Snow cover monitoring in the Northern Patagonia Icefield using MODIS satellite images (2000–2006)
Introduction
The Northern Patagonia Icefield (NPI) and the Southern Patagonia Icefield (SPI) are important in evaluating the impact of climate change in the southern hemisphere (Casassa et al., 2002). They represent the largest temperate ice masses of South America (Warren and Sugden, 1993) and provide an important source of fresh water. Rignot et al. (2003) showed that the average ice thinning rates more than doubled during the 1995–2000 period compared to the average rate over 1975–2000, reaching an equivalent sea level rise of 0.105 ± 0.011 mm per year. To improve our understanding of the response of glaciers in South America to climate change, several studies have been conducted in Patagonia, most particularly on the NPI. Some of them consider surface monitoring, front and volume changes with remote sensing data such as aerial photographs or high-resolution satellite images (Lliboutry, 1956, Aniya, 1988, Aniya, 2007). It has been suggested that the origin of the thinning, retreat, and loss of glacier surface could result from a change in the icefield's snow accumulation regime (Aniya, 1988), however in order to confirm this suggestion, mass balance measurements are necessary on the NPI.
To monitor glacier mass balance, either field measurements (ablation stakes and accumulation pits) or differential analysis of the multi-temporal Digital Elevation Model (DEM) are required (Bamber and Payne, 2004, Berthier et al., 2004). Applying both of these methods to the NPI is a complex issue because access to the area is difficult and there is a paucity of DEM. Nevertheless, given the sensitivity of snow cover to precipitation and temperature, its study can provide more insights on the evolution of the icefield accumulation regime. The main problem in monitoring the NPI snow cover is the difficulty in making field observations because of unfavourable meteorological conditions (Peña and Escobar, 1987) as well as the large size of the icefield. Remote sensing offers good tools to evaluate the earth's conditions and therefore is particularly appropriate to analyse such large ice bodies.
Romanov and Tarpley (2003) analysed seasonal changes of the snow covered area of South America between May 2000 and November 2001 using data from Geostationary Operational Environmental Satellite 8 (GOES-8) Imager at 1 km and 4 km of spatial resolution. Foster et al., 2002, Foster et al., 2003 analysed the seasonal snow cover extent and depth (for the 1988–2001 and 1979–2001 periods, respectively) in the middle latitudes of South America using data from the Scanning Multichannel Microwave Radiometer (SSMR) instrument onboard the Nimbus 7 satellite and from the Special Sensor Microwave Imager (SSMI) sensors onboard Defense Meteorological Satellite Program (DMSP) satellites. These studies have provided important information on the seasonal behaviour of the snow extent in South America in a global scale. The spatial resolution of the sensors used on those studies is not refined enough for studying the snow cover extent of the NPI.
Snow cover has a highly dynamic behaviour (Lliboutry, 1956) and therefore needs to be monitored frequently. The NPI is very often covered by clouds, and as a consequence, only very few high-resolution images are available. In order to partially address this issue, the Moderate resolution Imaging Spectroradiometer (MODIS) onboard the Terra spacecraft was used because it provides imagery with a repeat time (2 days) and resolution (250, 500 and 1000 m) that are suitable for monitoring the snow cover extent of the NPI.
The National Snow and Ice Data Center distributes the MOD10A1 Snow Products which are daily snow cover extent maps based in MODIS images. The MODIS MOD10A1 Snow Products make it possible to study the daily changes in snow cover. However, they fail to map some snow because of confusion between clouds and snow due to the MOD35 cloud mask which is an input to the MOD10A1 algorithm (Hall et al., 2002). This may hamper the analysis of the snow cover distribution (Romanov and Tarpley, 2003). The capabilities of MODIS Snow Products to identify snow and ice in the NPI were evaluated by testing the MODIS MOD10A1 daily Snow Products in the area of study. We confirmed the fact that too many pixels were mapped as cloud when they should have been mapped as snow. Therefore, the monitoring of the snow cover extent of the NPI over time was derived from raw cloud-free MODIS-Terra data sets over the 6-year period (2000–2006). Due to complex terrain, a topographical correction had to be applied to the MODIS-Terra images.
