Recent glacier mass balance and area changes in the Kangri Karpo 1 Mountains derived from multi-sources of DEMs and glacier 2 inventories

Due to the effect of Indian monsoon, the Kangri Karpo Mountains, located in southeast 14 Tibetan Plateau, are the most humid region of Tibetan Plateau, and become one of the most 15 important and concentrated regions with maritime (temperate) glaciers development. Glacier mass 16 loss in Kangri Karpo Mountains is important contributor to global mean sea level rise, and it 17 changes runoff distribution, increases risk of glacial lake outburst floods (GLOFs). Because of 18 their difficult accessibility and high labor costs, the knowledge of glaciological parameters of 19 glaciers in the Kangri Karpo Mountains is still limited. This study presents glacier elevation 20 changes in the Kangri Karpo Mountains, by utilizing geodetic methods based on digital elevation 21 models (DEM) derived from Topographic Maps (1980), the Shuttle Radar Topography Mission 22 (SRTM) DEM (2000), and TerraSAR-X/TanDEM-X (2014). Glacier area and length changes were 23 derived from Topographical Maps and Landsat TM/ETM+/OLI images between 1980 and 2015. 24 Our results show that the Kangri Karpo Mountains contain 1166 glaciers, with an area of 2048.50 25 ± 48.65 km 2 in 2015. Ice cover in the Kangri Karpo Mountains diminished by 679.51 ± 59.49 km 2 26 (24.9% ± 2.2%) or 0.71% ± 0.06% a -1 from 1980 – 2015, however, with nine glaciers in advance 27 from 1980–2015. Glaciers with area of 788.28 km 2 in the region, as derived from DEM 28 differencing, have experienced a mean mass loss of 0.46 ± 0.08 m w.e. a -1 from 1980 – 2014. 29 These glaciers showed slight accelerated shrinkage and significant accelerated mass loss during 3


Introduction
Glaciers in the Tibetan Plateau (TP), as a key components in cryosphere system (Li et al., 2008), are the water resource of many major rivers and lakes (Immerzeel et al., 2010).Glacier mass balance is a valuable indicator to understand the climate variability on the TP (Oerlemans, 1994;Yao et al., 2012a).Under the background of global warming, many mountains glaciers have progressively shrunk in mass and extent in past decades (IPCC, 2013).However, glaciers with positive mass balance have also been reported in recent years, especially in the central Karakoram, eastern Pamir and the western TP (Bao et al., 2015;Gardelle et al., 2012b;Gardelle et al., 2013;Kä ä b et al., 2015;Neckel et al., 2014;Yao et al., 2012a).The relationship between glacier mass balance and climate change, and the knowledge of water source in cryosphere and its disaster risks, are become an advanced research hotspot.
Glaciers in the Kangri Karpo Mountains are known as temperate characteristic under a warming climate and abundant precipitation from Indian Monsoon (Li et al., 1986;Shi and Liu, 2000).Based on digitized glacier inventories from Topographic Maps and remote sensing images, or in situ measurements, glaciers in the Kangri Karpo Mountains have experienced an intense area reduction, continued terminus retreat from 1980-2013, and mass deficit from 2005-2009(Li et al., 2014;;Yang et al., 2010;Yang et al., 2008;Yao et al., 2012a).While some previous studies showed that the phenomenon of glacier advance exists in the Kangri Karpo Mountains.Based on aerial photographs, China -Brazil Earth Resources Satellite (CBERS) image and Landsat Thematic Mapper (TM) image, about 60% of glaciers in the region have been losing their mass and other glaciers have advanced during 1980 to 2001 (Liu et al., 2006).Shi et al. (2006) attributed the complex behavior of glacier dynamics to the increase of precipitation suppressing glacier melting.
Glacier mass balance can be acquired through glaciological, hydrological and geodetic methods (Ye et al., 2015).Due to high altitude and harsh climatic conditions, it is hard to carry out widespread in-situ measurements.Meanwhile, satellite remote sensing is a promising alternative to conduct glacier mass balance through geodetic method, even in remote mountainous terrains and over several glaciers at a time (Paul and Haeberli, 2008).Glaciers in the Kangri Karpo Mountains were almost completely mapped by Topographic Maps that made from aerial photographs in October 1980, and then mapped by X-band SAR Interferometry (InSAR) in February 2000 during the SRTM resulting in a Digital Elevation Model.Glaciers in the region were mapped again by single-pass X-band InSAR from TerraSAR-X on 18 February 2014 and 13 March 2014, and its add-on for digital elevation measurements (TanDEM-X) (Krieger et al., 2007).In this study, the approach of Differential Synthetic Aperture Radar Interferometry (DInSAR) was used to estimate glacier mass balance in the Kangri Karpo Mountains between 1980 and2014.

