MONITORING GLACIER FLOW VELOCITY BY SAR INTERFEROMETRY AND TEXTURE TRACKING METHOD USING ALOS PALSAR DATA AROUND MT. EVEREST REGION Nikhil R. Poudyal*a, Ryutaro Tateishib, Bambang Setiadic Center for Environmental Remote Sensing, Chiba University 1-33 Yayoi-cho, Inage-ku, Chiba-shi, Japan a <[email protected]>, b <[email protected]>, c <[email protected]> ABSTRACT Glacier flow velocity is one of the important parameter for knowing the glacier dynamics and its recession due to negative mass balance. This research exploits both the SAR (Synthetic Aperture Radar) interferometry and feature tracking techniques in two important glaciers (Khumbu & Kangshung) around Mt. Everest region for estimating glacier flow velocity. The research uses four ALOS PALSAR scene (2007-2009) with two interferometric pair. Shuttle Radar Topography Mission (SRTM -4, Void Filled SRTM Data) elevation data were used for digital elevation information of the area which has a spatial resolution of 90m. Kangshung has good coherence as compared to Khumbu Glacier. The Kangshung Glacier which flows East of China from Everest massif shows velocity up to 8cm/day, where as Khumbu Glacier which flows north to south into Nepal from Everest shows some decorrelation problem and is almost stagnant except some part near the Khumbu falls area. Keywords: SAR Interferometry, Texture Tracking, Glacier, Himalayan, Velocity INTRODUCTION Glacial lake outburst flood (GLOF) is one of the potential hazard that has been alarming due to recent climate changes around the Himalayan Regions of Nepal. The consequence could bring destruction to downhill communities, hydropower stations and other natural settlement of plants and animals. Nepal and surrounding areas in Tibet (China) has experienced more than 30 GLOF since 1964 (WWF Nepal Program March, 2005). Additionally, glaciers are important natural sensors to detect climate change. The GLOF potential threats and the detection of climate change both depends upon on the dynamics of glaciers such as volume, area, morphometry, its morine and valley, and the glacier velocity. Glacier surface velocities are an important part of mass balance modeling of glaciers and therefore constitute an important parameter for monitoring how glaciers respond to the changing climate (Liestol, 1969; Lefauconnier & Hagen 1991). Application of Satellite remote sensing for monitoring glacier velocities is less expensive and has larger coverage than traditional surveying in remote areas. Although Using optical satellite imagery have improved for deriving Himalayan Glacier flow rate but it always has limitation due to its cloud coverage and shadow in deep valleys. Furthermore, the aerial photos requires areas of high contrast in order to work (Lefauconnier et al. 1994; Kääb & Funk 1999). SAR data provides high resolution and precise mapping capabilities which can even detect slight changes or movement on the ground with the support of independency on weather conditions on acquisition period. SAR interferometry (InSAR) has been widely used to document the spatial pattern of velocities typically measured over interval of few days on valley glaciers (Michel & Rignot, 1999). Though this technique has been vastly used around the polar region, only few has been carried out in Himalayan glaciers and reported using either InSAR or offset tracking approaches (Strozzi, 2002). Furthermore the technique of Radar Feature Tracking is also implemented for output comparison. This technique also have been developed considerably in recent years (Lucchitta, 1995; Luckman, 2003), which derives the glacier flow from repeat-pass satellite radar imagery. But the result from Feature Tracking in this research is very preliminary and may not be technically correct. STUDY AREA AND DATA In the current study we put our focus on two glacier near Mt. Everest region, Kangshung Glacier moving east into China and Khumbu Glacier moving from North to South into Nepal as shown in Figure 1. These regions have been showing some downwasting and are potential for forming glacial lake in future. Also we can refer to various previous studies and data (Luckman and others, 2007; D.J. Quincey and others,2009) which can give us the opportunities for our result comparison and also these series of finding can be interpolated for glacier's future health. The Khumbu Glacier is the main tracking route for climbing Mt. Everest and is approximately 