The objective of this study was to analyse the changes in the extent of the snow cover on the NPI and its relations with the atmospheric conditions. Meteorological data were analysed to elucidate the source of the snow cover fluctuations. This required the icefield to be divided into a western and an eastern side based on the topographic and climatic differences between the two sides of the NPI.
Section snippets
Study area
The Northern Patagonia Icefield (Fig. 1) is located between 46°30′ and 47°30′ of the southern latitudes. It covers a total surface area of 4197 km2 (including rock outcrops) and it extends for nearly 125 km north–south between Grosse and Steffen Glaciers, with a maximum width of 71 km in the west–east direction between the frontal tongues of San Quintín and Soler Glaciers (Rivera et al., 2007).
The highest point is the San Valentin summit, 3910 m above sea level (asl) which is located in the
MODIS images
MODIS images belong to the Earth Observation System (EOS) and they were used for this study, because they offer a suitable temporal, spectral and spatial resolution (Table 1). MODIS has 36 bands and is a cross-track-scanning imaging radiometer. The instrument field of view is ± 55° from the nadir. The sensor observes the earth from a sun-synchronous position near the polar orbit at an altitude of 705 km. MODIS-Terra presents global coverage in 2 days. The datasets used are Level 1 B products
Snow cover extraction method
The first task was to evaluate the accuracy of MODIS Snow Products for snow and ice identification in the NPI. Ten MODIS Snow Products (MOD10A1) of different months and years were compared with their corresponding raw images (Fig. 3). In every MODIS Snow Product tested, the whole NPI was mapped as a cloud area when it is really snow and ice as seen on the Fig. 3A.
It is not the purpose of this study to make a thorough analysis of the performance of MODIS MOD10A1 Snow Products. However, on the
Snow extent changes over the entire NPI
This study established the frequency with which the icefield was totally cloud-covered, partially cloud-covered or free of clouds. On average during the 2000–2006 period, the entire icefield was covered by clouds 273 days per year, partially cloud-covered (at least part of the icefield covered by clouds, very often the western side) 73 days per year, and free of clouds 19 days per year. The cloudiest months of the year were January, September and December. The year least [most] cloud-covered
Concluding remarks
The snow cover extent changes between 2000 and 2006 were successfully monitored applying the NDSI and Red/NIR ratio to cloud-free MODIS images. The results showed a seasonal variation in snow cover extent, with a maximum snow cover extent totalling an area of 11,623 km2 (August 2001) and a minimum extension of 3600 km2 (March 2000). Over the yearly cycle, the snow cover extent correlated well with the Reanalysis air temperature (R2 = 0.76). Large fluctuations in snow cover extent over the entire
Acknowledgments
Paulina Lopez conducted this study within a PhD program sponsored by a IRD (Institut de Recherche pour le Développement) scholarship. Two Chilean institutions were associated with this study: the Centro de Estudios Cientificos de Santiago (CECS), which provided the mosaic based on Landsat ETM+ images and the Direccion General de Aguas (DGA) de Aysén.
The MODIS images and the MODIS Snow Products used in this study were acquired as part of NASA's Earth Science Enterprise. The algorithms were
References (42)
Spectral signatures of alpine snow cover from the Landsat Thematic Mapper
Remote Sensing of Environment
(1989)- et al.
Development of Methods for Mapping global snow cover using Moderate Resolution Imaging Spectroradiometer Data
Remote Sensing of Environment
(1995) - et al.
MODIS snow-cover products
Remote Sensing of Environment
(2002) - et al.
High-latitude surface temperature estimates from thermal satellite data
Remote Sensing of Environment
(1997) - et al.
Evaluation of topographic effects in remotely sensed data
Remote Sensing of Environment
(1989) - et al.
Ice elevation and areal changes of glaciers from the Northern Patagonia Icefield, Chile
Global and Planetary Change
(2007) - et al.
