Position opening: Using machine learning to build a climate data record of soil moisture

Category : Non classé

About CESBIO

The CESBIO (Centre d’Etudes Spatiales de la Biosphère) is a joint research unit of Paul Sabatier University, the Centre National de la Recherche Scientifique (CNRS), the Centre National d’Etudes Spatiales (CNES) and the Institut de Recherche pour le Développement (IRD). The laboratory aims at doing research in the domains of observation and modeling of the continental surfaces, addresses the interface between physical and biological sciences and participates in the specification of space missions and the treatment of remotely sensed data to develop the knowledge on continental biosphere dynamics and functioning at various temporal and spatial scales. CESBIO hosts the PIs for two European Space Agency (ESA) satellite missions (SMOS and Biomass missions) and for the French-Israeli Venus satellite.

The CESBIO is the lead « Expert Support Laboratory » for land applications with SMOS for ESA and « Expertise Center » for the SMOS French ground segment, the CATDS (Centre Aval de Traitement de Données SMOS).

Scientific background

The SMOS (Soil Moisture and Ocean Salinity) satellite is a passive microwave interferometer that is providing L-band (1.4 GHz) observations of the Earth since 2009. SMOS is the first radiometer that performed systematic L-band measurements from the space. It has been followed by the NASA Soil Moisture Active Passive (SMAP) satellite, launched in 2015 and also carrying an L-Band radiometer. Both satellites have a maximum revisit time of 3 days at the equator.

The soil moisture content is an essential climate variable that has to be monitored over long periods of time (~30 years), however due to the limited lifetime of satellite missions, several satellites are needed to reach this goal. ESA has implemented a Climate Change Initiative (CCI) with a soil moisture chapter. Based on its expertise in using neural networks to retrieve soil moisture from multi-source data (Rodriguez-Fernandez et al. 2015, 2017a, 2017b, 2018), the passive radiometry team of CESBIO has shown that machine learning techniques such as neural networks can be used to build long time series using SMOS as reference to retrieve soil moisture from other sensors such as the ASMR-E radiometer (Rodriguez-Fernandez et al. 2016). This data set, which provides SMOS-like soil moisture data set since 2003, is a level 4 product distributed by CATDS.

Aim of this work

The main goal of this position is to improve the long time series level 4 product by CATDS. By construction, the AMSR-E neural network product shows no global bias with respect to the Level 3 SMOS soil moisture CATDS product but local bias may exist. A systematic search will be done and several methods will be studied to correct possible local bias. In addition, a new machine learning algorithm will be proposed to extend the time series back to the 80’s using data from other multifrequency radiometers such as TRMM/TMI (from 1997) and those of the SSM/I family (the first one was launched in 1978). The reference soil moisture dataset should be the Level 3 SMOS soil moisture product by CATDS or a merged SMOS/SMAP product. For this goal, a comparison of SMOS/SMAP brightness temperatures should be performed and a consistent retrieval algorithm will be proposed. Results of these analyses are opened for publication.

Expected profile of the applicants:

The candidate should ideally have a PhD in applied mathematics or physics or an Engineering degree with experience in data science. Knowledge in remote sensing (ideally in microwave radiometry) is an asset. The applicant should also be rather autonomous, and creative. The ability to read and write technical documentation in English language is required. The post-doctoral fellow will be based in Toulouse but he will work in collaboration with other international scientists and ESA, meaning that the applicant should fully master English as as spoken language.

Salary will depend on qualifications and experience following the CNRS grid. Social security benefits are provided. Inquires and applications (resume and motivation letter) should be sent by e-mail before 31/Jan/2019 to the following contacts list:

Contacts:

Nemesio Rodriguez-Fernandez
CESBIO
18 avenue. Edouard Belin,
bpi 2801,
31401 Toulouse cedex 9
Tel: 05 61 55 85 22
email: nemesio.rodriguez@cesbio.cnes.fr

Philippe Richaume
CESBIO
18 avenue. Edouard Belin,
bpi 2801,
31401 Toulouse cedex 9
Tel: 05 61 55 74 87
email: philippe.richaume@cesbio.cnes.fr

References:

  • Rodríguez-Fernandez, N. J., de Rosnay, P., Albergel, C., Aires, F., Prigent, C., Richaume, P., Kerr, Y., & Drusch, M. (2018, July). SMOS Neural Network Soil Moisture data assimilation. In International Geoscience and Remote Sensing Symposium (IGARSS), 2018 IEEE (pp. 5548-5551)
  • Rodríguez-Fernández, N. J., de Souza, V., Kerr, Y. H., Richaume, P., & Al Bitar, A. (2017, July). Soil moisture retrieval using SMOS brightness temperatures and a neural network trained on in situ measurements. In International Geoscience and Remote Sensing Symposium (IGARSS), 2017b,IEEE (pp. 1574-1577)
  • Rodríguez-Fernández, N. J., Muñoz-Sabater, J.; Richaume, P.; Albergel, C.; de Rosnay, P. & Kerr, Y. H. Evaluation of the SMOS Near-Real-Time soil moisture, Hydrology and Earth System Sciences, 2017a, 21, 5201-5216

  • Rodriguez-Fernández, N. J., Kerr, Y. H., van der Schalie, R., Al-Yaari, A., Wigneron, J. P., de Jeu, R., et al. (2016). Long Term Global Surface Soil Moisture Fields Using an SMOS-Trained Neural Network Applied to AMSR-E Data. Remote Sensing, 8(11), 959.

  • Rodriguez-Fernandez, NJ, Aires F., Richaume, P., Kerr, Y. Prigent, C,, Kolassa, J., Cabot F. Jimenez, C., Mahmoodi, A., Drusch, M. (2015) Soil moisture retrieval using neural networks: application to SMOS, IEEE Transactions on Geoscience and Remote Sensing, 53(11), 5991-6007

Post a comment

download wordpress themes