The accurate estimation of snow water equivalent (SWE) in the Canadian sub-arctic is integral to climate variability studies and water availability forecasts for economic considerations (drinking water, hydroelectric power generation). Common passive microwave (PM) snow water equivalent (SWE) algorithms that utilize the differences in brightness temperature (Tb) at 37 GHz – 19 GHz falter in lake-rich tundra environments because of the inclusion of lakes within PM pixels. The overarching goal of this research was to investigate the use of multiple platforms and methodologies to observe and quantify the effects of lake ice and sub-ice water on passive microwave emission for the purpose of improving snow water equivalent (SWE) retrieval algorithms. Using in situ snow and ice measurements as input, the Helsinki University of Technology (HUT) multi-layer snow emission model was modified to include an ice layer below the snow layer. Emission for 6.9, 19, 37 and 89 GHz were simulated at horizontal and vertical polarizations, and were validated by high resolution airborne passive microwave measurements coincident with in situ sampling sites over two lakes near Inuvik, Northwest Territories (NWT). Overall, the general magnitude of brightness temperatures were estimated by the HUT model for 6.9 and 19 GHz H/V, however the variability was not. Simulations produced at 37 GHz exhibited the best agreement relative to observed temperatures. However, emission at 37 GHz does not interact with the radiometrically cold water, indicating that ice properties controlling microwave emission are not fully captured by the HUT model. Alternatively, active microwave synthetic aperture radar (SAR) measurements can be used to identify ice properties that affect passive microwave emission. Dual polarized X-band SAR backscatter was utilized to identify ice types by the segmentation program MAGIC (MAp Guided Ice Classification). Airborne passive microwave transects were grouped by ice type classes and compared to backscatter measurements. In freshwater, where there were few areas of high bubble concentration at the ice/water interface Tbs exhibited positive correlations with cross-polarized backscatter, corresponding to ice types (from low to high emission/backscatter: clear ice, transition zone between clear and grey ice, grey ice and rafted ice). SWE algorithms were applied to emission within each ice type producing negative or near zero values in areas of low 19 GHz Tbs (clear ice, transition zone), but also produced positive values that were closer to the range of in situ measurements in areas of high 19 GHz Tbs (grey and rafted ice). Therefore, cross-polarized X-band SAR measurements can be used as a priori ice type information for spaceborne PM algorithms, providing information on ice types and ice characteristics (floating, frozen to bed), integral to future tundra-specific SWE retrieval algorithms.

Time series of brightness temperature (TB) measurement obtained at various frequencies by the Advanced Microwave Scanning Radiometer–Earth Observing System (AMSR-E) are investigated to determine ice phenology parameters and ice thickness on Great Bear Lake (GBL) and Great Slave Lake (GSL), Northwest Territories, Canada. TB measurements from the 6.9, 10.7, 18.7, 23.8, 36.5, and 89.0 GHz channels (H- and V- polarization) are compared to assess their potential for detecting freeze-onset (FO)/melt-onset (MO), ice-on/ice-off dates, and ice thickness on both lakes. The sensitivity of TB measurements at 6.9, 10.7, and 18.7 GHz to ice thickness is also examined using a previously validated thermodynamic lake ice model and the most recent version of the Helsinki University of Technology (HUT) model, which accounts for the presence of a lake-ice layer under snow. This study shows that 18.7 GHz H-pol is the most suitable AMSR-E channel for detecting ice phenology events, while 18.7 GHz V-pol is preferred for estimating lake ice thickness on the two large northern lakes. These two channels therefore form the basis of new ice cover retrieval algorithms. The algorithms were applied to map monthly ice thickness products and all ice phenology parameters on GBL and GSL over seven ice seasons (2002-2009). Through application of the algorithms much was learned about the spatio-temporal dynamics of ice formation, decay and growth rate/thickness on the two lakes. Key results reveal that: 1) both FO and ice-on dates occur on average 10 days earlier on GBL than on GSL; 2) the freeze-up process or freeze duration (FO to ice-on) takes a comparable amount of time on both lakes (two to three weeks); 3) MO and ice-off dates occur on average one week and approximately four weeks later, respectively, on GBL; 4) the break-up process or melt duration (MO to ice-off) lasts for an equivalent period of time on both lakes (six to eight weeks); 5) ice cover duration is about three to four weeks longer on GBL compared to its more southern counterpart (GSL); and 6) end-of-winter ice thickness (April) on GBL tends to be on average 5-15 cm thicker than on GSL, but with both spatial variations across lakes and differences between years.

