Effects of Topographic Corrections on Snow Cover Monitoring In Himalayan Terrain Using MODIS Data

Effects of Topographic Corrections on Snow Cover Monitoring In Himalayan Terrain Using MODIS Data

Manpreet Kaur*, V D Mishra**, J K Sharma*

*Rayat Institute of Engineering & Information Technology, S B S Nagar Punjab 144 533, India.

**Snow and Avalanche Study Establishment, Defence Research and Development Organization, Chandigarh 160036, India.

Abstract

Snow cover is the most important and widely used parameter for meteorological, hydrology and avalanche study. In the present work topographic effects on snow cover assessment was studied using MODIS sensor data from October 2010 to June 2011. NDSI has been used to estimate the snow cover area of topographic corrected and uncorrected satellite images. It has been found that the snow cover area of topographic corrected image is less than the snow cover area of the uncorrected image. The quantitative estimation of snow cover was further analysed for North and South aspects of the NW Himalaya after apply the topographic correction. Comparative analysis between the snow cover area of North and South aspect were evaluated respectively with and without topographic correction. The results show that in North aspect the effect of topography is maximum.

Keywords: MODIS, Topographic correction, Slope matching, NDSI.

1. Introduction

   Snow cover is the largest component of the earths cryosphere and covers up to 33 percent of total land surface. About 98 percent of the total snow cover is located in the Northern Hemisphere [1]. The highly reflective nature of snow combined with its large surface coverage makes snow an important component of Earths radiation balance [2-3]. Study of snow cover plays an important role in hydropower

   generation system, water management and other developmental activities. In addition, varying snow cover area cause formation of avalanches, which in turn strongly affect the safety of human lives and property. In India, Himalaya is well known for the occurrence of snow avalanches, particularly in the North Western parts. Therefore, monitoring of areal extent of snow is an important.

   To obtain snow cover information in highly inaccessible mountainous regions, use of ground based conventional data collection technique is quite difficult, time consuming and area specific only. Satellite remote sensing based techniques are useful tool for real-time, year-round and large spatial coverage for monitoring [4-7]. Different optical sensors of Indian Remote Sensing satellites (IRS) like LISS-III (Linear Imaging Self Scanning), WiFS (Wide Field Sensor) and AWiFS (Advance Wide Field Sensor) are used for the mapping of snow cover area in Himalaya [8-11]. MODIS (Moderate Resolution Imaging Spectroradiometer sensor) is one of the few space borne sensors [7] with an improved ability to map snow cover in good spatial, temporal and radiometric resolution.

   Satellite imagery in mountainous terrain is affected due to sharp variations in the topographic parameters such as altitude, slope and aspect. The topographic variability causes a problem of differential illumination due to steep and varying slopes in rugged Himalayan terrain. Sun-facing illuminated slopes (south aspect) show more than expected spectral radiance or reflectance, whereas the effect is opposite in shaded relief area (north aspect) [12].

   Therefore, topographic corrections are essential for qualitative and quantitative analysis of snow cover applications. A wide range of commonly used topographic correction methods have been reported in the literature [13-21].These are cosine method, c- correction, Minneart correction and two stage normalization. It is reported [22] that slope match is most suitable technique for topographic correction on Himalayan terrain.

   In this present work MODIS sensor data was used to monitor seasonal snow cover. The effect of topographic corrections on quantitative snow cover analysis was investigated using slope matching method.

2. Study Area

   The study area of NW Himalaya considered (Figure1) in the present work is located between 30o1500N to 36o0000N latitude and 73o1500E to 79o4500E longitudes. This area is divided into three distinct zones on the basis of varying climate conditions [23] i.e. Upper, Middle and Lower Himalaya. The Upper Himalaya zone is characterized by comparatively low temperature, light snowfall and severe wind activity. The Middle Himalaya zone is characterized by fairly cold temperature, heavy and dry snow fall and strong wind action. The Lower Himalaya zone is characterized by moderate temperature, heavy snow fall and short winter period. The area is densely forested in between 2400 and 3100m altitude. Beyond 3100m forest is scanty; however, below 2400m area is hebitated.

3. Satellite Data

   Forty five Cloud free satellite images of MODIS sensor from October 2010 to June 2011 are used in the present work to understand the topographic effect on snow cover monitoring. The sensor scans an area 2330km wide swath. Temporal resolution of MODIS sensor is once in a day. The other salient specifications of MODIS sensor are given in Table 1.

4. Methodology

   1. Satellite Data Processing

      The master image of MODIS was generated by geo- referencing it with the help of reference of AWiFS sensor image and 1:50,000 toposheet. ASTER (Advances Spaceborn Thermal Emission and Reflection Radiometer) DEM was used to derive slope, Aspect and Illumination for topographic correction.

      Figure 1 Study area and different zones of North- West Himalaya on MODIS image.

