Download the entire proceedings as an Adobe PDF - Eastern Snow ...
Download the entire proceedings as an Adobe PDF - Eastern Snow ...
Download the entire proceedings as an Adobe PDF - Eastern Snow ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
154<br />
63 rd EASTERN SNOW CONFERENCE<br />
Newark, Delaware USA 2006<br />
still underestimates SWE in densely forested are<strong>as</strong>. Tedesco et al. (2004) developed <strong>an</strong>d tested <strong>an</strong><br />
inversion technique for retrieval of SWE <strong>an</strong>d dry snow depths b<strong>as</strong>ed on artificial neural networks<br />
(ANN) by using 19- <strong>an</strong>d 37-GHz SSM/I me<strong>as</strong>ured brightness temperatures.<br />
Hallikainen et al 2003 combined active (QuikSCAT/SeaWinds) <strong>an</strong>d p<strong>as</strong>sive (SSMI/DMSP) data<br />
for monitoring key snow parameters in Finl<strong>an</strong>d. The results show that combined active <strong>an</strong>d<br />
p<strong>as</strong>sive microwave sensors provide useful diurnal <strong>an</strong>d se<strong>as</strong>onal information. These results are<br />
more accurate th<strong>an</strong> those obtained by only p<strong>as</strong>sive microwave. In <strong>an</strong>o<strong>the</strong>r research Hallikainen<br />
showed that using space borne scatterometer (QuikSCAT onboard SeaWinds) for dry snow<br />
conditions, <strong>the</strong> backscattering coefficient incre<strong>as</strong>es with incre<strong>as</strong>ing SWE. For wet snow condition<br />
backscattering coefficient decre<strong>as</strong>es with incre<strong>as</strong>ing SWE. Ku-b<strong>an</strong>d scatterometer were used<br />
successfully to determine <strong>the</strong> onset <strong>an</strong>d <strong>the</strong> end of snow melt, <strong>an</strong>d to derive time series for <strong>the</strong><br />
fraction of snow-free ground during <strong>the</strong> se<strong>as</strong>onal snow melt period (Hallikainen et al 2004).<br />
Syn<strong>the</strong>tic Aperture Radar (SAR) particularly C-b<strong>an</strong>d SAR h<strong>as</strong> shown <strong>the</strong> potential for<br />
monitoring snow <strong>an</strong>d ice for more th<strong>an</strong> two decades. The high spatial resolution <strong>an</strong>d <strong>the</strong><br />
independence of <strong>the</strong> sensors from sun illumination <strong>an</strong>d cloud cover make SAR <strong>an</strong> ideal tool for<br />
snow studies. Launched in 1995, Radarsat-1 offers spatial resolutions between 10m to 100m <strong>an</strong>d a<br />
swath up to 500km. To estimate SWE using C-b<strong>an</strong>d SAR, Bernier et al. (1998) introduced <strong>an</strong><br />
approach b<strong>as</strong>ed on <strong>the</strong> fact that snow cover characteristics influence <strong>the</strong> underlying soil. The snow<br />
influence on soil temperature affects <strong>the</strong> dielectric properties of <strong>the</strong> soil which h<strong>as</strong> a major role on<br />
<strong>the</strong> backscattered signal. To recover <strong>the</strong> SWE from SAR data <strong>an</strong> algorithm made of two equations<br />
w<strong>as</strong> used. The first equation defines a linear relationship between <strong>the</strong> snow <strong>the</strong>rmal resist<strong>an</strong>ce <strong>an</strong>d<br />
<strong>the</strong> backscattering ratio between a winter image <strong>an</strong>d a reference (snow-free) image in DB. The<br />
snow-free image helps to eliminate <strong>the</strong> radiometric distortion due to topography <strong>as</strong> well <strong>as</strong> to<br />
minimize <strong>the</strong> effect of soil roughness on <strong>the</strong> signal. The second equation is a linear relationship<br />
between <strong>the</strong>rmal resist<strong>an</strong>ce <strong>an</strong>d <strong>the</strong> SWE. To estimate SWE from <strong>the</strong>rmal resist<strong>an</strong>ce <strong>the</strong> me<strong>an</strong><br />
density of <strong>the</strong> snowpack h<strong>as</strong> to be derived. This approach h<strong>as</strong> been applied for cold winter<br />
conditions <strong>an</strong>d dry snow (Bernier et al. 1999). The critical variables influencing <strong>the</strong> algorithm are<br />
variety of l<strong>an</strong>d cover, specifically forest density, <strong>Snow</strong>pack properties (depth>2m), <strong>an</strong>d severe<br />
topography. In a research on p<strong>as</strong>sive <strong>an</strong>d active airborne microwave remote sensing of snow cover<br />
Sokol et al. (2003) showed that SAR sensors are highly sensitive to ch<strong>an</strong>ges in <strong>the</strong> dielectric<br />
const<strong>an</strong>t <strong>an</strong>d have better spatial resolution th<strong>an</strong> <strong>the</strong>ir p<strong>as</strong>sive counterparts. They concluded that<br />
p<strong>as</strong>sive techniques estimate SWE most accurately under dry snow conditions with minimal<br />
stratified snow structures (Sokol et al. 2003).<br />
The focus of this research is estimating <strong>Snow</strong> Water Equivalent (SWE) in Great Lakes area by<br />
using active <strong>an</strong>d p<strong>as</strong>sive microwaves. Different approaches were examined for SWE estimations<br />
by RADARSAT SAR <strong>an</strong>d also QuikSCAT-Ku along <strong>an</strong>d p<strong>as</strong>sive SSM/I.<br />
STUDY AREA<br />
It is also located on <strong>the</strong> tr<strong>an</strong>sitional zone for snow me<strong>an</strong>ing that <strong>the</strong> nor<strong>the</strong>rn part of <strong>the</strong> study<br />
area is covered by snow for <strong>the</strong> whole winter se<strong>as</strong>on however for <strong>the</strong> sou<strong>the</strong>rn part <strong>the</strong>re is a<br />
pattern of snow-fall <strong>an</strong>d snow melt within <strong>the</strong> se<strong>as</strong>on. In addition to snow pattern, <strong>the</strong> l<strong>an</strong>d cover<br />
type varies a wide r<strong>an</strong>ge including, Evergreen Needle leaf forest, Deciduous Broadleaf forest,<br />
cropl<strong>an</strong>d, woodl<strong>an</strong>d <strong>an</strong>d dry l<strong>an</strong>d (Figure 1).