Skip to main content

Resultados relacionados

Resultados totales (Incluyendo duplicados): 3
Encontrada(s) 1 página(s)

DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

Climatologies of seawater CO2 chemistry variables: A neural network approach

  • Broullón, Daniel
  • Pérez, Fiz F.
  • Velo, Antón
  • Suzuki, Toru
1 poster presented at the 10th International Carbon Dioxide Conference, Interlaken, Switzerland, 21 August 2017 - 25 August 2017.-- Daniel Broullón ... et al., For decades, the anthropogenic modification of the carbon cycle has been widely studied. More recently, ocean acidification studies have increased significantly. Establishing robust climatologies of seawater CO2 chemistry variables and building models are a key point for a better understanding of the associated processes. The availability and quality of data is crucial for the evaluation of climate models and, consequently, to improve their predictions. Version 2 of the Global Ocean Data Analysis Project (GLODAPv2) is an internally consistent data product composed of data from 724 scientific cruises covering the entire global ocean. Among others, it contains seawater CO2 chemistry variables such as total alkalinity (AT), total dissolved inorganic carbon (TCO2) and pH. This sparse dataset has been subjected to extensive quality control and different interpolation techniques have been applied to extend the data coverage on a homogeneous grid (Lauvset et al. 2016). We propose a novel neural network approach to generate annual and monthly climatologies of AT, TCO2, pH and both calcite and aragonite saturation state from the GLODAPv2 dataset for the preindustrial and current periods. These climatologies are fitted to the World Ocean Atlas 2013 version 2 (WOA13v2) grid. WOA13v2 is a set of objectively analyzed (1° grid) climatological fields of different oceanographic variables (but not CO2 system) at standard depth levels for annual, seasonal, and monthly compositing periods for the World Ocean. A feed-forward neural network was chosen in a multi-layer architecture, which allows linear and nonlinear variability to be assimilated by the network. The proposed configuration is able to approximate most functions arbitrarily well (Hagan et al., 2014). We have tested different neural network designs and sizes to obtain the minimum error. For that, the number of neurons in the network was varied and different training techniques were used. The input variables introduced in the network, which must be related to AT and TCO2 variability, were latitude, longitude, depth, potential temperature, phosphate, nitrate, silicate, year, month and atmospheric pCO2. First, the network was trained with GLODAPv2 data and then AT and TCO2 fields were derived from WOA13v2 data. Saturation states and pH were computed from these two variables. The monthly pre-industrial climatology will be generated by eliminating anthropogenic carbon from the ocean., This research was supported by Ministerio de Educación, Cultura y Deporte (FPU grant FPU15/06026), Ministerio de Economía y Competitividad through the A. RIOS (CTM2016-76146-C3-1-R) project co-funded by the Fondo Europeo de Desarrollo Regional 2014-2020 (FEDER) and EU Horizon2020 through the AtlantOS project (grant agreement 633211)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

Towards comprehensive observing and modeling systems for monitoring and predicting regional to coastal sea level