Study Area
The Kangri Karpo Mountains, located on the eastern end of the Nyainqentanglha Mountains in southeast Tibetan Plateau, extent about 280 km from northwest to southeast.This region is located in the south of Bomi County, and near to Motuo, Zayu and Basu County (Fig. 1).The north of this region, Purlung Zangbo, is a tributary of the Yalung Zangbo River, and the other side is Gongri Gabo River which belongs to west tributary of the Zayu River.Because the east section faces the entrance of the moist southwest monsoon from Indian Ocean, which enters into the plateau at the Grand Bend of the Yarlung Zangbo River, and the terrain forces the air flow to rise, it is the region with the maximum precipitation and highest moisture on the plateau and hence glaciers are well-developed (Shi et al., 2008a).
During winter and spring, the westerly jet in the Northern Hemisphere is blocked by Tibetan Plateau and divided into two branches; southern branch develops into a trough in study area after bypassing the Himalaya Mountains.Moisture in the Bay of Bengal region is attracted by the trough, land on TP and form strong snowfall.In summer, due to the effect of topography flattening, abundant precipitation is transported to study area by the Indian monsoon (Li et al., 1986).Hence, the Kangri Karpo Mountains are the most humid region of Tibetan Plateau, and become one of the most important and concentrated regions with maritime (temperate) glaciers development (Shi and Liu, 2000).It is estimated that the mean summer air temperature at the equilibrium-line altitude (ELA) of glaciers in the region is usually above 1 °C , and annual precipitation is about 2500-3000 mm (Shi et al., 1988).Most glaciers in study area are in the state of pressure melting point, glacier surface ablation is intensive and the velocity of glacier movement is fast (Li et al., 1986).Due to the large ablation and accumulation rates in this maritime region, glacier mass turnover is large in the Kangri Karpo Mountains.
According to the first Chinese Glacier Inventory, the Kangri Karpo Mountains contain 1320 glaciers, with a total area and volume of 2655.2 km 2 and 260.3 km 3 , respectively (Mi et al., 2002).
Yalong Glacier (CGI code: 5O282B37) is the largest one among these glaciers (191.4 km 2 in surface area and 32.5 km in length), while the Ata Glacier (CGI code: 5O291B181), located on the south slope of the Kangri Karpo Mountains, is the glacier with the lowest terminus at 2450 m, 16.7 km in length and 13.75 km 2 in area (Liu et al., 2006).Comparison of photographs taken in different time found that tongue position of Ata Glacier, ice volume and glacial surface conditions have changed greatly in the past decades (Yang et al., 2008).

Topographic Maps
Five topographic maps on 1:100000 scale and fifty topographic maps on 1:50000 scale generated from aerial photos acquired in October 1980 by the Chinese Military Geodetic Service was employed in the present study.Using a seven parameter transformation method, these maps georeferenced into the 1954 Beijing Geodetic Coordinate System (BJ54) with geoid (datum level is Yellow Sea mean sea level at Qingdao Tidal Observatory in 1956) were re-projected into World Geodetic System 1984 (WGS1984)/Earth Gravity Model 1996 (EGM96) (Xu et al., 2013).The contour lines of the maps were manually digitized and then converted into raster DEM (TOPO DEM) with a 30 m grid cell by employing the Thiessen polygon method (Shangguan et al., 2010;Wei et al., 2015b;Zhang et al., 2016a).According to the national photogrametrical standard issued by the Standardization Administration of the People's Republic of China (GB/T12343.1, 2008), the nominal vertical accuracies of these topographic maps were controlled within 3-5 m for the flat (with slopes < 2°) and hilly areas (with slope 2-6°) and controlled within 8-14 m for the mountain (with slope 6-25°) and high mountain areas (with slope >25°).Since the slopes of the most of the glacierized areas in the Kangri Karpo Mountains were gentle (~19°), the vertical accuracy of the TOPO DEM is better than 9 m on glaciers.

Shuttle Radar Topography Mission
The SRTM was conducted to acquire interferometric synthetic aperture radar (InSAR) date simultaneously in the C-band and X-band frequencies from 11 to 22 February 2000 (Farr et al., 2007).The SRTM DEM can be referred to the glacier surface in the last balance year (1999) with slight seasonal variances (Gardelle et al., 2013;Pieczonka et al., 2013;Zwally et al., 2011).The X-band SAR system was operated with a swath width of 45 km leaving large data gaps in the resulting X-band DEM (Rabus et al., 2003).Unfortunately, only 23% of the Kangri Karpo glaciers are covered by the data set.The unfilled finished SRTM C-band DEM, with a swath width of 225 km and 1 arc-second resolution (approximately 30 m) in WGS84/EGM96, is freely available for scientific purposes (http://earthexplorer.usgs.gov/).Hence, the unfilled finished SRTM C-band DEM was employed to study ice surface elevation change.