17 km long (Flowing North to South, less sensitive in case of InSAR) . This Glacier is noticed to be downwasting and contains many supraglacial lakes on its surface (Nakawo and others, 1999; Wassels and others 2002). Early investigation of ablation in bare ice areas in Khumbu were measured to be 2030 mm d-1 (Nakawo and others, 1999). Similar results can be seen in (Luckman and others, 2007), but this results showed more stagnant parts on the glacier surface. The Kangshung Glacier is oriented from West to East (more sensitive in case of InSAR). It is one of the most remote glacier around the Everest region and has been less explored and studied. This glacier doesn't appear to have developed many supraglacier lakes normally associated with stagnation (Luckman and others, 2007). Four ALOS PALSAR (Advanced Land Observation Satellite Phased Array type L-band Synthetic Aperture Radar) scenes were selected for the research keeping some parameters in mind like least number of repeat pass, perfect baseline and seasons. This criteria left us with very few options for scene selection. The number of scenes used in this research and their details can be seen in Table 1. Additionally, digital elevation models (DEM) with adequate resolution and precision have been available through the NASA's Satellite Radar Topography Mission (SRTM4, Void Filled SRTM Data set). These DEMs after re-sampling were used to model the topographic phase and to isolate the flow related component from the topography. Figure 1: Landsat image to show overview, location and orientation of the study area Table 1: ALOS PALSAR Data used for the study area Pair 1 & 2 used for DInSAR, Pair 2 used for Amplitude tracking Satellite Sensor Process Level Off Nadir Angle Observation Date Pair 1 ALOS PALSAR 1.0 34 2007/12/13 ALOS PALSAR 1.0 34 2008/1/28 Pair 2 ALOS PALSAR 1.0 34 2009/11/2 ALOS PALSAR 1.0 34 2009/12/18 Repeat pass period Perpendicular Baseline (m) 46 Days 280 46 Days 212 Frame Center Number Passes 540 Ascending 540 Ascending 540 Ascending 540 Ascending METHODOLOGY SAR Interferometry: The first methodology applied in this research is SAR Interferometry. In this approach interferograms are generated from two complex SAR images obtained by slightly different orbit configuration with different time interval. This interferogram gives the phase difference between the two acquisition which represent both the surface topography and the coherent displacement along the look vector. The differential use of two interferograms with similar displacement allows the removal of the topographic related phase from the interferogram to derive a displacement map. (Joughin I & other, 1996; Wegmüller U & other, 1998). Two pair of ALOS PALSAR data (Each 46 day interval, See Table 1) were used for the analysis. SAR data were processed using Sarscape tool in ENVI. The raw data (ALOS PALSAR 1.0) were processed first and changed to Single Look Complex (SLC) data products. Then the SLC data were coregistered to prepare for the interferogram process. These interferogram were refined and flattened using the ENVI tool. This process is important for precise transformation of unwrapped phase information into displacement values. It allows for correcting inaccuracies and to calculate the phase offset. These two generated interferogram were suitable for differential analysis since they have different baselines and gave good coherence for most of the glacier zone. Flattened interferogram, intensity image and the coherence image can be seen in Figure 4,5 & 6. The glacier surface displacement in the line of sight (LOS) was calculated after differentiating the two interferogram using dual pair differential Interferometry (DInSAR) tool. This allows to differentiate between topographic phase and the phase actually created by surface displacement. The Interferogram Flattening is done using the reference of SRTM DEM. A set of geometric ground control points with reference to Master image (acquisition on 2007.12.13) were generated to correct the SAR data with respect to the reference DEM. During these processes geo-coded DEM is also generated by converting the unwrapped phase to height by using the phase to height conversion tool in the software (Figure 7). Texture (Amplitude) tracking: Second methodology applied in this research is Amplitude tracking method. This method calculates the shift between master and slave data in pixel