Our approach to geosciences is from a pure signal processing and machine learning perspective. Despite the many successful results and developments, there are still strong limitations for the general adoption of machine learning algorithms for predicting and understanding Earth observation (EO) data. Machine learning and signal processing have advanced enormously in the last decade (both at a theoretical and applied levels) but have not moved forward the field of EO data analysis in all its real potential.
The current statistical treatment of biophysical parameters is strongly limited by the quantity and quality of EO data, as well as because of the abuse of standard off-the-shelf methods, which in general are not well-adapted to the particular EO data characteristics. Specifically, current regression models used for EO applications are still deficient because they rely on limited amount of meteorological and remote sensing data, do not observe the particular data characteristics, and often make strong assumptions of linearity, homoscedasticity or Gaussianity. These limitations translate into certain risks of overfitting and unreasonably large uncertainties for the predictions, suggesting a lack of explanatory variables and deficiencies in model specification. Graphical models have been seldom used in EO data analysis. The few works restrict to local studies, use limited amount of data and explanatory variables, consider remote sensing input features only, apply standard structure learning algorithms driven by univariate (often unconditioned) dependence estimates, and do not extract causal relations or identify new drivers in the problem.
We advocate that machine learning algorithms for EO applications need to be guided both by data and by prior physical knowledge. This combination is the way to restrict the family of possible solutions and thus obtain nonparametric flexible models that respect the physical rules governing the Earth climate system. We are equally concerned about the ‘black-box' criticism to statistical learning algorithms, for which we aim to design self-explanatory models and take a leap towards the relevant concept of causal inference from empirical EO data.
Our main goal is to develop new machine learning models for the efficient treatment of biophysical land parameters and related covariates at local and global scales. This main scientific goal translates into the following objectives:
Improve prediction models by adaptation to Earth Observation data characteristics. We typically rely on the framework of kernel learning, which has emerged as the most appropriate framework for remote sensing data analysis in the last decade. The new retrieval models are adapted to the particular signal characteristics, such as unevenly sampled time series and missing data, non-Gaussianity, presence of heteroscedastic and non-stationary processes, and non-i.i.d. (spatial and temporal) relations. Models based on kernels and GPs allow us to advance in uncertainty quantification using predictive variances under biophysical constraints. Advances in sparse, reduced-rank and divide-and-conquer schemes address the computational cost problem. The proposed kernel framework aims to improve results in terms of accuracy, reduced uncertainty, consistency of the estimations and computational efficiency.
Discover knowledge and causal relations in Earth observation data. We investigate graphical causal models and regression-based causal schemes applied to large heterogeneous EO data streams. This requires improved measures of (conditional) independence, designing experiments in controlled situations and using high-quality data. Learning the hierarchy of the relations between variables and related covariates, as well as their causal relations, may in turn allow the discovery of hidden essential variables, drivers and confounders. Moving from correlation to dependence and then to causation concepts is fundamental to advance the field of Earth Observation and the science of climate change.
Current related projects
>SEDAL: Statistical Learning for Earth Observation Data Analysis
ERC Consolidator Grant, G. Camps-Valls, 01/15 - 12/19
Mapping and the citizen sensor
ICT COST Action, 01/13 - 12/16
Cloud detection in the cloud
Google Earth Engine Research Award, L. Gomez-Chova, 01/16 - 12/17
- Recent advances in the retrieval of terrestrial vegetation properties with optical remote sensing - A review. J. Verrelst, G. Camps-Valls, J. Muñoz-Marí, J.P. Rivera, F. Veroustraete, J. Clevers, and J. Moreno.
- A Survey on Gaussian Processes for Earth Observation Data Analysis. G. Camps-Valls, J. Muñoz-Marí, J. Verrelst, F. Mateo, J. Gomez-Dans, IEEE Geoscience and Remote Sensing Magazine, 2016.
- Uncertainty analysis of Gross Primary Production predictions using Random Forests, remote sensing and eddy covariance data. Gianluca Tramontana, Kazuito Ichii, Gustau Camps-Valls, Enrico Tomelleri and Dario Papale. Remote Sensing of Environment, Volume 168, Pages 360-373, October 2015.
- An emulator toolbox to approximate radiative transfer models with statistical learning. Rivera, JP; Verrelst, J.; Gomez-Dans, J.; Muñoz-Marí, J.; Moreno, J; and Camps-Valls, G. Remote Sensing. 7, p. 9347-9370, 2015.
- Optical remote sensing and the retrieval of terrestrial vegetation bio-geophysical properties - A review. Verrelst, J.; Camps-Valls, G.; Muñoz-Marí, J.; Rivera, JP; Veroustraete, F.; Clevers, J.P.G.W., Moreno, J. ISPRS Journal of Photogrammetry and Remote Sensing, Vol 108, Pages 273-290, 2015.
- Experimental Sentinel-2 LAI estimation using parametric, non-parametric and physical retrieval methods - A comparison. Verrelst, J.; Rivera, JP; Veroustraete, F.; Muñoz-Marí, J.; Clevers, J.P.G.W.; Camps-Valls, G.; Moreno, J. ISPRS Journal of Photogrammetry and Remote Sensing. Vol 108, Pages 260-272, 2015.
- Mapping Leaf Area Index with a Smartphone and Gaussian Processes. Manuel Campos-Taberner, Franciso Javier García-Haro, Álvaro Moreno, María Amparo Gilabert, Sergio Sánchez-Ruiz, Beatriz Martínez and Gustau Camps-Valls. IEEE Geoscience and Remote Sensing Letters. In press, 2015.