TerraSAR-X/TanDEM-X
TerraSAR-X was launched in June 2007 by the German Aerospace Center (DLR).TerraSAR-X and its add-on for digital elevation measurements (TanDEM-X) are flying in a close orbit formation acting as a flexible single-pass SAR interferometer (Krieger et al., 2007).
Interferometric data acquisition can be performed in the pursuit monostatic mode, the bistatic mode and the alternating bistatic mode.Current baseline for operational DEM generation is the bistatic mode which minimizes temporal decorrelation and makes efficient use of the transmit power (Krieger et al., 2007).
The experimental Co-registered Single look Slant range Complex (CoSSC) files, acquired in bistatic InSAR stripmap mode on 18 February 2014 and 13 March 2014, were employed in this study (Fig. 2 and Table 1).The CoSSC files have been focused and co-registered at the TanDEM-X Processing and Archiving Facility (PAF).The GAMMA SAR and interferometric processing software was employed for the interferometric processing of the CoSSC files (Werner et al., 2000).

Landsat images
In order to analyze the relationship between glacier mass balance and changes in glacier extent, two Landsat Thematic Mapper (TM) scenes, one Landsat Enhanced Thematic Mapper Plus (ETM+) scene and three Landsat Operational Land Imager (OLI) scenes were employed in this study (Table 1).It is better that Landsat images were acquired in the same year with SRTM and TerraSAR-X/TanDEM-X acquisitions, while because of the low quality of Landsat images in 2000 and 2014, Landsat TM/ETM+ images in 2001 and Landsat OLI images in 2015 with high quality were selected in this study.All Landsat images are available from the United States Geological Survey (USGS), and are orthorectified with the SRTM and ground control points from the Global Land Survey 2005 (GLS2005) dataset.Amongst the Landsat images, the co-registered TerraSAR-X coherence image, the SRTM-X DEM and Topographic Maps, almost no horizontal shift was observed.For Landsat ETM+/OLI images, pan-sharpening employing principal component analysis was performed to enhance the spatial resolution to 15 m.

Glacier Inventory
The outlines of glaciers in October 1980 were delineated manually from Topographic Maps.These maps were geo-referenced and rectified with a kilometer grid, and validated by reference to the original aerial photographs to update the first Chinese Glacier Inventory (Wu et al., 2016b).
Global inventory of glacier outlines can be available through the Randolph Glacier Inventory (Arendt et al., 2015).For the Kangri Karpo Mountains, these glacier outlines are taken from "the Second Chinese Glacier Inventory" (CGI2) that delineated from Landsat TM on 8 September 2005 (Guo et al., 2015).An inventory of high-mountain Asian glaciers named "Glacier Area Mapping for Discharge from the Asian Mountains" was compiled from 356 Landsat ETM+ scenes in 226 path-row sets (Nuimura et al., 2015).The GAMDAM outlines are nearly all from 1999-2003 and thus conform with the recommendation for the compilation of glacier inventory to select imagery as close to 2000 as possible (Arendt et al., 2015;Paul et al., 2009).Hence, Landsat TM/ETM+ scenes were used to validate and update the CGI2 and GAMDAM glacier inventory, and generated the 2000 inventory of the detailed study area.
A semi-automated approach using the TM3/TM5 band ratio was applied to delineate glacier outlines in 2015 using Landsat OLI images (Bolch et al., 2010b;Paul et al., 2009;Racoviteanu et al., 2009).To ensure that ice patches were larger than 0.01 km 2 , a 3 by 3 median filter was applied to eliminate isolated pixel (Bolch et al., 2010b;Wu et al., 2016b).Then, the derived glacier polygons are checked manually against images from adjacent years with less or no snow and cloud-free, to discriminate proglacial lakes, seasonal snow, boulders on the glacier and debris-covered ice (Fig. 3).The final contiguous ice coverage was divided into individual glacier polygons using topographical ridgelines (TRLs), which were automatically generated based on the SRTM-C DEM (Guo et al., 2011).
The best way to assess the accuracy of glacier outlines is to compare extracted results with independently digitized glacier outlines from high resolution aerial imagery at random locations (Bolch et al., 2010a;Paul et al., 2003).Previous studies indicated that the average offsets between glacier outlines derived from Topographic Maps and Corona-image outlines was ±6.8 m (Wu et al., 2016b), and the average offsets between Landsat-image outlines and real-time kinematic differential GPS (RTK-DGPS) measurements, Google Earth TM images with a spatial resolution better than 1 m, were ±10 m and ±30 m for the delineation of clean and debris-covered ice, respectively (Guo et al., 2015).On the basis of these average offsets, mean relative errors of ±1.3%, ±2.0% and ±2.4% were calculated for glacier area in 1980, 2000 and 2015, respectively.