  • Ponte, Rui M.
  • Carson, Mark
  • Cirano, Mauro
  • Domingues, Catia M.
  • Jevrejeva, Svetlana
  • Marcos, Marta
  • Mitchum, Gary
  • van de Wal, R. S. W.
  • Woodworth, Philip L.
  • Ablain, Michaël
  • Ardhuin, Fabrice
  • Ballu, Valérie
  • Becker, Mélanie
  • Benveniste, Jérôme
  • Birol, Florence
  • Bradshaw, Elizabeth
  • Cazenave, Anny
  • Mey-Frémaux, Pierre de
  • Durand, Fabien
  • Ezer, Tal
  • Fu, Lee-Lueng
  • Fukumori, Ichiro
  • Gordon, Kathy
  • Gravelle, Médéric
  • Griffies, Stephen M.
  • Han, Weiqing
  • Hibbert, Angela
  • Hughes, Chris William
  • Idier, Déborah
  • Kourafalou, Villy
  • Little, Christopher M.
  • Matthews, Andrew
  • Melet, Angélique
  • Merrifield, Mark
  • Meyssignac, Benoit
  • Minobe, Shoshiro
  • Penduff, Thierry
  • Picot, Nicolas
  • Piecuch, Christopher G.
  • Ray, Richard D.
  • Rickards, Lesley
  • Santamaría-Gómez, Álvaro
  • Stammer, Detlef
  • Staneva, Joanna
  • Testut, Laurent
  • Thompson, Keith
  • Thompson, Philip
  • Vignudelli, Stefano
  • Williams, Joanne
  • Williams, Simon D. P.
  • Wölppelmann, Guy
  • Laure, Zanna
  • Zhang, Shuebin
A major challenge for managing impacts and implementing effective mitigation measures and adaptation strategies for coastal zones affected by future sea level (SL) rise is our limited capacity to predict SL change at the coast on relevant spatial and temporal scales. Predicting coastal SL requires the ability to monitor and simulate a multitude of physical processes affecting SL, from local effects of wind waves and river runoff to remote influences of the large-scale ocean circulation on the coast. Here we assess our current understanding of the causes of coastal SL variability on monthly to multi-decadal timescales, including geodetic, oceanographic and atmospheric aspects of the problem, and review available observing systems informing on coastal SL. We also review the ability of existing models and data assimilation systems to estimate coastal SL variations and of atmosphere-ocean global coupled models and related regional downscaling efforts to project future SL changes. We discuss (1) observational gaps and uncertainties, and priorities for the development of an optimal and integrated coastal SL observing system, (2) strategies for advancing model capabilities in forecasting short-term processes and projecting long-term changes affecting coastal SL, and (3) possible future developments of sea level services enabling better connection of scientists and user communities and facilitating assessment and decision making for adaptation to future coastal SL change., RP was funded by NASA grant NNH16CT00C. CD was supported by the Australian Research Council (FT130101532 and DP 160103130), the Scientific Committee on Oceanic Research (SCOR) Working Group 148, funded by national SCOR committees and a grant to SCOR from the U.S. National Science Foundation (Grant OCE-1546580), and the Intergovernmental Oceanographic Commission of UNESCO/International Oceanographic Data and Information Exchange (IOC/IODE) IQuOD Steering Group. SJ was supported by the Natural Environmental Research Council under Grant Agreement No. NE/P01517/1 and by the EPSRC NEWTON Fund Sustainable Deltas Programme, Grant Number EP/R024537/1. RvdW received funding from NWO, Grant 866.13.001. WH was supported by NASA (NNX17AI63G and NNX17AH25G). CL was supported by NASA Grant NNH16CT01C. This work is a contribution to the PIRATE project funded by CNES (to TP). PT was supported by the NOAA Research Global Ocean Monitoring and Observing Program through its sponsorship of UHSLC (NA16NMF4320058). JS was supported by EU contract 730030 (call H2020-EO-2016, “CEASELESS”). JW was supported by EU Horizon 2020 Grant 633211, Atlantos.
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

Inteligencia artificial contra el cambio climático

  • Broullón, Daniel
  • Pérez, Fiz F.
  • Velo, Antón
VI International Symposium on Marine Sciences, Vigo (Spain), 20-22th June 2018, El ciclo del carbono inorgánico en el océano juega un papel fundamental en el cambio climático. La necesidad de conocer con detalle su funcionamiento es esencial para realizar evaluaciones con rigor. Las climatologías globales de las variables de este sistema son una herramienta básica para conocer el estado del problema, principalmente de la acidificación oceánica. En este trabajo se ha utilizado el potencial de las redes neuronales y las bases de datos Global Ocean Data Project Version 2 (GLODAPv2; Key et al. 2015; Olsen et al. 2016) y Lamont-Doherty Earth Observatory database (LDEO; Takahashi et al., 2017) para generar climatologías mensuales de alcalinidad total (AT) y carbono inorgánico disuelto total (TCO2). Una red neuronal feed-forward con una arquitectura de dos capas y 128 neuronas fue entrenada para cada variable con los datos de GLODAPv2 y LDEO para aprender las relaciones entre predictores relacionados con las variables y éstas. Una función sigmoidal y una lineal fueron utilizadas en cada capa para modelar la variabilidad no lineal. El comportamiento de cada red fue testado en un set independiente extraído de las bases de datos y en las diversas series temporales de las estaciones fijas de medida distribuidas por el océano global. Las climatologías de AT y TCO2 fueron diseñadas sobre las climatologías de los predictores utilizados en el entrenamiento que ofrece el World Ocean Atlas 2013 (WOA13) en una malla global de 1ºx1º y 102 niveles de profundidad. La red mostró un buen comportamiento al modelar los datos de los sets independientes y de las series temporales con errores cercanos a la incertidumbre de las medidas. La capacidad de la red para modelar el ciclo estacional en las series temporales permitió generar climatologías mensuales supeditadas a la calidad del diseño de WOA13. Este producto se ofrece para ser introducido en futuros modelos, Este trabajo de investigación está financiado por el Ministerio de Educación, Cultura y Deporte a través del contrato FPU15/06026, por el Ministerio de Economía y Competitividad a través del proyecto A.RIOS (CTM2016-76146-C3-1-R) y por el Fondo Europeo de Desarrollo Regional 2014-2020 (FEDER) y EU Horizon 2020 a través del proyecto AtlantOS (grant agreement 633211), Peer reviewed