14:30-16:00
Place:
Meeting Room 1+2, 2nd floor, Conference Building, Yokohama Institute
Speaker 1:
Swadhin Behera
Title:
Air-sea interaction in the Gulf of Alaska: ENSO and beyond
Abstract:
The interannual variation of sea surface temperature (SST) in Gulf of Alaska are said to be linked to the ENSO variability. The ENSO teleconnection, typically seen as the PNA pattern, is such that an El Nino teleconnection strengthens westerlies in the midlatitude North Pacific causing cold SST anomaly in the central and western subpolar gyre. At this time, it is noted that a strengthened subpolar gyre advects anomalously warm water northward along the eastern boundary into the Gulf of Alaska. While this teleconnection could explain some of the warm events, it does not explain all the warm events in the Gulf. This is also reflected in the fact that Nino3 index is not so-well-correlated with the Alaska coastal SST index derived in this study. Other mechanisms, including local air-sea interactions, play important roles in the development of warm and cold anomalies in the Gulf of Alaska.
Speaker 2:
Yasumasa Miyazawa
Title:
Applying a variant of four-dimensional ensemble-based variational method (4DEnVAR) to the Kuroshio variations south of Japan
Abstract:
Operational ocean nowcast/forecast systems require real-time sampling of oceanic data for representing realistic oceanic conditions. The satellite altimetry plays a key role in detecting mesoscale ocean current variations. A major sampling period of 10-day and a maximum horizontal gap between the altimetry tracks of 100km cause difficulty to capture shorter-term/smaller-scale ocean current variations. Our current operational systems based on a three-dimensional variational method (3DVAR) simply omit spatio-temporally varying information of the observation data within data assimilation time windows. Four-dimensional frameworks of data assimilation could be considered as a possible direction toward more effective utilization of the available observation data by extracting dynamical information especially from the satellite altimetry data. We are developing a variant of four-dimensional ensemble-based variational method (4DEnVAR) for our operational applications. Preliminary experiments targeting the Kuroshio variations south of Japan within a time window of 9-day indicate that our 4DEnVAR scheme surely improves 3DVAR results by more effectively assimilating the satellite altimetry data.

APL Guest Seminar

Date & Time:
27 March, Wednesday, 14:30-15:30
Place:
Room 1+2, 2nd floor, Conference Building, Yokohama Institute
Speaker:
Dr. Manali Pal (Indian Institute of Technology, Kharagpur, India)
Title:
The Long-lead Prediction of El Nino Modoki using Machine Learning algorithms
Abstract:
The focus of this study is to evaluate the efficacy of the Machine Learning algorithms in the long-lead prediction of the El Nino Modoki index after identification of the significantly contributing variables. We used three Machine Learning methods namely Random Forest (RF), Support Vector Regression (SVR) and Artificial Neural Network (ANN) to deal with the highly non-linear and complex behavior of the ocean-atmosphere coupled processes and their efficacies are assessed at four different lags i.e. 6, 12, 18 and 24 months to predict the Modoki index. The contributing variables are identified using Kendall rank correlation coefficient (also known as Kendall's tau), which is a non-parametric measure of relationships. Furthermore, the supervised principal component analysis is used to assess the contribution of each identified variable. The results of long-lead prediction show all the three models to capture the peaks of the Modoki index with lags up to 12 months. For the lags beyond 12 months, the models showed limitation in the predictability perhaps owing to over fitting issue in the short-period datasets.

