PUNTOS ANALISIS
6. PLAN ESTRATÉGICO
6.2 RECURSOS DEL PLAN ESTRATÉGICO
6.2.4 Movimiento Y Patio De Contenedores
for large-scale monitoring
In contrast to terrestrial measurements, satellite-based sensors can cover large areas, which makes them ideally suited for monitoring the remote and largely unpopulated permafrost areas in the Arctic. Since the launch of the first weather satellites in the 1960s, the capabilities of earth observation satellites have ex- panded rapidly. In 2009, the European Space Agency initiated the “DUE Per- mafrost” project (www.ipf.tuwien.ac.at, 2010), which is dedicated to providing a range of new remote sensing products for permafrost research. Despite of the manifest potential in this approach, many remote sensing products are still in the validation stage in arctic regions, so that terrestrial measurements are even more required to achieve a satisfactory performance of satellite-based earth sys- tem monitoring in permafrost areas. In the following, a short synopsis of remote sensing applications in permafrost research is given, with a particular focus on the surface energy budget and the requirements of a large-scale permafrost mon- itoring scheme.
Radiation: In case of cloud-free conditions, the outgoing short- and long-wave radiation are accessible through various remote sensing platforms, that operate at different temporal and spatial resolutions. As the radiation is modified by scattering, absorption and emission in the atmosphere of the earth, a radiomet- ric correction similar to the scheme applied for the thermal imaging system (see Sect. 2.7) is required. However, as satellite sensors perform spectrally resolved
measurements, e.g. 36 spectral bands between 400 nm and 15 µm for MODIS (King et al., 1992), a coincident characterization of the atmosphere and clouds is possible by combining the information from spectral bands, that are modi- fied differently in the atmosphere. With Terra and Aqua MODIS, atmospheric profiles of temperature and water vapor as well as information on aerosols and ozone can be retrieved with a spatial resolution of 5 km for clear-sky condi- tions (Seemann et al., 2003; Gao and Kaufman, 2003). With a radiation trans- fer code (see Sect. 2.6), both incoming short- and long-wave radiation could thus be evaluated. The outgoing long-wave radiation is derived in the context of LST evaluation from satellite sensors. The albedo and even Bidirectional Reflectance Distribution Functions (BRDF) are available from several satellite sensors, e.g. from MODIS at a resolution of 500 m (Schaaf et al., 2002), so that the outgoing short-wave radiation can be evaluated. Therefore, the radiation budget under clear-sky conditions can in principle be evaluated from satellite measurement.
Prolonged periods with cloudy conditions, where measurements from satellites are not available, are a strong limitation for determining the radiation budget through remote sensing in the Arctic. In this thesis, the implications of the fre- quent cloud cover for the accuracy of remotely sensed land surface temperatures have been demonstrated (see Sect. 4.5). Despite of the great potential of re- mote sensing, terrestrial radiation measurements should be routinely conducted at many more sites in the Arctic to support both the improvement of satellite retrieval algorithms and modeling schemes.
Turbulent fluxes: A number of studies have evaluated possibilities to derive sensible and latent heat fluxes from remote sensing applications (e.g. Hall et al., 1992; Bastiaanssen et al., 1998; Friedl, 2002). However, as direct measurements of turbulent fluxes are not possible from satellites, the approaches are based on the surface energy budget equation or even make use of empirically derived re- lations in terms of e.g. surface temperature and soil moisture. Therefore, direct measurements of turbulent fluxes on the ground cannot be replaced by remote sensing at the moment.
Soil properties: Similarly, direct measurements of the ground heat flux or the soil composition from satellites are not possible. However, a few promising approaches have been made to characterize some of the soil parameters that are required for modeling the thermal dynamics of the soil:
• The surface soil moisture can be assessed from passive or active microwave sensors, e.g. AMSR–E (Njoku et al., 2003) or Envisat/ASAR (Baghdadi et al., 2006). However, the result is not independent of the surface cover, so that it is problematic to derive absolute values for the volumetric soil water content, which would be required to determine soil thermal proper- ties. The penetration depth of the signal varies depending on the surface cover and the soil water content, so that the measured volume cannot be determined properly. Furthermore, typical penetration depths are on the order of a few centimeters, so that the by far largest part of the relevant soil region is not covered. Despite of these limitations, space-borne microwave sensors can offer an extremely valuable assessment of the soil moisture
regime in arctic regions, that may become of fundamental importance to improve modeling of the soil thermal regime.
