Deep learning and process understanding for data-driven Earth system science

Machine learning approaches are increasingly used to extract patterns and insights from the ever-increasing stream of geospatial data, but current approaches may not be optimal when system behaviour is dominated by spatial or temporal context. Here, rather than amending classical machine learning, we argue that these contextual cues should be used as part of deep learning (an approach that is able to extract spatio-temporal features automatically) to gain further process understanding of Earth system science problems, improving the predictive ability of seasonal forecasting and modelling of long-range spatial connections across multiple timescales, for example. The next step will be a hybrid modelling approach, coupling physical process models with the versatility of data-driven machine learning.Complex Earth system challenges can be addressed by incorporating spatial and temporal context into machine learning, especially via deep learning, and further by combining with physical models into hybrid models.

[1]  Tim Appelhans,et al.  Improving the accuracy of rainfall rates from optical satellite sensors with machine learning — A random forests-based approach applied to MSG SEVIRI , 2014 .

[2]  Prabhat,et al.  Application of Deep Convolutional Neural Networks for Detecting Extreme Weather in Climate Datasets , 2016, ArXiv.

[3]  Robert Pincus,et al.  Parameter estimation using data assimilation in an atmospheric general circulation model: From a perfect toward the real world , 2013 .

[4]  Ronald M. Welch,et al.  A neural network approach to cloud classification , 1990 .

[5]  P. E. O'connell,et al.  River flow forecasting through conceptual models part III - The Ray catchment at Grendon Underwood , 1970 .

[6]  Alexei A. Efros,et al.  Image-to-Image Translation with Conditional Adversarial Networks , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[7]  Omar Bellprat,et al.  Objective calibration of regional climate models: OBJECTIVE CALIBRATION OF RCMS , 2012 .

[8]  P. Ciais,et al.  Spatiotemporal patterns of terrestrial gross primary production: A review , 2015 .

[9]  Robert Jacob,et al.  Statistical emulation of climate model projections based on precomputed GCM runs , 2013 .

[10]  B. Santer,et al.  Statistical significance of climate sensitivity predictors obtained by data mining , 2014 .

[11]  D. A. Kenny,et al.  Correlation and Causation , 1937, Wilmott.

[12]  Ken Perlin,et al.  Accelerating Eulerian Fluid Simulation With Convolutional Networks , 2016, ICML.

[13]  Luis Alonso,et al.  Retrieval of Vegetation Biophysical Parameters Using Gaussian Process Techniques , 2012, IEEE Transactions on Geoscience and Remote Sensing.

[14]  Gabriele Moser,et al.  Multimodal Classification of Remote Sensing Images: A Review and Future Directions , 2015, Proceedings of the IEEE.

[15]  N. Guttman ACCEPTING THE STANDARDIZED PRECIPITATION INDEX: A CALCULATION ALGORITHM 1 , 1999 .

[16]  Jane R. Foster,et al.  Application of imaging spectroscopy to mapping canopy nitrogen in the forests of the central Appalachian Mountains using Hyperion and AVIRIS , 2003, IEEE Trans. Geosci. Remote. Sens..

[17]  Honglak Lee,et al.  Action-Conditional Video Prediction using Deep Networks in Atari Games , 2015, NIPS.

[18]  B. Frey,et al.  Predicting the sequence specificities of DNA- and RNA-binding proteins by deep learning , 2015, Nature Biotechnology.

[19]  Prabhat,et al.  ExtremeWeather: A large-scale climate dataset for semi-supervised detection, localization, and understanding of extreme weather events , 2016, NIPS.

[20]  Philippe Ciais,et al.  The status and challenge of global fire modelling , 2016 .

[21]  Ryan Kelly,et al.  Linking big models to big data: efficient ecosystem model calibration through Bayesian model emulation , 2016, Biogeosciences.

[22]  P. Swain,et al.  Neural Network Approaches Versus Statistical Methods In Classification Of Multisource Remote Sensing Data , 1990 .

[23]  Derek T. Anderson,et al.  Comprehensive survey of deep learning in remote sensing: theories, tools, and challenges for the community , 2017 .

[24]  R. Valentini,et al.  A new assessment of European forests carbon exchanges by eddy fluxes and artificial neural network spatialization , 2003 .

[25]  Zoubin Ghahramani,et al.  Probabilistic machine learning and artificial intelligence , 2015, Nature.

[26]  Yann LeCun,et al.  Deep multi-scale video prediction beyond mean square error , 2015, ICLR.

[27]  Philippe Ciais,et al.  A framework for benchmarking land models , 2012 .

[28]  Alain Chedin,et al.  A Neural Network Approach for a Fast and Accurate Computation of a Longwave Radiative Budget , 1998 .

[29]  Prabhat,et al.  Deep Neural Networks for Physics Analysis on low-level whole-detector data at the LHC , 2017, Journal of Physics: Conference Series.