APL Guest Seminar

Date & Time:
4 March, Monday, 14:00-15:00
Place:
5th floor meeting room, IT Building, Yokohama Institute
Speaker:
Prof. X. San Liang (Nanjing Institute of Meteorology, Nanjing, China)
Title:
Causality, Information Flow, and Quantitative Causal Inference with Time Series
Abstract:
Causal inference is a fundamental problem lying at the heart of scientific research. Recently, a rigorous formalism has been established ab initio for information flow/transfer and causality within dynamical systems. The “principle of nil causality” that reads, an event is not causal to another if the evolution of the latter is independent of the former, which classical formalisms fail to verify in many situations, turns out to be a proven theorem here. Moreover, the so-obtained information flow and causality is invariant upon nonlinear coordinate transformation, indicating that it should be an intrinsic property in physical world. For linear systems, there exists a remarkably concise formula, which asserts analytically that causation implies correlation, while correlation does not imply causation, resolving the long-standing philosophical debate over causation versus correlation ever since George Berkeley (1710).
This formalism has been validated with many benchmark systems (e.g., baker transformation, Hénon map, etc.), and touchstone series purportedly generated with one-way causality that defies the classical approaches. It has also been applied to the investigation of many real world problems. For example, we have studied the cause-effect relation between the two climate modes El Niño and Indian Ocean Dipole (IOD). In general, these modes are mutually causal, but the causality is asymmetric. To El Niño, the information flowing from IOD manifests itself as a propagation of uncertainty from the Indian Ocean.
In the second example, an unambiguous one-way causality is found between CO2 and the global mean temperature anomaly. While it is confirmed that CO2 indeed drives the recent global warming, on paleoclimate scales the cause-effect relation may be completely reversed.
Also presented will be the causation and information exchange during the interaction between transient storms and large-scale atmospheric flows, the identification of PM2.5 sources, network deconvolution, financial time series analysis, to name a few.

51th APL Open Forum

Date & Time:
22 February, Friday, 15:45-17:15
Place:
Meeting Room 1+2, 2nd floor, Conference Building, Yokohama Institute
Speaker 1:
Toru Miyama
Title:
Role of river inflows from Kamchatka Peninsula in the Okhotsk Sea
Abstract:
The Okhotsk Sea is a unique ocean in that sea ice forms at the lowest latitude in the world. The sea ice formation plays important roles in water mass formations not only in the Okhotsk Sea but also in the North Pacific Ocean. The river inflows from the surrounding coasts affect the sea ice formation by changing stratification. Especially, the contribution from the river inflow from the Amur River on the western side of the Okhotsk Sea has been widely discussed. The roles of the Amur River in chemical and biological cycles have been also investigated. On the hand, the role of the river inflows from the Kamchatka Peninsula on the eastern side of the Okhotsk Sea is not well known. In this study, we simulated the northern part of the Okhotsk Sea to investigate the roles of river inflows from the Kamchatka Peninsula. FVCOM (Finite-Volume, Primitive equation Community Ocean Model; Chen et al. 2003) was used. JRA55-do (Tsujino et al., 2018) was used for the atmospheric forcing. The river inflow dataset also came from JRA55-do based on Suzuki et al. (2017). The unstructured horizontal grid spacing was decreased from about 9.0 km at the southern boundary to about 1.4 km along the northern boundary. The model runs with and without the river inflows from the Kamchatka Peninsula were compared. The salinity significantly increased without the river inflow along the coast of Kamchatka (by more than 1 unit at maximum). The salinity increase also spread toward the western Okhotsk Sea. With these changes in salinity, ice formation in the Okhotsk Sea was also affected.
Speaker 2:
Sergey Varlamov
Title:
Regional JCOPE modeling with Multi Scale 3DVar data assimilation
Abstract:
The regional JCOPE-T models are quite intensively used for evaluation and forecast of oceanic circulation and thermodynamic conditions for number of practical applications, like real-time navigation and planning of marine activities. As examples, we could mention an informational support of JAMSTEC drilling ship "Chikyu" ongoing operations. It makes quite important to keep model state close to real observed oceanic conditions by performing observational data assimilation.
In the versions of regional model prior to 2018 it was a simple nudging method to the assimilative coarse resolution "mother" model JCOPE2M used for these purposes. From this year direct Multi Scale 3DVar assimilation method replaced nudging. It allowed to utilize modern satellite high resolution SST data provided from Himawari-8 geostationary satellite and other observed data. Details of realized assimilation process and some cases of solved problems during tuning the method would be reported.