   2. Estimation of Reflectance

      To generate snow cover maps, first scale integer (SI) values are converted into reflectance using equation (1) [1].

      R.cos = reflectance_scales (SI reflectance_offset)

      (1)

      Where, Ri refers to the reflectance of a pixel, indicates to the solar zenith angle in degrees, the values of reflectance_scales and reflectance_offsets can be inferred from meta information of the satellite data [3].

   3. Topographic Correction

      The topographic correction was applied on MODIS images using slope matching technique [22]. Flow chart for topographic correction is shown in Figure 2. The topographically corrected reflectance using slope matching is estimated using equation (2) [22][24].

      Where Rnij is the spectral reflectance of topographically corrected image, Rij is spectral reflectance on the tilted surface, Rmax and Rmin is maximum and minimum spectral reflectance and estimated from topographically uncorrected reflectance image, cosiji is illumination image for

      Table 1 Salient specification of MODIS sensor

      Spectral bands	Spectral wavelength (nm)	Spatial Resolution (m)	Quantization (bit)	Radiance Scale (mw/cm2/sr/m)	Radiance Offset	Solar Exoatmostpheric spectral Irradiance (mw/cm2/sr/m)
      B1	620-670	250	12	0.0026144	0	160.327
      B2	841-876	250	12	0.0009926	0	98.70
      B3	459-479	250	12	0.0027612	0	209.071
      B4	545-565	250	12	0.0021087	0	186.4
      B5	1230-150	250	12	0.0021087	0	47.6
      B6	1628-1652	250	12	0.0002572	0	23.8
      B7	2105-2155	250	12	0.0000787	0	8.7

      pixel illumination on the south aspect, cosi is illumination (IL) image[20]. C is [22] normalization coefficient for different satellite bands and estimated using equation (3);

      Where S is the mean reflectance value on sunny slopes after first stage normalization, N is the mean reflectance value on shady slopes in uncorrected image, N is the mean reflectance value on shady slopes after first stage normalization.

      MODISs Raw Image

      infrared (SWIR) region respectively. Additionally, the reflectance of clouds remains high in SWIR band, thus NDSI allows in discriminating clouds and snow also. NDSI is estimated using equation (4) ;

      (4)

      Where RGreen (MODIS band 4) and RSWIR (MODIS band 6) are the reflectances of the green and shortwave infrared bands respectively. A threshold value for NDSI 0.4 [25] is selected to identify the pixels that are approximately 50% or greater covered by snow. This threshold value merges the water body in snow cover area. Water body can be masked using the reflectance of band NIR>11% and the threshold value of NDSI 0.4. NDSI varies from -1 to +1.

      Geometric Correction

      Projection Transformation

      Estimation of Reflectance

      DEM (Digital Elevation Model

      Slope and Aspect

      Topographic Correction

      4.5 Snow Cover on North and South Aspect

      The snow cover area was retrieved on North and South aspect using topographically uncorrected and corrected reflectance and NDSI image. The North aspect is taken between -45 degree to +45 degree and the south aspect is taken from 135 degree to 225 degree from true North direction. Flow chart for snow cover retrieval on North and South aspect is shown in Figure 3.

      Algorithm for Snow Cover Monitoring

      Figure 2 Flow chart for topographic correction

   4. Algorithm for Snow Area

      Uncorrected

      Algorithm for Snow Cover Monitoring

      S Reflectance Image South and North aspect

      Topographic Corrected reflectance Image

      South and North aspect

      Normalized Difference Snow Index (NDSI) method is used for snow cover mapping using satellite data [3][7][10]. NDSI uses the high reflectance and low reflectance of snow in visible (Green) and shortwave

      Figure 3 Flow chart for snow retrieval on North and South aspect.

5. Result and Discussion

   1. Snow Cover Mapping

      Figure 4 shows the total percentage SCA (snow cover area) estimated for entire North-West Himalayas using NDSI technique. Snow cover started building up in the month of October 2010 and maximum snow cover area occurs in the month of January 2011. It is also observed that April onwards snow cover has started depleted show onset of melt.

      Figure 4 Seasonal snow cover variation in NW Himalaya using MODIS data

   2. Topographic Effects

      1. Solar Zenith, Solar Azimuth, Terrain Slope and Aspect and Illumination

         The solar geometry parameters, viz., solar zenith angle and solar azimuth angle, were computed for each pixel using the equation given in the literature [14][26]. Solar zenith angle for MODIS images from October 2010 to June 2011 varying from 13o to 55o. Solar azimuth angle varies approximately from 121o to 170o. Terrain slope, aspect and illumination (IL) are the three important inputs required in topographic correction models for Himalayan terrain. The terrain slope and aspect image of the study area are shown in Figure 5a and Figure 5b respectively. The altitude in the entire study area varies approximately from 12m to 7000m with a mean 3129m. The slope in the study area varies from 10-80o with mean 15o and aspect ranges from 00-360o with mean 180o.

