|TSM/TDM Science Team Meeting 2016|
|Science Team Meetings|
|How to Submit a Proposal|
|COFUR Price List (scientific use)|
|TanDEM-X Science Service System|
|Title||Examination of the hydrology of a low-gradient High Arctic wetland at the regional scale|
|Investigator||Muster, Sina - Stiftung Alfred-Wegener-Institut für Polar- und Meeresforschung, Geowissenschaften|
The project’s overall goal is to examine the hydrology of a low-gradient wetland at the regional scale (i.e. Polar Bear Pass, Bathurst Island, ~75°40’N, 98°30’W) to determine its temporal and spatial response to water inputs (meltwater, icemelt and rainfall) and losses (evaporation and drainage).
Goals of the project were to
(i) identify the different morphological and hydrological terrain units (ponds, lakes, late-lying snowbeds, stream channels and plateau areas) at Polar Bear Pass (Fig.1), (ii) map seasonal changes in snow cover extent, duration and melt, surface water boundaries and fluctuations, soil moisture and seasonal thaw depth, extent of wetland components’ hydrologic connectivity prior and during the freeze-back period, (iii) isolate key parameters (e.g. surface roughness, albedo, surface temperature, soil moisture) for the quantification of evapotranspiration and upscale them via remote sensing methods from the plot scale (< 1m) to the meso (1m – 10km) and regional scale (>10km) and (iv) evaluate existing and implement new models for the estimation of evapotranspiration from remote sensing data.
1. Data and methods
1.1 Optical satellite and aerial imagery
Concurrent to TSX imagery, optical satellite and aerial imagery was acquired during 2009 and 2010. Optical satellite data includes following scenes: June 29 (SPOT5), August 3 (ALOS AVNIR) and August 25, 2009 (SPOT5), July 3 (ALOS AVNIR & PRISM) and August 16 (SPOT4) in 2010. In 2009 high-resolution visible and near-infrared aerial images of a transect from the dry ridge to the wetland were taken at one point in time with our paraglider.
1.1 Image processing and image analysis
TSX images with HH polarization were processed using the software GAMMA SAR, rendering geocoded radar intensity images with a final resolution of 5m. Image analysis was done in ENVI 4.7. A density slicing technique was applied to extract the water bodies and inundated surface areas. Density slices were then used for intra-annual change detection. Optical satellite and aerial imagery are used to validate and help interpret the radar signal.
Interferometric processing was done in GAMMA SAR using a DEM to subtract the topography related phase.
1.2 Ground campaigns in 2009 and 2010
Ground measurements were made both in 2009 and 2010 concurrent to image acquisition. The field campaigns’ goal was to quantify the water and energy balance components for the prominent land cover and terrain types, i.e. wet and mesic meadow, hummock seepage slope, mesic ridge and dry polar desert (Fig. 1). In 2009, point measurements as well as spatially integrated measurements were conducted along a transect of about 1km length reaching from the ridge into the wetland measuring surface and ground temperature, surface moisture, and evapotranspiration. In 2010, randomly selected plots in the main land cover units were mapped for an area of 5x5 km.
Ground measurements included
1.1.2 Climate and soil stations 25 HOBO data loggers have successfully measured a full year (July 2009 to July 2010) of surface and soil temperature in the main land cover units. The climate station set up in the wetland provides a 2-year record of air temperature, soil and surface temperature, and surface moisture. Together with Eddy covariance data and point measurements from 2008, energy and water fluxes can be estimated for the ridge and the wetland.
1.1.2 Eddy Covariance station An Eddy Covariance station was set up in the wetland for a period of 6 weeks in 2009. The Eddy station consists of a Campbell CSAT 3D sonic anemometer and a LiCor LI–7500 CO2 and H2O gas analyzer and recorded sensible and latent heat fluxes as well as CO2 fluxes integrating over an area of approximately 200 m². (Fig. 2)
1.1.3 Ground-penetrating radarA multi-channel ground-penetrating radar (GPR) was used to estimate average volumetric soil water content and thaw depth of the active layer. (Fig. 3) The GPR measurements were assisted by punctual time domain reflectometry (TDR) data of volumetric soil water content and thaw depth measurements conducted with a frost probe. Along the main transect, GPR and TDR measurements were done synchronously with a TerraSAR-X overflight in order to relate the ground-based geophysical measurements to remote sensing data.