The accurate estimation of snow water equivalent (SWE) in the Canadian sub-arctic is integral to climate variability studies and water availability forecasts for economic considerations (drinking water, hydroelectric power generation). Common passive microwave (PM) snow water equivalent (SWE) algorithms that utilize the differences in brightness temperature (Tb) at 37 GHz – 19 GHz falter in lake-rich tundra environments because of the inclusion of lakes within PM pixels. The overarching goal of this research was to investigate the use of multiple platforms and methodologies to observe and quantify the effects of lake ice and sub-ice water on passive microwave emission for the purpose of improving snow water equivalent (SWE) retrieval algorithms. Using in situ snow and ice measurements as input, the Helsinki University of Technology (HUT) multi-layer snow emission model was modified to include an ice layer below the snow layer. Emission for 6.9, 19, 37 and 89 GHz were simulated at horizontal and vertical polarizations, and were validated by high resolution airborne passive microwave measurements coincident with in situ sampling sites over two lakes near Inuvik, Northwest Territories (NWT). Overall, the general magnitude of brightness temperatures were estimated by the HUT model for 6.9 and 19 GHz H/V, however the variability was not. Simulations produced at 37 GHz exhibited the best agreement relative to observed temperatures. However, emission at 37 GHz does not interact with the radiometrically cold water, indicating that ice properties controlling microwave emission are not fully captured by the HUT model. Alternatively, active microwave synthetic aperture radar (SAR) measurements can be used to identify ice properties that affect passive microwave emission. Dual polarized X-band SAR backscatter was utilized to identify ice types by the segmentation program MAGIC (MAp Guided Ice Classification). Airborne passive microwave transects were grouped by ice type classes and compared to backscatter measurements. In freshwater, where there were few areas of high bubble concentration at the ice/water interface Tbs exhibited positive correlations with cross-polarized backscatter, corresponding to ice types (from low to high emission/backscatter: clear ice, transition zone between clear and grey ice, grey ice and rafted ice). SWE algorithms were applied to emission within each ice type producing negative or near zero values in areas of low 19 GHz Tbs (clear ice, transition zone), but also produced positive values that were closer to the range of in situ measurements in areas of high 19 GHz Tbs (grey and rafted ice). Therefore, cross-polarized X-band SAR measurements can be used as a priori ice type information for spaceborne PM algorithms, providing information on ice types and ice characteristics (floating, frozen to bed), integral to future tundra-specific SWE retrieval algorithms.

Time series of brightness temperature (TB) measurement obtained at various frequencies by the Advanced Microwave Scanning Radiometer–Earth Observing System (AMSR-E) are investigated to determine ice phenology parameters and ice thickness on Great Bear Lake (GBL) and Great Slave Lake (GSL), Northwest Territories, Canada. TB measurements from the 6.9, 10.7, 18.7, 23.8, 36.5, and 89.0 GHz channels (H- and V- polarization) are compared to assess their potential for detecting freeze-onset (FO)/melt-onset (MO), ice-on/ice-off dates, and ice thickness on both lakes. The sensitivity of TB measurements at 6.9, 10.7, and 18.7 GHz to ice thickness is also examined using a previously validated thermodynamic lake ice model and the most recent version of the Helsinki University of Technology (HUT) model, which accounts for the presence of a lake-ice layer under snow. This study shows that 18.7 GHz H-pol is the most suitable AMSR-E channel for detecting ice phenology events, while 18.7 GHz V-pol is preferred for estimating lake ice thickness on the two large northern lakes. These two channels therefore form the basis of new ice cover retrieval algorithms. The algorithms were applied to map monthly ice thickness products and all ice phenology parameters on GBL and GSL over seven ice seasons (2002-2009). Through application of the algorithms much was learned about the spatio-temporal dynamics of ice formation, decay and growth rate/thickness on the two lakes. Key results reveal that: 1) both FO and ice-on dates occur on average 10 days earlier on GBL than on GSL; 2) the freeze-up process or freeze duration (FO to ice-on) takes a comparable amount of time on both lakes (two to three weeks); 3) MO and ice-off dates occur on average one week and approximately four weeks later, respectively, on GBL; 4) the break-up process or melt duration (MO to ice-off) lasts for an equivalent period of time on both lakes (six to eight weeks); 5) ice cover duration is about three to four weeks longer on GBL compared to its more southern counterpart (GSL); and 6) end-of-winter ice thickness (April) on GBL tends to be on average 5-15 cm thicker than on GSL, but with both spatial variations across lakes and differences between years.

No description available.

Scene classification is performed on countless remote sensing images in support of operational activities. Automating this process is preferable since manual pixel-level classification is not feasible for large scenes. However, developing such an algorithmic solution is a challenging task due to both scene complexities and sensor limitations. The objective is to develop efficient and accurate unsupervised methods for classification (i.e., assigning each pixel to an appropriate generic class) and for labeling (i.e., properly assigning true labels to each class). Unique from traditional approaches, the proposed bootstrap approach achieves classification and labeling without training data. Here, the full image is partitioned into subimages and the true classes found in each subimage are provided by the user. After these steps, the rest of the process is automatic. Each subimage is individually classified into regions and then using the joint information from all subimages and regions the optimal configuration of labels is found based on an objective function based on a Markov random field (MRF) model. The bootstrap approach has been successfully demonstrated with SAR sea-ice and lake ice images which represent challenging scenes used operationally for ship navigation, climate study, and ice fraction estimation. Accuracy assessment is based on evaluation conducted by third party experts. The bootstrap method is also demonstrated using synthetic and natural images. The impact of this technique is a repeatable and accurate methodology that generates classified maps faster than the standard methodology.

Scene classification is performed on countless remote sensing images in support of operational activities. Automating this process is preferable since manual pixel-level classification is not feasible for large scenes. However, developing such an algorithmic solution is a challenging task due to both scene complexities and sensor limitations. The objective is to develop efficient and accurate unsupervised methods for classification (i.e., assigning each pixel to an appropriate generic class) and for labeling (i.e., properly assigning true labels to each class). Unique from traditional approaches, the proposed bootstrap approach achieves classification and labeling without training data. Here, the full image is partitioned into subimages and the true classes found in each subimage are provided by the user. After these steps, the rest of the process is automatic. Each subimage is individually classified into regions and then using the joint information from all subimages and regions the optimal configuration of labels is found based on an objective function based on a Markov random field (MRF) model. The bootstrap approach has been successfully demonstrated with SAR sea-ice and lake ice images which represent challenging scenes used operationally for ship navigation, climate study, and ice fraction estimation. Accuracy assessment is based on evaluation conducted by third party experts. The bootstrap method is also demonstrated using synthetic and natural images. The impact of this technique is a repeatable and accurate methodology that generates classified maps faster than the standard methodology.