units to estimate the displacement in the satellite view direction with help of the amplitude or intensity data from SAR. We use the same Sarscape tool for this purpose. The shift parameter is calculated by means of coregistration process using the coherence first and then using amplitude cross correlation where the coherence value is below the threshold value. But due to lack of proper interval of SAR data the result in this methodology is not satisfactory. On these glaciers with low flow rates, surface velocity is more appropriately measured over longer time frames of between 6 months and to several years (Luckman and others, 2007). But our data has either 46 days repeat pass or more than 2 years. We tried with both image pair and obtained no result from 2 years period (Null values on the glacier surface) whereas some values are obtained in case of 46 days repeat pass which we have presented in Figure 2d & 3b. RESULT AND DISCUSSION The velocity map of both the glaciers and their corresponding mid section displacement and elevation profile are shown in Figure 2a,2b,2c,2d (Khangshung Glacier) and Figure 3a,3b,3c,3d (Khumbu Glacier). These velocity map have been superimposed on intensity image of their corresponding glacier. The Elevation profile is totally obtained from InSAR process and no smoothing process is applied on it. The green line on the displacement graph is the linear model applied to displacement values along the pixels. Khanshung glacier shows a good coherence over the glacier surface and hence clearly visible fringes are obtained (Figure 4). It shows various values of velocity along its surface. On a linear scale it varies from 8 cm/day near accumulation zone to 5 cm/day. But from velocity map (InSAR method) we can observe that the velocity is almost uniform throughout the glacier and the average is calculated to be about 6.5 cm/day. In case of Amplitude tracking method the velocity varies from 0 to 6 cm/day only. The velocity seems to be uniformly distributed along the glaciers except some part where there are higher velocities. These higher velocity can also be equally seen in InSAR method. These zone either falls at confluence zone or near stiff range where there are possibilities of avalanche. By observing its moving trend and almost significant movement till the terminus projected that Khangshung is a healthier glacier. On the other hand Khumbu glacier shows a trend of stagnation before reaching its terminus. Though it has comparatively a high velocity near its accumulation zone but after coming half the way the glacier seems to be stagnant. But we should keep in mind that this glacier being almost perpendicular to the line of sight there could arise some error along the glacier surface. On a linear scale the variation can be observed from 5 to 0 cm/day, but the mid section profile from the InSAR measurement shows the movement is about 7 cm/day at higher elevation and it decreases to 1 cm/day towards the terminus. In case of Amplitude tracking method it shows similar kind of pattern where the velocity ranges from 6 to 0 cm/day. Unlike in case of Khangshung glacier, Khumbu glacier's flow pattern and the velocity are quite comparable in both the methods. These result illustrate both the InSAR and Amplitude Tracking method are useful and can successfully derive the useful velocity information for calculation of the ice mass balance at the glaciers. These two methods has been a complimentary to each other and has been explained and used by (Kaab, 2005; Luckman and others 2007). Amplitude tracking has an advantage in deriving velocity of Himalayan glacier. Any coregistered multi-temporal SAR imagery irrespective of baseline and perfect coherence can be used to compute the glacier velocity with competitive accuracy. In this research due to limited data set we used same time period (46 days repeat pass) for both techniques. But it is recommended to use at least six month or more period difference of satellite acquisition. On the other hand InSAR is considered for giving high level of accuracy if perfect image (good baseline, high coherrence, lower period of repeat pass) is selected for the process. However in the Himalayan region these conditions are difficult to maintain at all times. Finally, these results obtained illustrates that there are much possibilities of applicability of both the technique on the high terrain valley glacier flow rate