APL Guest Seminar

Date & Time:
20 February, Wednesday, 15:00-16:00
Place:
5th floor meeting room, IT Building, Yokohama Institute
Speaker:
Mr. Givo Alsepan (Hokkaido University)
Title:
Relation between interannual variability of regional-scale Indonesian precipitation and large-scale climate modes in the last half century
Abstract:
Regional–scale precipitation responses over Indonesia to major climate modes in the tropical Indo–Pacific Oceans, i.e., canonical El Niño–Southern Oscillation (ENSO), El Niño Modoki, and Indian Ocean Dipole (IOD), are investigated using a terrestrial precipitation dataset, Asian Precipitation–Highly Resolved Observational Data Integration Towards Evaluation (APHRODITE). The spatial resolution of the precipitation data is 0.5° x 0.5°, much higher than the precipitation datasets used in previous studies, allowing us to clarify spatial patterns of precipitation responses in greater detail. The explained variance ratios of climate modes to precipitation generally have stronger magnitudes from July to November, corresponding to the dry season and transition period from the dry to wet. Since the climate modes are not statistically independent, we employ standard and partial correlation analyses. A standard correlation for a climate mode is regarded as the correlation when the corresponding climate mode is the primary driver. On the other hand, partial correlation is the correlation that is intrinsic to the mode even after removal of other modes’ influences. In order to know seasonal development of climate modes’ influence, we calculate the standard and partial correlations between precipitation and climate modes in each calendar month after smoothing with a three–month running mean. The standard correlations show that canonical ENSO reduces precipitation in a wide area from western to eastern Indonesia from July to November and northern Indonesia slightly after it from December to April. El Niño Modoki also causes precipitation reduction with weaker correlations than those of the canonical ENSO from July to November, but conversely induces precipitation increase in western Indonesia for some months in the wet season. IOD also reduces precipitation with an overlapping spatial pattern to the canonical ENSO, and this influence is limited from July to November. Interestingly, the partial correlations produce a more limited spatial pattern of intrinsic influence of canonical ENSO and IOD to precipitation than those in standard correlations, while for El Niño Modoki the standard and partial correlations produce similar features. Intrinsic influence of canonical ENSO to precipitation reduction is confined in central and eastern Indonesia from July to November, and for IOD it is mainly pronounced in western Indonesia. Most of these precipitation reductions are strongly associated with the contraction of convergence zones from their mean state, except for positive precipitation response to El Niño Modoki.

APL Guest Seminar

Date & Time:
18 February, Monday, 14:00-15:30
Place:
Meeting Room 1+2, 2nd floor, Conference Building, Yokohama Institute
Speaker 1:
Dr. Sébastien Masson (LOCEAN/IPSL)
Title:
Impact of the current / wind stress coupling on the eddy activity
Abstract:
Ocean mesoscale eddies are major players of the ocean circulation, and transport of heat, salt, and biogeochemical tracers. These coherent structures are characterized by rotating currents that imprint the low atmosphere and feedback to the wind power input to the ocean. With a hierarchy of coupled models, this study shows that the eddy kinetic energy (EKE) of the ocean is decreased by ~30% in average by current feedback. The large impact of this feedback is often described as the “eddy killing effect”. Our simulations show that the amplitude (in %) of the eddy killing effect is mainly controlled by the large scale wind. Locally this effect is modulated by the balance between wind work at mesoscale, and ocean intrinsic instabilities that are the main source of EKE. The efficiency of current feedback appears quite insensitive to model resolution, and in particular to the atmospheric resolution. These results provide guidance for required processes, and resolution for a correct representation of air-sea energy transfers.
Speaker 2:
Dr. Clement Rousset (LOCEAN/IPSL)
Title:
What does it take to model the cold oceans?
Abstract:
The cold oceans occupy roughly 35 million km2 which represents about 10% of the global ocean surface. They are characterized by a seasonal presence of sea-ice which acts as an insulator between the atmosphere and the ocean by reflecting most of the incoming solar radiation and damping the air-sea momentum and heat exchanges. They are also the main locations of dense water formation in the world. Their role for climate is thus crucial. In this talk I will set out the key ingredients to model these regions, with a focus on the NEMO model.
Speaker 3:
Dr. Eric Maisonnave (CERFACS)
Title:
NEMO 4.0 performance: most recent improvements
Abstract:
A non-intrusive instrumentation of the NEMO code and the development of a simplified configuration (called BENCH) brought information about MPI communications cost and structure. It helped us to identify the most appropriate incremental developments that model needs to enhance its scalability. We prioritised the reduction of extra calculations and communications required at the North Polar folding, the grouping of boundary exchanges and the replacement of global communications by alternative algorithms. Appreciable speed up (x2 in some cases) is measured. Scalability limit is pushed below a size of 7x7 grid points per sub-domain, showing that the limitation of the North Polar folding solution can be compared with the supposed icosahedral grid one. However, disk access and per-user bandwidth availability clearly limits over-10,000 CPU resource performance. We consider that scalability is not the major well of future performance gain, neither horizontal resolution increase, whereas potentiality of extra developments accelerating cache access (horizontal domain tiling and single precision computations) is favourably evaluated.

