MuSenALP: Multi-Sensor Application for LSWT Processing

The MuSenALP project facilitated a Lake Surface Water Temperature (LSWT) plugin for ESA’s Sentinel Applications Platform (SNAP) and data from a variety of spaceborne thermal radiometers. It was carried out as a Swiss ‘Mésure de Positionnement’ in 2016-2018.

LSWT is the first Essential Climate Variable for lakes in WMO’s Global Climate Observation System (GCOS), and a key driver for aquatic ecosystem processes. Using an extensive collection of AVHRR imagery from NOAA and MetOp satellites, the Remote Sensing Research Group (RSGB) of the University of Bern has a strong focus on climate studies, but provides also near-real time LSWT products for the CORESIM project. Through collaboration with Odermatt & Brockmann, a SNAP plugin was developed that enables such products from several other data sources, and through deployment in distributed computing infrastructures.

The MuSenALP plugin is based on SNAP’s Python library snappy. It is usable via the common SNAP GUI or using the command line tool provided with SNAP. It enables the use of data from AVHRR, Landsat-8 TIRS, MODIS, Sentinel-3 SLSTR and VIIRS. Two different retrieval algorithms are implemented to be selected by the user depending on the number of available thermal bands. Auxiliary data from radiative transfer simulations can be ingested in various manners, including approximation through a regionalization concept. A quality level procedure is included that allows for the improvement of retrieval accuracy through selection of favourable observations. The plugin were installed at a testbed  for bugfixing and initial testing in the scope of several application studies, and will be assessed and developed further in the scope of the project partner’s research and service activities.

The plugin’s multi-sensor capabilities and the adequacy of a regionalized parameterization concept are verified at the example of Lake Geneva. RSGB Scientists compared LSWT from several hundred Sentinel-3 SLSTR datasets to the AVHRR long-term climate time series, with reference to in situ measurements provided by the Aquatic Physics Group of EPFL and for quality levels optimized for AVHRR. The resulting performance for SLSTR is already quite robust and will further improve once sensor-specific qualiy level optimization is available for SLSTR.

Top: LSWT map, retrieved from NOAA-AVHRR sensor and processed with MuSenALP;Bottom: Comparison of AVHRR LSWT (left) and SLSTR LSWT (right) with in-situ measurements at Lake Geneva.

Top: LSWT map, retrieved from NOAA-AVHRR sensor and processed with MuSenALP;Bottom: Comparison of AVHRR LSWT (left) and SLSTR LSWT (right) with in-situ measurements at Lake Geneva.


Further testing included the use of 100 m Landsat-8 TIRS data, which have recently undergone a recalibration process for the correction of straylight artifacts (Gerace and Montanaro, 2017). However the LSWT output was still subject to banding effects, and a significant bias from ferry measurements on Lake Constance was observed. It is thus still not recommended using Landsat-8 for LSWT retrieval, but it is anticipated that the availability of a high-resolution thermal radiometer on Sentinel-8 might soon lift the current limitations in spatial resolution for LSWT products.


Gerace, A., and Montanaro, M. (2017). Derivation and validation of the stray light correction algorithm for the thermal infrared sensor onboard Landsat 8. Remote Sensing of Environment 191, 246–257

Lieberherr, G., Riffler, M., and Wunderle, S. (2017). Performance Assessment of Tailored Split-Window Coefficients for the Retrieval of Lake Surface Water Temperature from AVHRR Satellite Data. Remote Sensing 9, no. 12


University of Bern

Project manager:
Daniel Odermatt

Odermatt & Brockmann

Swiss State Secretariat for Education, Research and Innovation, Swiss Space Office

Mésures de Positionnement 2016