Publications

phd students/postdoc under my supervision or co-supervision

Preprints

Peer-reviewed journal articles

Questioning the “glacier as water resource” narrative

Glaciers are often portrayed as an important water resource in public debate. However, the scientific literature calls for caution when presenting this idea. While shrinking glaciers can increase water insecurity, their contribution to runoff is often minor.


Figure: Respective contribution of rain, snow, snow on ice, and exposed glacier ice to runoff in three large Asian catchments above 2000 m altitude (from Armstrong et al., 2018).

How much snow "disappears" into the atmosphere? 

The Indus basin is considered as the one with the highest dependence on snowmelt runoff in High Mountain Asia. Over 2000–2016, basin-average snowmelt was 101 ± 11 Gt.a−1 (121 ± 13 mm.a−1), which represents about 25–30% of basin-average annual precipitation. Snow sublimation accounts for 11% of the mean annual snow ablation, but with a large spatial variability across the basin. 

Code to reproduce the study: https://github.com/sgascoin/HMA-Snow-Reanalysis-scripts

Estimating the snow cover fraction in operational context from Sentinel-2 

We propose a simple equation to compute the snow fraction from Sentinel-2 and evaluate its accuracy using several datasets including high resolution satellite data, terrestrial cameras, lidar scans and crowd-sourced ODK data collect.

Snow depth from very high resolution stereo satellite

In this study we continued the work of Marti et al. (2016) with much better validation dataset and demonstrate that satellite photogrammetry stands out as a convenient method to estimate the spatial distribution of snow depth in high mountains.

At what resolution should I run my distributed snow model? 

We studied the sensitivity of an energy balance model to the resolution of the model grid in the High Atlas. We used a time series of 8‐m resolution snow cover area maps with an average revisit time of 7.5 days to evaluate the model results. The digital elevation model was generated from Pléiades stereo images and resampled from 8 to 30, 90, 250, 500, and 1,000 m. The results indicate that the model performs well from 8 to 250 m but the agreement with observations drops at 500 m. This is because significant features of the topography were too smoothed out to properly characterize the spatial variability of meteorological forcing, including solar radiation. This result is consistent with the shape of the semivariogram of the topographic slope.

Figure 1 of this paper was featured in cover of WRR issue 7 vol. 56.

Theia snow products

The Sentinel-2 mission observes the land surface at 20 m every 5 days. The frequency of observations can be further increased with Landsat-8. In this paper, we describe and evaluate the Theia Snow collection, a collection of snow maps made from Sentinel-2 and Landsat-8. The data are routinely produced over several mountain areas and freely distributed via http://theia.cnes.fr. 

Modeling the snow water equivalent in the High-Atlas without station data

Rivers from the High-Atlas provide key water resources to the semi-arid region of Marrakech. In this paper we assess a method to compute in near real time (< 1 month) the snow water equivalent without weather station data. The study area is the catchment of the snow dominated Ourika river, a major tributary of the Tensift river.

Monitoring the decline of glaciers in the Pyrenees using Pleaides

There remain three glaciers around the emblematic Vignemale peak in the French Pyrenees: Ossoue, Oulettes and Petit Vignemale. We used digital elevation models from Pleaides stereoscopic imagery to compute their elevation changes from 2013 to 2017. 

Lire la news en français sur le site du Cnes : Les satellites Pléiades suivent le déclin de glaciers des Pyrénées françaises

Aru mega-avalanches 

An international collaboration to understand the mechanisms behind these extraordinary events. This work was featured in many news media (NYT, SA), my favorite is the story by A. Voiland in Nasa's blog Earth Matters. Otherwise you can read my blog posts here. 

Un article en français est paru dans Pour la Science.

Snow observations in Lebanon 

Although the snow melt is a critical water resource in Lebanon, there is no operational in situ observations of the climate conditions in the high elevation areas, and very few available snow measurements. This article describes new weather station data, snow course surveys and remote sensing observations of the snow cover area in Mount-Lebanon. The data are archived and publicly available in Zenodo. Snow courses data were collected by Abbas Fayad (photo) as part of his PhD (CESBIO/CNRS-L, IRD grant). See also the field work section.

Fifty shades of grey 

In Iceland, large variations of the ice caps reflectivity (albedo) may occur due to the deposition of volcanic ash. Here we evaluated the MCD43 MODIS albedo product using measurements from ten automatic weather stations on Vatnajökull and Langjökull ice caps. The MODIS data show that the albedo decreased significantly after the 2010 Eyjafjallajökull and 2011 Grímsvötn eruptions on all the main ice caps except the northernmost Drangajökull. Large reductions of the summer albedo by up to 0.6 is observed over large regions of the accumulation areas, causing a strong increase in the mass losses of the ice caps during these years (Bjornsson et al., 2013)

Why precipitation does not always increase with elevation in the dry Andes of Chile?