APL Guest Seminar

Date & Time:
28 January, Monday, 11:00-12:00
Place:
Room 1+2, 2nd floor, Conference Building, Yokohama Institute
Speaker:
Dr. Ibrahim Hoteit (Professor, King Abdullah University of Science and Technology, Thuwal, Saudi Arabia)
Title:
An Integrated Data-Driven Modeling System to Study and Predict the Circulation, Productivity and Climate of the Red Sea
Abstract:
The talk will present the integrated data-driven modeling and forecasting system that we have developed to study and understand the physical and biological variability of the Red Sea. I will first describe the modeling system and summarize our key findings on the Red Sea general circulation, including the striking seasonally overturning circulation, the dominant eddy activity, and the occasional northern deep water formation events, and discuss their impact on the Red Sea ecosystem. I will then focus on our efforts to develop an efficient ensemble data assimilation and forecasting system for the Red Sea, presenting recent algorithmic developments and results, and discussing our future plans.

50th APL Open Forum

Date & Time:
25 January, Friday, 15:45-17:15
Place:
Meeting Room 1+2, 2nd floor, Conference Building, Yokohama Institute
Speaker 1:
Hideharu Sasaki
Title:
A Quasi-Global Eddying Hindcast Ocean Simulation of OFES2
Abstract:
Outputs from global eddying oceanic simulations have been widely used to study various oceanic features with broad spatiotemporal scales from mesoscale to large scales and from intraseasonal to decadal scales. Long-term hindcast outputs from OFES have been bringing out a number of research achievements through its release to the public. However, there are several issues with unrealistic properties in OFES. Therefore, I have conducted a quasi-global eddying oceanic simulation using a new version of OFES, which we call OFES2. In this presentation, I show the descriptions and simulated oceanic fields in OFES2 compared to OFES. We implemented the sea ice model and tidal mixing scheme in OFES2 and OFES2 is forced by newly created atmospheric reanalysis data called JRA55-do. Although a few issues remain, we found several improvements in OFES2: small biases in global sea surface temperature, sea surface salinity, and water properties in the Indonesian Seas compared to OFES. Time series of El Niño and Indian Ocean Dipole indexes are a bit better simulated in OFES2 than OFES. The well-verified outputs from OFES2 are expected to be widely used to study various oceanic phenomena with broad spatiotemporal scales.
Speaker 2:
Pascal Oettli
Title:
Surface air temperatures in Japan: A story
Abstract:
The mean areal surface air temperature anomalies are used to group 762 AMeDAS stations, taken throughout Japan, into homogeneous clusters. Clustering is performed seasonally, with a focus on boreal winter (DJF) and boreal summer (JJA) and associated time series of temperature anomalies are created. Subsequently, correlation between these indices and different climatic variables anomaly fields are calculated to explore the possible source(s) of the interannual variability of surface air temperature anomalies.