Glacier Length
Glacier length is a key parameter in glacier inventory and its vector representation (glacier centerlines) is important to model future glacier evolution and calculate glacier ice volume (Le Bris and Paul, 2013).Some researchers defined it as the central flowline from the highest glacier elevation to the terminus, whereas others regard the length of the longest flowline as glacier length (Kienholz et al., 2014;Leclercq et al., 2014).The length change can be calculated by intersecting the central flowline with the respective glacier outline (Paul and Svoboda, 2010), or calculated as the average length from the intersection of the glacier outlines with the stripes which drawn parallel to the main flow direction of the glaciers (Koblet et al., 2010).
In this study, a new strategy based on a glacier axis concept from glacier morphology perspective was applied that requires only glacier outlines and a DEM as input (Yao et al., 2015).
As the base for the glacier axis concept we assume that the main direction of any given glaciers can be defined as a curved line from its highest to its lowest elevation.At first, the outline of given glacier is divided into two curved lines by its highest and its lowest point.Using the two curved lines, the polygon of given glacier is divided into two regions by Euclidean distance.Glacier axis is the common boundary of the two regions, and it can be defined as the glacier centerline.An error estimation of the resulting glacier centerlines was performed that compared the semi-automatically generated results to high resolution aerial imagery at the terminus of glacier centerlines.A Corona image with a resolution of 4 m and Google Earth TM images with a resolution better than 1 m were used to evaluate the accuracy of glacier centerlines.Hence, the uncertainties of glacier centerlines from Topographic Maps and Landsat images are no more than 6 m and 7.5 m, respectively.
The interferometric phase of the single-pass TerraSAR-X/TanDEM-X interferogram could be described by where ∆ ∅TSX TDX ⁄ is the difference of phases of phases ∅TSX and ∅TDX simultaneously acquired by TerraSAR-X and TanDEM-X.∆ ∅orbit is the phase difference induced by the different acquisition geometry of the SAR sensors, and ∆ ∅topo is the phase difference induced by topography.∆ ∅atm and ∆ ∅scat are the phase differences induced by atmospheric conditions and different scattering on the ground.As the data of TerraSAR-X/TanDEM-X were acquired simultaneously, the same atmospheric conditions and scattering are assumed for both SAR antennas, which set ∆ ∅atm and ∆ ∅scat in Eq. ( 1) to zero.∆ ∅orbit could be removed from the interferogram by subtracting a simulated flat-earth phase trend (Rosen et al., 2000).
The DInSAR approach can be described by where ∆ ∅SRTM−C is the interferometric phase of the February 2000 SRTM-C acquisition.Due to the unavailable of the raw interferometric data of the SRTM-C acquisition, ∆ ∅SRTM−C was simulated from SRTM-C DEM data using the satellite geometry and baseline model of the TerraSAR-X/TanDEM-X pass.Therefore the differential phase ∆ ∅diff is solely based on changes in ∆ ∅topo between data acquisitions (Neckel et al., 2013b).
In order to improve the phase-unwrapping procedure and minimize errors, the unfilled finished SRTM C-band DEM were employed in this study.Before generating the differential interferogram, precise horizontal offset registration and fitting between the SRTM C-band DEM and the TerraSAR-X/TanDEM-X acquisitions is necessary.Based on the relation between the map coordinates of the SRTM C-band DEM segment covering the TerraSAR-X/TanDEM-X master file, and the SAR geometry of the respective master file, an initial lookup table was calculated.
While the areas of radar shadows and layover in the TerraSAR-X/TanDEM-X interferogram would induce gaps in the lookup table, a method of linear interpolation between the gap edges in each line of the lookup table was used to fill these gaps.The offsets between the master scene and the simulated intensity of the SRTM C-band DEM, were calculated using cross correlation optimization of the simulated SAR images employing GAMMA's offset_pwrm module.The horizontal registration and geocoding lookup table were refined by these offsets.The SRTM C-band DEM was translated from geographic coordinates into SAR coordinates via the refined geocoding lookup table, and conversely, the final difference map was translated from SAR coordinates into geographic coordinates.Then a differential interferogram was generated by the TerraSAR-X/TanDEM-X interferogram and the simulated phase of the co-registered SRTM C-band DEM.An adaptive filtering approach was used to filter the differential interferogram (Goldstein and Werner, 1998), and then GAMMA's minimum cost flow (MCF) algorithm was employed to unwrap the flattened differential interferogram.According to the computed phase-to-height sensitivity and select ground control points (GCPs) from the respective off-glacier pixel locations in the SRTM C-band DEM, the unwrapped differential phase was converted to absolute differential heights.While, a residual not covered by the baseline refinement would be existed, and can be regarded as a linear trend that estimated by a two dimensional first order polynomial fit in off-glacier regions.The linear trend and a constant vertical offset were removed from the maps of absolute differential heights.Finally, the resulting data sets were translated to a metric cartographic coordinate system with 30 m × 30 m pixel spacing (Neckel et al., 2013a).
The same method of DInSAR was employed to acquire the glacier elevation change from 1980 to 2014 with the data sets of TOPO DEM and TerraSAR-X/TanDEM-X acquisitions.
For the glacier elevation change from 1980 to 1999, a difference map was constructed by common DEM differencing with TOPO DEM and SRTM C-band DEM (Bolch et al., 2011;Nuth and Kä ä b, 2011;Pieczonka et al., 2013;Wei et al., 2015a).Relative horizontal and vertical distortions between the two data sets, can be corrected by statistical approaches that based on the relationship between elevation difference and slope, aspect (Nuth and Kä ä b, 2011).Elevation differences in off-glacier regions were used to analyze the consistency of TOPO DEM and SRTM C-band DEM (Fig. 4).Outliers of elevation differences, usually around data gaps and near DEM edges, were omitted with values exceeding ±100 m (Berthier et al., 2010;Bolch et al., 2011).The vertical biases and horizontal displacements could be adjusted simultaneously using the substantial cosinusoidal relationship between standardized vertical bias and topographical parameters (slope and aspect).And the biases that caused by different spatial resolutions between DEMs, could be adjusted by the relationship between elevation differences and maximum curvatures (Gardelle et al., 2012a;Nuth and Kä ä b, 2011).
The penetration depth of SRTM C-band radar beam into snow and ice need to be considered for elevation changes of glacier surfaces (Berthier et al., 2006;Gardelle et al., 2012a;Pieczonka et al., 2013).Penetration depth can range from 0 to 10 m depending on a variety of parameters such as snow temperature, density and water content (Berthier et al., 2006;Dall et al., 2001).As a first approximation, the penetration depth of SRTM X-band radar beam is much smaller than C-band, the elevation difference between these two data sets can be considered as the penetration of SRTM C-band radar beam into snow and ice (Gardelle et al., 2012a).The elevation differences between