Estimating fractional snow cover from MODIS using the Normalized Difference Snow Index
Remote Sensing of Environment
(2004) Glacier inventory for the Northern Patagonia Icefield, Chile and variations 1944/45 to 1985/86
Arctic and Alpine Research
(1988)Glacier variations of Hielo Patagonico Norte, Chile, for 1944/45–2004/05
Bulletin of Glaciological Research
(2007)- Bamber, J., Payne, A. (Dir.), 2004. Mass Balance of the Cryosphere. Observations and Modelling of Contemporary and...
Satellite monitoring of equilibrium lines in Northern Patagonia Icefield: 1979 satellite monitoring of equilibrium lines in Northern Patagonia Icefield: 1979–2003
Recent rapid thinning of the Mer de Glace glacier derived from satellite optical images
Geophysical Research Letters
An assessment of the differences between three satellite snow cover mapping techniques
Hydrological Processes
Reconstructed North American, Eurasian, and Northern Hemisphere snow cover extent, 1915–1997
The Patagonian Icefields. A Unique Natural Laboratory for Environmental and Climate Change Studies
Satellite discrimination of snow/cloud surfaces
International Journal of Remote Sensing
Monitoring snow-cover dynamics in Northern Fennoscandia with SPOT VEGETATION images
International Journal of Remote Sensing
La prima traversata della Cordigliera Patagonica del lago Argentino al fiordo Falcon
La Vie d'Italia e del Mondo
Recent climate fluctuations in Patagonia
The shuttle radar topography mission
Reviews of Geophysics
Cited by (49)
Human impact changes hydrological connectivity in a Patagonian fluvial basin
2023, Journal of Hydrology: Regional StudiesAssessment of daily MODIS snow cover products to monitor snow cover dynamics over the Moroccan Atlas mountain range
2015, Remote Sensing of EnvironmentCitation Excerpt :In addition, Driouech, Déqué, and Sánchez-Gómez (2010) have highlighted a possible drop of up to 15% in winter precipitation, with an increase in dry periods by 2050 over Morocco. This possible change that could be associated with a shift in time of the snow season (Lopez, Sirguey, Arnaud, Pouyaud, & Chevallier, 2008) would have a significant impact on the future role of “water tower” played by snow in the South Mediterranean area. Regarding the importance of water held in snowpack over the Moroccan Atlas Mountains, there is thus an imperious need for better monitoring of snow-covered areas.
Sequence-based mapping approach to spatio-temporal snow patterns from MODIS time-series applied to Scotland
2015, International Journal of Applied Earth Observation and GeoinformationAnalysis of sub-pixel snow and ice extent over the extratropical Andes using spectral unmixing of historical Landsat imagery
2014, Remote Sensing of EnvironmentCitation Excerpt :For snow and ice, several tools have been developed that help to directly or indirectly estimate the presence and volume of snow over mountainous regions. The NASA MODIS (Moderate Resolution Imaging Spectroradiometer) product suite (Hall, Riggs, Salomonson, DiGirolamo, & Bayr, 2002) has been used for snow covered area estimation and applied to several hydrological analyses, including snowmelt modeling (Tekeli, Akyürek, Arda Şorman, Şensoy, & Ünal Şorman, 2005), hydrologic modeling (Parajka & Blöschl, 2008; Powell, Blesius, Davis, & Schuetzenmeister, 2011; Ragettli, Cortés, McPhee, & Pellicciotti, 2013) and glacier monitoring (König et al., 2001; Lopez, Sirguey, Arnaud, Pouyaud, & Chevallier, 2008). However, the original MODIS suite of products has retrieval issues such as failing to detect snow presence in the higher elevations at the end of the ablation season, which can result in biases (Rittger, Painter, & Dozier, 2013) or underestimation of the actual snow covered areas due to the use of a binary classification of each pixel instead of more precise fractional estimates.
Spatiotemporal changes in snow cover and their relationship with drought events in the Lake Urmia basin
2024, Hydrological Sciences Journal