         Figure 5 (a) Slope of the study area

         Figure 5 (b) Aspect of the study area

         Figure 5c shows the IL image of the study area for topographic correction and computed using the equation given in the literature [12][20]\. The IL vary from -1(low illumination) to +1 (high illumination) which is rescaled to a range of 0255 for the topographic models. The mean IL for MODIS imageries from October 2010 to June 2011 vary from 2100 to 2450.

      2. Estimation of Coefficients for Topographic Models

         The results of coefficient C values were estimated from equation (3) for different date images of MODIS and shown in Table 2. It can be inferred that these coefficients are not same for different date images. These differences in the coefficients are attributed to different Illumination condition, solar zenith and azimuth angle and temporal changes in snow surface physical properties.

         Figure 5(c) Illumination (IL) image

   3. Visual Analysis

      The Visual analysis of topographic corrected image using slope matching method show flat view of the terrain and the three dimensional relief effects is minimized as shown in Figure 6a,b.

   4. Graphically Representation of Topography Effects on SCA

      Figure 4 shows the snow cover area with and without topographic correction, it is observed that the estimated snow cover area is less by using topographic correction than without topographic correction. It is inferred that snow cover area is overestimated without inclusion of topographic corrections.

      (a) (b)

      Figure 6 MODIS images of (a) Uncorrected (b) Topographically corrected

   5. Effect of Topographic Correction on Different Aspects

      After evaluating the topographic effects in NW Himalaya, Aspect wise snow cover monitoring has been carried out to understand the topographic influence on it. This analysis not only provides the aspect-wise snow cover mapping but also provides qualitative and quantitative changes in the snow cover on these two aspects. It can be observed from the Figure 7 that topographic influence on the south aspect is not significant and almost show comparable results of snow cover estimation. Whereas, the result on the north aspect is significant as shown in the Figure 8. Inclusion of topographic models for NDSI using spectral reflectance improve the snow cover analysis on the north aspect.

      Table 2 Values of normalization coefficient for different MODIS passes.

      Dates	B1	B2	B3	B4	B5	B6	B7
      09-Oct-2010	.045	.138	.032	.048	.204	.273	.214
      07-Nov-2010	.021	.100	-.005	.008	.261	.298	.240
      05-Dec-2010	.046	.110	.141	.035	.210	.251	.251
      10-Jan-2011	.226	.286	.189	.212	.399	.437	.365
      08-Mar-2011	-.070	.016	-.080	-.062	.074	.132	.111
      08-Apr-2011	-.212	-.058	-.24	-.213	0	.170	.135
      14-May-2011	-.0295	-.083	-.33	-.306	.173	.490	.480
      15-June-2011	-.107	.02	-.085	-.116	.066	.285	.282

      Figure 7 Snow cover variation in South aspect

      Figure 8 Snow cover variation in North aspect

6. Conclusion

The influence of topography has been studied on the quantitative estimation of seasonal snow cover variation using MODIS data. The analysis is considered for the complete winter period from October to June. Slope matching model for topographic correction is implemented for estimation of NDSI. The snow cover area is overestimated by considering NDSI>=0.4 and reflectance in NIR band

>11% without considering topographic model. The results show no significant change in snow cover area on the south aspect with topographic inclusion. The influence of topography on snow cover on the north aspect is significant. It is concluded that considering topographic corrections improve the results of snow cover analysis. Topographic correction are very useful for further application as sub-pixel snow mapping, climatic modelling, snow melt run-off modelling, change detection analysis, avalanche hazard analysis, etc.

Acknowledgement

The authors would like to thank Director Snow Avalanche Study Establishment (SASE), Department of Defence Research and Development Organization (DRDO) to grant permission to pursue M.Tech thesis in Remote Sensing Research group. Thanks are also due to Vaibhav Sharma, Senior Research Fellow, SASE for dedicated help and scientific discussion.

References

1. NSIDC Snow Cover products 2003 MODIS Snow Cover products Documents Available online at: http://nsidc.org/data/docs/daac/mod10_modis_snow.gd.htm l (accessed November 2011)

2. D. A. Robinson and G. Kukla, Maximum surface albedo of seasonally snow covered lands in the northern Hemisphere, J. Climate and Applied Meteorology, 1985, vol. 24, pp. 402411.

3. H. S. Negi, Snemani and N. K. Thukar, Operational Snow Cover Monitoring in NW Himalaya using Terra and Aqua MODIS Imageries; proc. of Int. workshop on Snow, Ice, Glacier and Avalanches, IIT Bombay, India, 2008, pp. 7-9.

4. M. Konig, J. G. Winther and E. Isaksson, Measuring snow and glacier properties from satellite, Rev. Geophys., 2001, vol. 39, pp. 127.