Mass balance and error estimation
In order to convert the derived surface elevation changes into the mass balance of glaciers, a density of ice/firn/snow should be considered.A value of 900 kg m -3 was applied to assess the mass changes in water equivalent (w.e.) from elevation differences, and then adding an ice density uncertainty of 17 kg m -3 (Gardner et al., 2013;Neckel et al., 2013b) .
For the accuracy assessment of TOPO DEM and SRTM C-band DEM, the elevation product of Geoscience Laser Altimeter System (GLAS) carried on-board the Ice Cloud and Elevation Satellite (ICESat) was utilized in this study (Neckel et al., 2013a).All available GLAS elevation data for the study area were obtained from the National Snow and Ice Data Center (NSIDC) (release 634; product GLA 14).Because of the effect of clouds during the time of data acquisition, some GLAS elevation data cannot represent the true altitude of ground surface.Outliers of elevation difference between GLA 14 and multi-source of DEMs in off-glacier regions were removed from the analysis with values exceeding ±100 m.Compared to the GLAS elevation data, a mean and standard deviation of 2.74 ± 1.73 m and 2.65 ± 1.48 m for TOPO DEM and SRTM C-band DEM, respectively.Due to the GCPs that convert the unwrapped TerraSAR-X/TanDEM-X interferogram into absolute heights was selected from the respective off-glacier pixel locations in the SRTM C-band DEM, vertical bias of TerraSAR-X/TanDEM-X DEM and GLA 14 is similar with the bias of SRTM C-band DEM and GLA 14.
For an error estimate of the derived surface elevation changes, the residual elevation differences were estimated in off-glacier regions assuming that these areas did not change in height between 1980 and 2014 and that elevations should be equal in TOPO DEM, SRTM C-band DEM and TerraSAR-X/TanDEM-X DEM.The mean elevation difference (MED) over off-glacier regions between the final difference maps was in the range of -1.42 to 0.75 m (Table 2).Because averaging over larger regions reduces the errors, the standard deviation (STDV) over off-glacier regions would probably overestimate the uncertainty for larger samples.Thus, the uncertainty can be estimated by the standard error of the mean (SE): where N is the number of the included pixels.To avoid the effect of autocorrelation, de-correlation length of 600 m and 200 m was employed for difference maps that derived by common DEM differencing and DInSAR (Bolch et al., 2011;Neckel et al., 2013a).Then, the overall errors of the derived surface elevation changes can be estimated using SE and MED over off-glacier regions: 4) Finally, the root of sum of squares of the estimated errors of glacier area and surface elevation changes, and ice density uncertainty of 17 kg m -3 , were used to estimate the overall errors of mass balance (Neckel et al., 2013a).