[30]  Athos Agapiou,et al.  Remote sensing heritage in a petabyte-scale: satellite data and heritage Earth Engine© applications , 2017, Int. J. Digit. Earth.

[31]  J. Pekel,et al.  High-resolution mapping of global surface water and its long-term changes , 2016, Nature.

[32]  Yoshua Bengio,et al.  Generative Adversarial Nets , 2014, NIPS.

[33]  Shihong Du,et al.  Learning multiscale and deep representations for classifying remotely sensed imagery , 2016 .

[34]  Nagiza F. Samatova,et al.  Theory-Guided Data Science: A New Paradigm for Scientific Discovery from Data , 2016, IEEE Transactions on Knowledge and Data Engineering.

[35]  Bjorn Stevens,et al.  Imprint of the convective parameterization and sea‐surface temperature on large‐scale convective self‐aggregation , 2017 .

[36]  Eric Guilyardi,et al.  Towards improved and more routine Earth system model evaluation in CMIP , 2016 .

[37]  D. Wilks Multivariate ensemble Model Output Statistics using empirical copulas , 2015 .

[38]  Mark J. Willis,et al.  Simultaneous parameter identification and discrimination of the nonparametric structure of hybrid semi-parametric models , 2017, Comput. Chem. Eng..

[39]  Peter Bauer,et al.  The quiet revolution of numerical weather prediction , 2015, Nature.

[40]  Markus Reichstein,et al.  Predicting carbon dioxide and energy fluxes across global FLUXNET sites with regression algorithms , 2016 .

[41]  Robert Pincus,et al.  On Constraining Estimates of Climate Sensitivity with Present-Day Observations through Model Weighting , 2011 .

[42]  Dit-Yan Yeung,et al.  Deep Learning for Precipitation Nowcasting: A Benchmark and A New Model , 2017, NIPS.

[43]  M. Pal,et al.  Random forests for land cover classification , 2003, IGARSS 2003. 2003 IEEE International Geoscience and Remote Sensing Symposium. Proceedings (IEEE Cat. No.03CH37477).

[44]  Jürgen Schmidhuber,et al.  Deep learning in neural networks: An overview , 2014, Neural Networks.

[45]  Atul K. Jain,et al.  Compensatory water effects link yearly global land CO2 sink changes to temperature , 2017, Nature.

[46]  Nicholas C. Coops,et al.  Prediction of eucalypt foliage nitrogen content from satellite-derived hyperspectral data , 2003, IEEE Trans. Geosci. Remote. Sens..

[47]  Jon Atli Benediktsson,et al.  Advances in Hyperspectral Image Classification: Earth Monitoring with Statistical Learning Methods , 2013, IEEE Signal Processing Magazine.

[48]  S. Vicente‐Serrano,et al.  A Multiscalar Drought Index Sensitive to Global Warming: The Standardized Precipitation Evapotranspiration Index , 2009 .

[49]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[50]  Patrick Gallinari,et al.  Deep learning for physical processes: incorporating prior scientific knowledge , 2017, ICLR.

[51]  Dit-Yan Yeung,et al.  Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting , 2015, NIPS.

[52]  P. Games Correlation and Causation: A Logical Snafu , 1990 .

[53]  M. Payne,et al.  A neural network-based estimate of the seasonal to inter-annual variability of the Atlantic Ocean carbon sink , 2013 .

[54]  Wojciech Samek,et al.  Methods for interpreting and understanding deep neural networks , 2017, Digit. Signal Process..

[55]  Pierre Gentine,et al.  Could Machine Learning Break the Convection Parameterization Deadlock? , 2018, Geophysical Research Letters.

[56]  Jia Deng,et al.  A large-scale hierarchical image database , 2009, CVPR 2009.

[57]  Baskar Ganapathysubramanian,et al.  Hierarchical Feature Extraction for Efficient Design of Microfluidic Flow Patterns , 2015, FE@NIPS.

[58]  Robert C. Wolpert,et al.  A Review of the , 1985 .

[59]  F. Woodward,et al.  Terrestrial Gross Carbon Dioxide Uptake: Global Distribution and Covariation with Climate , 2010, Science.

[60]  Anuj Karpatne,et al.  Physics-guided Neural Networks (PGNN): An Application in Lake Temperature Modeling , 2017, ArXiv.

[61]  Robert M. May,et al.  Simple mathematical models with very complicated dynamics , 1976, Nature.

[62]  Sangram Ganguly,et al.  Generating High Resolution Climate Change Projections through Single Image Super-Resolution: An Abridged Version , 2018, IJCAI.

[63]  Nuno Carvalhais,et al.  Enhanced seasonal CO2 exchange caused by amplified plant productivity in northern ecosystems , 2016, Science.

[64]  Neus Sabater,et al.  Emulation of Leaf, Canopy and Atmosphere Radiative Transfer Models for Fast Global Sensitivity Analysis , 2016, Remote. Sens..