- Replacing radiative transfer models by surrogate approximations through machine learning. Jochem Verrelst, Juan Pablo Rivera, Jose Gómez-Dans, Gustau Camps-Valls and Jose Moreno. IEEE International Geoscience and Remote Sensing Symposium 2015 (IGARSS 2015). 26-31 July, 2015, Milan, Italy.
- Ranking drivers of global carbon and energy fluxes over land. Gustau Camps-Valls , Martin Jung, Kazuhito Ichii, Dario Papale, Gianluca Tramontana, Paul Bodesheim, Christopher Schwalm, Jakob Zscheischler, Miguel Mahecha, and Markus Reichstein. IEEE International Geoscience and Remote Sensing Symposium 2015 (IGARSS 2015). 26-31 July, 2015, Milan, Italy.
- Sensitivity analysis of Gaussian processes for oceanic chlorophyll prediction. Katalin Blix, Robert Jenssen and Gustau Camps-Valls. IEEE International Geoscience and Remote Sensing Symposium 2015 (IGARSS 2015). 26-31 July, 2015, Milan, Italy.
- Large-scale random features kernel regression. Valero Laparra, Diego Marcos Gonzalez, Devis Tuia and Gustau Camps-Valls. IEEE International Geoscience and Remote Sensing Symposium 2015 (IGARSS 2015). 26-31 July, 2015, Milan, Italy.
- Biophysical parameter retrieval with warped Gaussian processes. Jordi Munoz-Marí, Jochem Verrelst, Miguel Lazaro-Gredilla and Gustau Camps-Valls. IEEE International Geoscience and Remote Sensing Symposium 2015 (IGARSS 2015). 26-31 July, 2015, Milan, Italy.
- Retrieval of Biophysical Parameters with Heteroscedastic Gaussian Processes. Miguel Lázaro-Gredilla, Michalis Titsias, Jochem Verrelst and G. Camps-Valls. IEEE Geoscience and Remote Sensing Letters, 11(4), 838-842, 2014.
- Global and time-resolved monitoring of crop photosynthesis with chlorophyll fluorescence. Luis Guanter, Yongguang Zhang, Martin Jung, Joanna Joiner, Maximilian Voigt, Joseph A. Berry, Christian Frankenberg, Alfredo Huete, Pablo Zarco-Tejada, Jung-Eun Lee, M. Susan Moran, Guillermo Ponce-Campos, Christian Beer, Gustavo Camps-Valls, Nina Buchmann, Damiano Gianelle, Katja Klumpp, Alessandro Cescatti, John M. Baker, and Timothy J. Griffis. Proceedings of the National Academy of Sciences, PNAS, 2014.
- Toward a Semiautomatic Machine Learning Retrieval of Biophysical Parameters. Juan Pablo Rivera, Jochem Verrelst, Jordi Muñoz-Marí, José Moreno, and G. Camps-Valls. IEEE Journal of Selected Topics in Applied Observations and Remote Sensing, 2014. DOI: 10.1109/JSTARS.2014.229875.
- Prediction of Daily Global Solar Irradiation using Temporal Gaussian Processes. Sancho Salcedo-Sanz, Carlos Casanova-Mateo, Jordi Muñoz-Marí, G. Camps-Valls. IEEE Geoscience and Remote Sensing Letters, 11(11), 1936-1940, Nov. 2014. 10.1109/LGRS.2014.2314315.
- An ensemble of global high-resolution products of energy fluxes over land. Martin Jung, Kazuhito Ichii, G. Camps-Valls, Dario Papale, Gianluca Tramontana, Sven Sickert, Christopher Schwalm, Markus Reichstein. 7th International Scientific Conference on the Global Water and Energy Cycle, GEWEX 2014. The Hague, the Netherlands, 14-17 July 2014.
- Advanced retrieval methods for leaf chlorophyll content in support of global mapping of vegetation fluorescence. Jochem Verrelst, Juan Pablo Rivera, Jordi Muñoz, Luis Alonso, G. Camps-Valls, Jose Moreno. GV2M (Global Vegetation Monitoring and Modelling) February 3-7 2014, Avignon, France.
- Advances in Non-linear Retrievals for IASI and MTG-IRS Hyperspectral Infrared Sounders. G. Camps, Valero Laparra, Jordi Muñoz, Luis Gómez, Xavier Calbet, Tim Hultberg, Thomas August. EUMETSAT Meteorological Satellite Conference, EUMETSAT 2014. Geneva, Switzerland, 22 - 26 September 2014.
- Gaussian processess Uncertainty Estimates in Experimental Sentinel-2 LAI and Leaf Chlorophyll Content Retrieval. Jochem Verrelst, Juan Pablo Rivera, José Moreno, G. Camps-Valls. ISPRS Journal of Photogrammetry and Remote Sensing, Volume 86, Dec. 2013.
- Gaussian process retrieval of chlorophyll content from imaging spectroscopy data. J. Verrelst, L. Alonso, P. Rivera, J. Moreno and G. Camps-Valls. IEEE Journal of Selected Topics in Applied Earth Observation and Remote Sensing, 6(2), Part 3, 2013.
- Mapping cropland GPP in the north temperate region with space measurements of chlorophyll fluorescence. L Guanter, Y Zhang, M Jung, J Joiner, M Voigt, AR Huete, P Zarco-Tejada, C Frankenberg, J Lee, JA Berry, SM Moran, G Ponce-Campos, C Beer, G Camps-Valls, NC Buchmann, D Gianelle, K Klumpp, A Cescatti, JM Baker, T Griffis. AGU Fall Meeting Abstracts, 2013.
- Estimation of Vegetation Chlorophyll Content with Variational Heteroscedastic Gaussian Processes. Miguel Lázaro-Gredilla, Michalis K. Titsias, Jochem Verrelst and G. Camps-Valls. IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2013. 21-26 July 2013. Melbourne, Australia.