In the arid north-central Chile, crop irrigation relies on the runoff from high elevation areas because the Andes Cordillera enhances the precipitation. However, we found that this orographic enhancement can vary significantly and even being canceled under specific meteorological conditions. These conditions are well captured by the Froude number of the atmospheric flow computed from climatic reanalyses.

DSCN5548.JPG

Measuring the volume of snow from satellite remote sensing

To date, there is no definitive approach to map snow depth in mountainous areas from spaceborne sensors. We used very-high-resolution stereo satellites imagery (Pléiades) to generate a map of snow depth in a small Pyrenean catchment. The validation results are promising and open the possibility to retrieve the snow depth at a metric horizontal resolution in remote mountainous areas, even when no field data are available. 

Articles grand public (en français) sur le site de l'INSU, sur le site du CNES et dans la Tribune Toulouse. 

The history of a pyrenean glacier

The evolution of the glacier d'Ossoue since 1850 was reconstructed based on historical documents, maps, field surveys and satellite imagery. The glacier has receded considerably since the end of the Little Ice Age, losing 40% of its length and 60% of its area. Assuming that the current ablation rate remains constant, Ossoue Glacier will disappear midway through the 21st century.

When and where is there snow in the Pyrenees?

We generated a gap-filled snow cover climatology for the Pyrenees from MODIS snow products. This figure shows the mean monthly snow cover duration that was calculated over 2000-2013 in four elevation bands. 

In addition we showed that a deficient snowpack in 2012 seems to have caused a drop in the national hydropower production.

image

More trees, less snow?

The Pyrenees are expected to undergo strong environmental perturbations over the 21st century because of climate change (rising temperatures) and the abandonment of agro-pastoral areas (reforestation). Both changes are happening at similar timescales and are expected to have an impact on snow cover. We analyzed the response of snow cover to a combination of climate and land cover change scenarios in a small Pyrenean catchment (Bassiès). Model projections indicate that both climate and land cover changes reduce the mean snow depth. However, the impact on the snow cover duration is moderated in reforested areas by the shading effect of trees on the snow surface radiation balance.

Looking at catchments in colors

Agro-hydrological models are heavily over-parameterized because they rely on a spatially-distributed representation of individual processes such as crop growth, nitrogen uptake, soil water flow... High-resolution data from Sentinel-2 mission data may provide useful information to better constrain the vegetation and soil parameters in such models.

Ossoue glacier DEM from Pléaides

Elevation differences between DGPS measurements (dots) and the model generated from a stereo pair acquired by Pleaides 1A satellite on the Ossoue glacier (Pyrenees, France)

Pourrier et al. 2014 Fig.2

From glacier melt to streamflow

In the semi-arid Andes most of the water resource originates from the high-elevation areas. This map shows the assemblage of a valley glacier (TG), a debris-covered glacier (DCG), two rock glaciers (RG, TPRG), and moraines (MC, OM) in the Tapado watershed (4000-5550 m asl). Each of these units shapes the hydrological response of the watershed to the climate.

Photo S. Ferrant

Irrigation et changement climatique en Inde du Sud

Lire la brève sur le site INSU Irrigated rice field (photo S. Ferrant)

Aussi repris dans le rapport annuel 2014 du CNRS

MODIS SCF versus maximum mean snow depth

MODIS snow cover fraction vs. snow depth observations

This graph shows that the snow cover extent varies faster during accumulation than during ablation periods in an alpine catchment. This hysteretic behavior in the snow cover depletion curve was introduced in the Catchment land surface model to improve the melting dynamics.

Wind transported snow (saltation and suspension) over the glacier area in Pascua-Lama

This map shows the quantity of snow transported by the wind as simulated by SnowTran-3D over a snow season in Pascua-Lama. The glacier contours are drawn in blue. We found that net deposition occurred for 43% of the glacier grid points, whereas it is only 23% of non-glacier grid points located above the minimum glacier altitude (4475 m). Snow transport may be a key "recharge" mechanism for glaciers, as it means that when snowfall is low in the area, glaciers would still receive preferential accumulation of drifting snow. We also found that the sublimation of blowing snow amounts to 18% of the total ablation over the study area.

How much runoff comes from the glaciers?

This figure shows that the mean annual glacier meltwater contribution to streamflow is greater than the glacier fractional area in five gauged catchments of the upper Huasco river, in the semi-arid Andes of Chile.

Soil albedo vs. top 5-cm soil moisture

This figure shows the negative correlation between the soil albedo and the soil water content in the top 5 cm of the Zongo glacier moraine. The average snow-free albedo value during the rainy season is 40% lower than during the dry season (0.16 vs. 0.26).

Books, chapters, reports

Not peer-reviewed

Science-fiction