Area change
According to the 2015 inventory, the Kangri Karpo Mountains contain 1166 glaciers, with an area of 2048.50 ± 48.65 km 2 , and the mean glacier size is 1.76 ± 0.04 km 2 (Fig. 5).The highest number of glaciers can be found in the size class 0.1-0.5 km 2 , whereas glaciers between 1-5 km 2 cover the largest area (Fig. 5A).Only two glaciers area larger than 50 km 2 , the largest glacier is Yalong Glacier and another is Xirinongpu Glacier, with glacier area of 173.00 ± 0.67 km 2 and 90.28 ± 0.23 km 2 , respectively.Glaciers area in Kangri Karpo Mountains present a normal hypsometry, about 76.9% of glacier area lies in the 4500-5500 m elevation range.Azha Glacier is the glacier with the lowest glacier tongue position, and the elevation of glacier tongue at 2551 m (Fig. 5B).
Median elevation of the glaciers in Kangri Karpo Mountains is situated at around 4852 m, 5215 m for glaciers on the north slope and 4639 m on the south.This is consistent with the tendency of equilibrium line altitude in southeast Tibetan Plateau (Su et al., 2014).The mean glacier surface slope in the Kangri Karpo Mountains is 24.1º , with most in the 12-32º range that accounts for 80.5% of the glaciers and 85.4% of their area.Most glaciers have a SE, S or SW aspects, account for 59.2% and 80.9% of glacier number and area, respectively.
Comparing the total area of all glaciers in 1980 with that in 2015, ice cover in the Kangri Karpo Mountains diminished by 679.51 ± 59.49 km 2 (24.9% ± 2.2%) or 0.71% ± 0.06% a -1 .Small glaciers shrunk in area in larger percentage (Fig. 5C).Meanwhile, absolute area loss was higher for larger glaciers.Analysis of the glacier hypsography indicated that ice coverage disappeared completely below 2500 m, the highest absolute area loss occurred in the 4500-4700 m a.s.l.altitude range, and ice coverage remains almost unchanged above 5800 m.The average minimum elevation of the glaciers increased by 106 m, while median elevation rose about 56 m from 4796 to 4852 m.

Length change
Comparing the termini of all glaciers, only nine glaciers advanced while others retreated in the Kangri Karpo Mountains from 1980 -2015.The nine glaciers experienced a mean advance of 14.8 m a -1 , and the length of centerlines increased between 103 m and 1547 m.The terminus elevation decrease of these advanced glaciers averaged at 191 m, varying between 34 m (the smallest lowering from 4796 m to 4762 m a.s.l.altitude) and 412 m (the largest one from 4362 m to 3949 m a.s.l.altitude) (Table 5).Based on different glacier size, slope and aspect, 86 glaciers were selected from all retreated glaciers to analyze the changes of length.These selected glaciers experienced a mean recession of 759 m (21.7 m a -1 ) with smallest of 6 m and largest of 3956 m.
Similar to area change of the glaciers in the case study region, these glaciers have shown accelerated retreat during the two periods from 1980 -2000 and from 2000 -2015 as measured in glacier length (Table 6).Glaciers experienced a mean length reduction of 21.0 m a -1 (varying from 2.5 m a -1 to 104.2 m a -1 ) in the period of 1980 -2000 and 22.6 m a -1 from 2000 -2015 (varied between 1.3 m a -1 and 144.8 m a -1 ).The recession of Yalong Glacier changed from 78.0 m a -1 in 1980 -2000 to 13.6 m a -1 in 2000 -2014, while the average rates of retreat increased significantly for Azha Glacier in the two sub-periods of 1980 -2000 and 2000 -2015, with a mean recession of 11.3 m a -1 and 144.8 m a -1 , respectively.

Mass balance
The average elevation change of the entire glacier surfaces in the case study area of Kangri Karpo Prominent thickening (elevation increase) was found on the termini of two glaciers on the southern slope of the Kangri Karpo Mountains (Fig. 6C, WGI: 5O291B0113 and 5O291B0117).
There have 3.79 km 2 and 3.70 km 2 debris-covered areas on the two glaciers, account for 20.6% and 31.4% of individual glacier areas, respectively.Meanwhile, the lengths of debris-covered regions account for 69.4% and 63.3% of individual glacier lengths.Probably the effect of debris cover, the glacier termini of 5O291B0113 and 5O291B0117 remain stable between October 1980 and October 2015.