[65]  Markus Reichstein,et al.  Improving canopy processes in the Community Land Model version 4 (CLM4) using global flux fields empirically inferred from FLUXNET data , 2011 .

[66]  Mohamed Akram,et al.  Sequence to Sequence Weather Forecasting with Long Short-Term Memory Recurrent Neural Networks , 2016 .

[67]  H. Glahn,et al.  The Use of Model Output Statistics (MOS) in Objective Weather Forecasting , 1972 .

[68]  Joachim Denzler,et al.  Predicting Landscapes as Seen from Space from Environmental Conditions , 2018, IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium.

[69]  Jürgen Kurths,et al.  Identifying causal gateways and mediators in complex spatio-temporal systems , 2015, Nature Communications.

[70]  Geoffrey E. Hinton,et al.  Deep Learning , 2015, Nature.

[71]  S. Seneviratne,et al.  Recent decline in the global land evapotranspiration trend due to limited moisture supply , 2010, Nature.

[72]  Kaiming He,et al.  Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[73]  Nils Thürey,et al.  tempoGAN: A Temporally Coherent, Volumetric GAN for Super-resolution Fluid Flow , 2018, ACM Trans. Graph..

[74]  Barak A. Pearlmutter,et al.  Reverse-mode AD in a functional framework: Lambda the ultimate backpropagator , 2008, TOPL.

[75]  C. Justice,et al.  High-Resolution Global Maps of 21st-Century Forest Cover Change , 2013, Science.

[76]  P. Cox,et al.  Sensitivity of tropical carbon to climate change constrained by carbon dioxide variability , 2013, Nature.

[77]  P. Mahadevan,et al.  An overview , 2007, Journal of Biosciences.

[78]  Luca Martino,et al.  Physics-aware Gaussian processes in remote sensing , 2018, Appl. Soft Comput..

[79]  R. Wood,et al.  Climatology of stratocumulus cloud morphologies: microphysical properties and radiative effects , 2014 .

[80]  Markus Reichstein,et al.  Soil respiration across scales: The importance of a model–data integration framework for data interpretation† , 2008 .

[81]  M. Körner,et al.  MULTI-TEMPORAL LAND COVER CLASSIFICATION WITH LONG SHORT-TERM MEMORY NEURAL NETWORKS , 2017 .

[82]  Fei-Fei Li,et al.  ImageNet: A large-scale hierarchical image database , 2009, 2009 IEEE Conference on Computer Vision and Pattern Recognition.

[83]  Noah D. Brenowitz,et al.  Prognostic Validation of a Neural Network Unified Physics Parameterization , 2018, Geophysical Research Letters.

[84]  H. Elsenbeer,et al.  Soil organic carbon concentrations and stocks on Barro Colorado Island — Digital soil mapping using Random Forests analysis , 2008 .

[85]  Martina Stockhause,et al.  CMIP6 Data Citation of Evolving Data , 2017 .

[86]  Lorenzo Bruzzone,et al.  Kernel methods for remote sensing data analysis , 2009 .

[87]  Beck Hylke,et al.  Global-scale regionalization of hydrologic model parameters , 2016 .

[88]  Bo Du,et al.  Deep Learning for Remote Sensing Data: A Technical Tutorial on the State of the Art , 2016, IEEE Geoscience and Remote Sensing Magazine.

[89]  Marvin N. Wright,et al.  SoilGrids250m: Global gridded soil information based on machine learning , 2017, PloS one.

[90]  Alexandre Tkatchenko,et al.  Quantum-chemical insights from deep tensor neural networks , 2016, Nature Communications.

[91]  Bjorn Stevens,et al.  Sensitivity of the summertime tropical Atlantic precipitation distribution to convective parameterization and model resolution in ECHAM6 , 2017 .

[92]  Tim R. McVicar,et al.  Global‐scale regionalization of hydrologic model parameters , 2016 .

[93]  Pierre Friedlingstein,et al.  Uncertainties in CMIP5 Climate Projections due to Carbon Cycle Feedbacks , 2014 .

[94]  Derek Anderson,et al.  Comprehensive survey of deep learning in remote sensing: theories, tools, and challenges for the community , 2017, ArXiv.

[95]  S. E. Haupt,et al.  Using Artificial Intelligence to Improve Real-Time Decision-Making for High-Impact Weather , 2017 .

[96]  J. Nash,et al.  River flow forecasting through conceptual models part I — A discussion of principles☆ , 1970 .

[97]  P. Baldi,et al.  Searching for exotic particles in high-energy physics with deep learning , 2014, Nature Communications.

[98]  Sergio M. Vicente-Serrano,et al.  The Standardized Precipitation-Evapotranspiration Index (SPEI): a multiscalar drought index , 2010 .

[99]  A. Arneth,et al.  Global patterns of land-atmosphere fluxes of carbon dioxide, latent heat, and sensible heat derived from eddy covariance, satellite, and meteorological observations , 2011 .