5. D. K. Hall, J. L. Foster, D. L. Verbyla, A. G. Klrin, and

   C. S. Benson, Assessment of snow-cover mapping accuracy in a variety of vegetation-cover densities in central Alaska, Remote Sens. Environ., 1998, vol. 66, pp. 129-137.

6. D. K. Hall, J. L. Foster, V. V.Salomonson, A. G. Klein and J. Y. L. Chein, Devolpment of a technique to assess snow-cover mapping errors from space, IEEE Transactions on Geoscien. and Remote Sens., 2001,vol. 39(2).

7. D. K. Hall, G. A. Riggs, V. V. Salomonson, N. DiGirolamo and K. J. Bayr, MODIS snow cover products, Remote Sens. of Environ., 2002, vol. 83, pp. 181-194.

8. A. V. Kulkarni, P. Mathur, B. P. Rathore, Suja Alex, Thakur N Manoj, Effect of global warming on snow ablation pattern in the Himalaya, current Science., 2002, vol. 83, pp. 120-123.

9. A.V. Kulkarni and B. P. Rathore, Snow cover monitoring in baspa basin using IRS WiFS data, MAUSAM, 2003, vol. 54, pp. 335-340.

10. A. V. Kulkarni, S. K. Singh, P. Mathur and V. D. Mishra, Algorithum to monitor snow cover using AWIFS

    data of RESOURCESAT-1 for Himalaya region, Int. J. Remote Sens., 2006, vol. 27, pp. 2449-2457.

11. R. P. Gupta, U. K. Haritashya, P. Singh, Mapping dry/wet snow cover in Indian Himalayas using IRS multispectral imagery, Remote Sens. of Environ., 2005, vol. 97, pp. 458-469.

12. D. Riano , E. Chuvieco, J. Salas and I. Aguado, Assessment of different topographic corrections in Landsat-TM data for mapping vegetation types, IEEE Trans. Geosci. Remote Sens., 2003, vol. 41(5), pp. 1056 1061.

13. M. Minneart, The reciprocity principle in lunar photometry, J. Astrophys., 1941, vol. 93, pp. 403410.

14. P. M. Teillet, B. Guindon and D. G. Good enough, On the slope aspect correction of multispectral scanner data, Canada. J. Remote Sens., 1982, vol. 8, pp. 84106.

15. D. K. Hall, A. T. C. Chang and H. Siddalingaiah Reflectances of glaciers as calculated using Landsat-5 Thematic Mapper data, Remote Sens. Environ., 1988, vol. 25, pp. 311321.

16. J. D. Colby Topographic normalization in rugged terrain, Photogramm. Eng. Remote Sens., 1991, vol. 57, pp. 531537.

17. C. Leprieur, J. M. Durand and J. L. Peyron Infuenceof topography on forest reflectance using Landsat Thematic Mapper and digital terrain data, Photogramm.Eng. Remote Sens., 1988, vol. 54, pp. 491 496.

18. J. A. Smith, T. L. Lin and K. J. Ranson The Lambertian Assumption and Landsat Data, Photogramm. Eng. Remote Sens., 1980, vol. 46(9), pp. 11831189.

19. T. Tokola, J. Sarkeala and M. Vander Linden Use oftopographic correction in Landsat TM-based forest inter- pretation in Nepal, Int. J. Remote Sens., 2001, vol. 22, pp. 551563.

20. D. L. Civco, Topographic normalization of Landsat Thematic Mapper digital imagery, Photogramm; Eng. Remote Sens., 1989, vol. 55, pp. 13031309.

21. J. Dozier Snow reflectance from Landsat Thematic Mapper, IEEE Trans. Geosci. Remote Sens., 1984, vol. 22(3), pp. 323-328.

22. V. D. Mishra, J. K. Sharma, K. K. Singh, N. K. Thakur and M. Kumar, Assessment of different topographic corrections in AWiFS satellite imagery of Himalaya terrain, J. Earth Syst.Sci., 2009, vol. 118, pp. 1126.

23. S. S. Sharma and A. Ganju, Complexities of avalanche forecasting in Western Himalayaan overview, Cold Regions Science and Technology, 2000, vol. 31, pp. 95102.

24. J. Nichol, L. K. Hang and W. M. Sing, Empirical correction of low sun angle images in steeply sloping terrain: a slope matching technique, Int. J. Remote Sens., 2006, vol. 27(34), pp. 629635.

25. X. Xiao, Z. Shen, and X. Qin, Assessing the potential of VEGETATION sensor data for mapping snow and ice cover: A normalized difference snow and ice index, Int. J. of Remote Sens., 2001, vol. 22, pp. 2479-2487.

26. F. Kasten, Table of solar altitudes for geographical latitudes, CRREL special Report, 57 U.S Army Corps Engineers, Hanover, New Hampshire, 1962.

.