Uncertainty
The uncertainty of glacier outlines was caused by positional and processing errors associated with glacier delineation (Bolch et al., 2010a;Racoviteanu et al., 2009).Seasonal snow, cloud and debris cover complicated glacier mapping precisely (Paul et al., 2013).The accuracy of glacier outlines in this study was estimated by compare extracted results with independently digitized glacier outlines from high resolution aerial imagery at random locations.An uncertainty model was employed to assess accuracy estimated in this study (Pfeffer et al., 2014).The delineation uncertainty of glaciers in the Kangri Karpo Mountains in 2015 was about 24.33 km 2 using the uncertainty model, that is smaller than the uncertainty of 48.65 km 2 in this study.Main reason for this discrepancy is probably that the delineation uncertainty of glaciers have been overestimated in this study, especially in the area of debris-covered ice and exposed bedrock that surrounded by ice cover.
For the uncertainty of mass balance, the penetration depth of SRTM C-band radar beam into snow and ice was critical issue when SRTM DEM was employed for geodetic mass balance calculations.The penetration depth of SRTM C-band radar beam can be estimated by comparing the SRTM C-band with the SRTM X-band DEM (Gardelle et al., 2012a;Kä ä b et al., 2012).
Previous studies indicated that the penetration depth decreases as temperature and water content of surface snow cover rise (Surdyk, 2002), and penetration depths at 10 GHz from 2.1 m to 4.7 m were measured in Antarctica (Davis and Poznyak, 1993).Glaciers in eastern Nyainqentanglha Mountains are predominantly monsoonal influenced and have more snow moisture and higher temperatures than the Antarctic ice sheet (Shi and Liu, 2000).Hence, the penetration correction is suitable under the assumption that the influence of slight penetration of the X-band is negligible.
The mean SRTM C-band penetration was 1.24 m in Kangri Karpo Mountains, led to mass changes on average of +0.06 m w.e. a -1 and -0.08 m w.e. a -1 for the periods of 1980 -2000 and 2000 -2014.Another issue is the lack of information in several regions due to data voids in accumulation area.Different suitable assumptions or elevation changes in the accumulation regions were used to fill the data voids and to assess the impact on mass balance (Pieczonka et al., 2013;Shangguan et al., 2015).In this study, the information of elevation change exists in all altitudinal zones from 2400 m to 6600 m a.s.l., and the area of data voids was too small (0.7% above 6000 m a.s.l. in area) to affect the mass balance significantly.Hence, data voids can be neglected when mass balance was estimated by all glaciers area in the detailed study area, average surface elevation change and ice density.

Glacier change of area and length
This study found ice cover in the Kangri Karpo Mountains diminished between 1980 and 2014 of about 0.71% ± 0.06% a -1 .In the period of 1980 -2000, glacier area decreased by 0.41% ± 0.06% a -1 in the case study area, and then this value increased to 0.52% ± 0.10% a -1 after 2000.
Our result is in agreement with previous study, that found shrinkage of 0.57% a -1 in the southeastern Tibetan Plateau from 1980-2001(Yao et al., 2012a).The shrinkage rate of previous study is larger than our result slightly, main reason for that is probably the difference of glacier size.The mean glacier sizes are about 2.07 km 2 and 6.54 km 2 in the Kangri Karpo Mountains and in the detailed study area in 1980, and greater relative loss for smaller glaciers was found in this study and previous studies (Wei et al., 2014;Wu et al., 2016b).
The location of glacier terminal is often measured by remote sensing and investigation in the field.Due to the differences in the periods studied and spatial scales, the length changes of glacier centerlines in this study are slower than previous studies, except for Azha Glacier (Liu et al., 2006;Yang et al., 2010;Yao et al., 2012a).The investigation of terminus variation on Parlung No. 10 Glacier was surveyed from 2006 -2008, the period of survey is too short to represent length change of glacier centerlines over long periods of time (Yang et al., 2010) For advanced glaciers, the mean glacier size is about 0.51 km 2 and mean glacier surface slope is about 27.9 º , most glaciers have an S or SW aspect, and mean accumulation area ratio (AAR) is 51.Previous studies also found advanced glaciers in Kangri Karpo Mountains (Liu et al., 2006;Shi et al., 2006).Compared with the CGI2 and GAMDAM glacier inventory, the location of most glacier termini in 2000 are very close to that in 2014, indicated that glacier advanced mainly occurs before 2000.Due to the special geographic location and climate feature, the qualities of Landsat MSS/TM images are too low to identify glacier termini.Fortunately, two Landsat TM scenes (LT51340401994189BKT00 and LT51340401988301BJC00) with high quality can be employed in this study.Compared the glacier termini that acquired from Landsat scenes, such as Glacier 5O282B0111 (Fig. 3B), glacier advanced mainly occurs before 1988, and glacier retreated continuously after that (Fig. 7).Main reason for this phenomenon is probably that the increase of high precipitation (Shi et al., 2006).Annual precipitation dataset from 1980 to 2012, collected from the three nearest meteorological stations (Bomi, Zuogong and Zayu), indicated that maximum precipitation (1153 mm in 1988) is 1.6 times the minimum precipitation (714 mm in 1981) at Bomi (29°52′N, 95°46′E, 2736 m a.s.l.), maximum precipitation (683 mm in 1987) is 2.3 times the minimum precipitation (302 mm in 1983) at Zuogong (29°40′N, 97°50′E, 3780 m a.s.l.), and maximum precipitation (1091 mm in 1988) is 1.7 times the minimum precipitation (624 mm in 1982) at Zayu (28°39′N, 97°28′E, 2423 m a.s.l.).Supposing that the precipitation fluctuation in high elevation glacier area had been consistent with that at the three nearest meteorological stations, the change of precipitation or glacier accumulation certainly have significant influence on terminus fluctuation of glaciers.Due to the complicated terrains, the accumulation of glaciers varies greatly, and the response of glacier movement is not quite the same, individual glaciers advanced during 1980 -1988.