1.1.4 (Sub)Surface characteristics
For a total of 82 plots of 0.5 m x 0.5 m size, we mapped following parameters within an extent of 5 km² : cover of plant functional type (i.e. shrub, herbaceaous, moss, lichens ) as well as bare ground and litter; vegetation height and microtopography as parameters for surface roughness: active layer depth (ALD); and moss layer thickness. (Fig. 4) Especially cover characteristics and surface roughness affect the backscatter of energy on the ground and thus the interpretation of satellite imagery. interferometry data. SAR interferometry has been successfully applied to TSX data showing displacements in the range of 1 cm in the area, most likely due to the thawing of the soil and the associated volumetric change of ground ice to water. (Fig. 2) In order to better understand the signals, transects were mapped in 2010 ranging from stable areas to areas showing subsidence. At intervals of about 20 m, ALD and moss thickness were measured and volumetric soil samples taken.
2.1 Land water classification
TerraSAR-X images are acquired during arctic summer season, from mid-June until late August, when temperatures rise above zero degrees. They display the annual melting process and show the current state of water bodies at different times during summer. Water bodies on radar images cause a low return. They are presented as dark areas and are very well distinguishable from bare ground and vegetation. At the beginning of snow melt, at mid-June (images from June 19, 2009 and June 17 2010) great areas of the valley are inundated. This is clearly visible on the radar images because of the low return and dark appearance of the inundated areas. The great lakes as well as the southern area show a very low, but not homogenous return on the image acquired on June 19, 2009. The same is visible in the corresponding image one year later, on June 17, 2010. The image acquired two weeks earlier, on June 6, 2010, before the beginning of snow melt, show a very low backscatter and appear dark grey all over the area. That clearly is caused by the low dielectric properties of ice and snow and indicates that snow melt has not begun yet. A comparison with climate data validates the assumption. On the slopes, late laying snow beds with wet surface cause a very low return and appear black between the bright slopes in the image acquired on June 19, 2009. Images acquired later in the season show the ongoing melting process: the late laying snow beds disappear.
The HH-polarization of TSX proves excellent to map water bodies, especially after snow melt is over. During snow melt, however, concurrent ground truth and optical imagery is needed to distinguish between wet snow, wet ice and inundated land areas.
2.2 Change detection
Change detection maps indicate no change in grey, a decrease in surface water in blue and an increase in red. Change detection was conducted for every year as well as between the two years. Change detection in 2009 shows a significant decrease of water content between the images acquired on June 19 and July 11 of about 22 percent. Inundated areas are drained. Between July 11 and July 22 a small increase of water area of about five percent is detectable, whilst between July 22 and August 24 the water area remains almost stable. Since SAR imagery is very sensitive to surface changes, the results have to be interpreted carefully. Water content can vary very widely depending on different influencing factors, such as precipitation events in form of rain, fog or snow, variable daily air temperatures or changing wind conditions. The change detection map between June 6 and June 17, 2010 shows a great increase of water content nearly all over the area of about 45 percent. That indicates the beginning of snow melt. Air temperatures rise about freezing point around June 10, 2010. Between June 17 and June 28, 2010 water area decreases rapidly by about 59 percent. Between June 28 and July 7, 2010 as well as between July 7 and July 31, 2010 water surface amount does not change significantly. The ice covering over the taller lakes is visibly melting. As in 2009, the ice surface seems to get wet recording to a low return, later the water surface becomes visible. Between July 31 and August 11, 2010 no significant changes are observable.
In permafrost landscapes the soil stores a lot of ground ice which thaws in summer. The melting of ice leads to a change of volume which in turn causes the ground to subside. Results at Polar Bear Pass show a subsidence of the ground of about 1 cm. Both 2009 and 2010 show a similar spatial but a different temporal pattern of subsidence. Subsidence maps also show the development of areas with a high decorrelation which indicates a high surface moisture. The timing of ground thaw and consequent subsidence will be evaluated with existing climate and soil data in future work.
4. Summary and Outlook
Three field seasons have provided a unique multi-scale and multi-spectral data set, ranging from sub-meter resolution (aerial imagery) to high-resolution satellite imagery of 10-20 m (ALOS, SPOT) and radar imagery with 3-6 m resolution. In conjunction with the ground survey of ecological, hydrological and energy budget parameters, the data allows for a detailed inter- and intra-annual eco-hydrological classification of the wetland.
The unforeseen cooperation with Gamma Remote Sensing provided the project with another time series of TerraSAR-X data and interferometric analyses. The project is still ongoing.
Further work will include
Back to list of proposals
|© DLR 2004-2016||