calculation. ACKNOWLEDGEMENT We heartily acknowledge and thank Prof. Dr. Fumio Yamazaki at Chiba University for providing his lab facility to use the software and process the data. We would like to thank Dr. Nguyen Thanh Hoan (Chiba University) for his valuable suggestions and sharing his enormous experiences with us. We would like to thank Dr. Bayan Alsaaideh (Chiba University) for assisting us in using GIS software. Last but not the least, we thank Mr. Ardi Ansyah (University of Indonesia) for sharing his knowledge of Sarscape tool with us during the research. Khangshung Glacier result demonstration Figure 2a Figure 2b Figure 2c Figure 2d Figure 2a: Velocity map generated by InSAR method (2nd Pair), overlaid on SAR intensity Image Figure 2b: Velocity graph along the mid section pixels of glacier with green line as linear model Figure 2c: InSAR generated elevation model of the glacier Figure 2d: Velocity map generated by Amplitude Tracking method, overlaid on SAR intensity Image Khumbu Glacier result demonstration Figure 3a Figure 3b Figure 3c Figure 3c Figure 3a: Velocity map generated by InSAR method (2nd Pair), overlaid on SAR intensity Image Figure 3b: Velocity map generated by Amplitude Tracking method, overlaid on SAR intensity Image Figure 3c: Velocity graph along the mid section pixels of glacier with green line as linear model Figure 3d: InSAR generated elevation model of the glacier Images from InSAR processing b a Figure 4: Flattened Interferogram of two test site imposed on intensity image with transparency of 40%, Image a: Khanshung Glacier, Image b: Khumbu Glacier Figure 5: Geo-coded SAR intensity (amplitude) image (2007/13/12) Figure 6: Geo-coded Coherence image (obtained from 2nd pair) Figure 7: Geo-coded DEM obtained from DInSAR REFERENCES [1] D.J. Quincey, A. Luckman, D. Benn. Quantification of Everest region glacier velocities between 1992 and 2002, using satellite radar interferometry and feature tracking, Journal of Glaciology, Vol. (55, No. 192), 2009. [2] Joughin I., Winebrenner D., Fahnestock M., Kwok R., and Krabill W., 1996. Measurement of ice-sheet topography using satellite-radar interferometry. Journal of Glaciology. 42(140): 10-22. [3] Kääb, A. (2005). Combination of SRTM3 and repeat ASTER data for deriving alpine glacier flow velocities in the Bhutan Himalaya. Remote Sensing of Environment, 94(4), 463−474. [4] Kääb, A. & Funk, M. 1999. Modelling mass balance using photogrammetric and geophysical data: A pilot study at Griesgletcher, Swiss Alps. Journal of Glaciology (45:151, 575–583). [5] Lefauconnier, B. & Hagen, J.O. 1991. Surging and Calving Glaciers in Eastern Svalbard. Norsk Polarinstitutt Meddelelser (116). Norsk Polarinstitutt, Oslo. [6] Lefauconnier, B., Hagen, J.O. & Rudant, J.P. 1994. Flow speed and calving rate of Kongsbreen glacier, Svalbard, using SPOT images. Polar Research (13:1, 59–65). [7] Liestøl, O. 1969. Glacier surges in West Spitsbergen. Canadian Journal of Earth Science (6, 895 –897). [8] Lucchitta, B. K., Rosanova, C. E., & Mullins, K. F. (1995). Velocities of Pine Island glacier, West Antarctica, from ERS-1 SAR images. Annals of Glaciology, B, (277−283). [9] Luckman, A., Murray, T., Jiskoot, H., Pritchard, H. D., & Strozzi, T. (2003). Automatic ERS SAR feature-tracking measurement of outlet glacier velocities on a regional scale in East Greenland. Annals of Glaciology, (36, 129−134). [10] Luckman, A., D.J. Quincey and S. Bevan. 2007. The potential of satellite radar interferometry and feature tracking for monitoring flow rates of Himalayan glaciers. Remote Sens. Environ., 111(2–3), 172–181. [11] Michel, R., Rignot, E., 1999. Flow of Glaciar Moreno, Argentina, from repeat-pass Shuttle Imaging Radar images: comparison of the phase correlation method with radar interferometry. Journal of Glaciology 45 (149), 93–100. [12] Nakawo, M., Yabuki, H., & Sakai, A. (1999). Characteristics of Khumbu Glacier, Nepal Himalaya: Recent changes in the debris-covered area. Annals of Glaciology, (28, 118−122). [13] Strozzi, T., Luckman, A., Murray, T., Wegmuller, U., Werner, C.L., 2002. Glacier motion estimation using SAR offset-tracking procedures. IEEE Transactions on Geoscience and Remote Sensing (40, 2384–2391). [14] Wessels, R. L., Kargel, J. S., & Kieffer, H. H. (2002). ASTER measurement of supraglacial lakes in the Mount Everest region of the Himalaya. Annals of Glaciology, (34, 399−408).
© Copyright 2024 Paperzz