• A highly practical approach is to assign representative soil properties to certain land cover elements, e.g. vegetation communities (e.g Walker et al., 2003). In many cases, land cover classes can be distinguished by their spec- tral characteristics in high-resolution multispectral images or the backscat- ter signature in Synthetic Aperture Radar images (e.g. Haack and Bechdol, 2000). This procedure inherently assumes that land cover classes deter- mined from surface characteristics are good indicators for subsurface prop- erties, which is not necessarily true. Furthermore, rather labor-intensive field campaigns are indispensable, which on the one hand must provide suitable training data sets for classifying the satellite images, while the representative thermal properties of the land cover classes must be evalu- ated on the other hand. Due to the wide range of ecosystems and climatic conditions found in permafrost areas, a pan-arctic classification scheme appears questionable. Therefore, regional landcover classes must be suc- cessively developed in a series of field campaigns in order to claim a con- vincing benefit for modeling. For Svalbard, a landcover map compiled from multispectral images of the Landsat Thematic Mapper is available (www.npolar.no, 2010). With 27 land cover classes, the map can be consid- ered an excellent starting point for modeling efforts, although the study does not assign thermal properties of the soil to the land cover classes. Considering the relative homogeneity of the surface cover on Svalbard compared to other permafrost areas, a supreme long-term effort would be required in order to accomplish such detailed work on a pan-arctic scale. Snow: The snow cover of the earth has been the target of remote sensing missions since the 1970s. Four different applications have a great potential for permafrost studies.
1. The presence of a snow cover can be determined with a high spatial reso- lution by multispectral images in the visible and near-infrared range of the electromagnetic spectrum (Hall et al., 1995). The formation and the dis- appearance of the perennial snow cover can therefore be determined with high temporal precision from satellite sensors such as MODIS (Hall et al., 2002). As with measurements of the land surface temperature, the cloud cover is the limiting factor for the temporal resolution. As the termination of the snow melt period with the associated change of the surface albedo is a crucial event in the annual cycle of the surface energy budget and the thermal dynamics of the permafrost, snow cover products may be critical for the success of permafrost monitoring schemes (see Sect. 5.3.1). 2. The snow water equivalent is accessible through space-borne passive mi-
crowave sensors, such as SMMR (1978–1987, Nimbus–7 satellite), SSM/I (1987–2002, DMSP satellites) and AMSR–E (2002–2010, Aqua satellite). These satellites record microwave emission from the ground due to black- body radiation in frequency bands, that experience a different amount of attenuation from a snow cover (Foster et al., 1984), e.g. frequency bands centered around 19 and 39 GHz for the AMSR–E algorithm (Kelly et al.,
2003). Due to the weak signal of the emitted microwave radiation, passive microwave sensors must integrate over large areas, which leads to lim- ited spatial resolutions of 25 km or more. For the study area, meaningful results for the snow water equivalent cannot be obtained, as sizable frac- tions of ocean and glaciers contribute to the footprint area of the satellite sensor. However, in less heterogeneous regions, the capabilities of passive microwave sensors to evaluate the snow water equivalent have been demon- strated in a number of studies (e.g. Derksen et al., 2003, 2005; Pulliainen, 2006).
3. The onset of the snow melt period can be obtained from active microwave sensors, such as QuickScat/SeaWinds (Long and Hicks, 2000), with which dry and wet snow can be distinguished. By exploiting the high temporal resolution of several measurements per day, Bartsch et al. (2007) developed a method to determine the first snow melt events as well as the end of daily freeze-thaw cycles on the snow surface. As with (1), such data may play an important role in permafrost monitoring schemes.
4. Bartsch et al. (2010) and Bartsch (2010) report the detection of rain-on- snow events in the backscatter signal of QuickScat/SeaWinds, which has great potential in permafrost modeling and monitoring, as rain-on-snow events can have a drastic impact on the winter soil temperatures (see Sect. 4.4). The representation of rain-on-snow events has been identified as the main limitation in the SEB model, as the employed precipitation record obtained from ERA reanalysis cannot properly resolve their occur- rence (see Sect. 4.6.3). Therefore, data on rain-on-snow events from ac- tive microwave sensors may contribute to progress in modeling the ground thermal regime in permafrost regions, that are affected by rain-on-snow events.