Glacier thinning and mass balance
A comparison of glacier thickness changes show that significant differences are found over eastern Nyainqentanglha Mountains.Using SRTM DEM and SPOT5 DEM (24 November 2011), glaciers experienced a mean thinning of 0.39 ± 0.16 m a -1 in eastern Nyainqentanglha Mountains (named Hengduan Shan) (Gardelle et al., 2013) (2015).Main reason for this discrepancy is the different estimation of SRTM C-band penetration.
An average SRTM C-band penetration of 1.24 m was used for Kangri Karpo Mountains that estimated from the difference of SRTM C-band and X-band DEMs (Gardelle et al., 2012a).While an average penetration of 8-10 m (7-9 m if based on the winter trends that might alternatively be assumed to reflect February conditions) was employed for eastern Nyainqentanglha Mountains (Kä ä b et al., 2015).Previous studies indicated that the penetration depth varies with temperature and water content (Surdyk, 2002), and penetration depths of SRTM C-band from 1.4 m to 3.4 m were estimated over the Pamir-Karakoram-Himalaya (Gardelle et al., 2013;Kä ä b et al., 2012).
The characteristics of glaciers in eastern Nyainqentanglha Mountains are similar with that in eastern Himalaya (Shi et al., 2008b).Therefore, the penetration in this study is more suitable under the assumption that the penetrations in eastern Nyainqentanglha Mountains and eastern Himalaya are similar.(Yang et al., 2008;Yao et al., 2012a).Based on SRTM DEM and TerraSAR-X/TanDEM-X acquisitions (18 February 2014), the two glaciers experienced large surface lowering from 2000 to 2014, with mean mass deficit of 0.65 ± 0.22 m w.e. a -1 and 0.67 ± 0.22 m w.e. a -1 .The comparison between filed measurements of mass balance and the result of this study indicated that there was high consistency of glacier mass deficit about Parlung No. 4 Glacier and Parlung No. 10 Glacier.
Interestingly is the bigger thinning on the debris-covered region of -0.99 ± 0.09 m a -1 on average than clean-ice region of -0.89 ± 0.08 m w.e. a -1 in the 2800 -5300 m a.s.l.altitude range from 1980 -2014 (Fig. 8).2800 m a.s.l. was the lowest altitude of clean-ice region and 5300 m a.s.l. was the highest altitude of debris-covered region.Due to complex surface conditions, such as supraglacial lakes, ice cliffs and heterogeneity of debris cover, the mass loss patterns on the debris-covered tongue are complicated (Pellicciotti et al., 2015).It is generally believed that ice ablation rate is highly reduced with the thick debris-cover due to the insulation effect of debris.
However, previous studies found that glacier ablation on debris-covered regions were greater than on the exposed ice regions, when debris thickness is less than critical thickness (Nakawo and Young, 1981;Pu et al., 2003;Ye et al., 2015;Zhang et al., 2011;Zhang et al., 2016a).The location of debris-covered regions, in the lower altitudes with higher temperature, and the development of supraglacial lakes and ice cliffs are probably the reasons for the larger mass loss on the debris-covered regions (Benn et al., 2012;Sakai and Fujita, 2010).
Overall, negative elevation changes were found in glacier tongue regions except for two glaciers in the southern slope of the Kangri Karpo Mountains (Fig. 6C).Compared the average elevation changes of the two glaciers tongue surface during 1980-2000 and 2000-2014, positive elevation changes were found between October 1980 and February 2000, and negative elevation changes were found after 2000.The two glaciers showed positive elevation changes at the termini, while elevation changes are unknown in accumulation area of the two glaciers due to data voids.This phenomenon could be interpreted as glacier surging (Cuffey and Paterson, 2010), or result from the increase of high precipitation (Shi et al., 2006).

Climatic considerations
The climate in Kangri Karpo Mountains is characterized be the westerly in winter and the Indian monsoon in summer (Li et al., 1986).While the effects of westerly are weak in the study area due to the block of Tibetan Plateau.Hence, the accumulation on glaciers in Kangri Karpo Mountains is supplied by summer monsoon precipitation (Bolch et al., 2010a;Yao et al., 2012a).
Previous studies indicated that the Tibetan Plateau has experienced an overall surface air warming since the mid-1950s (Duan et al., 2015;Li et al., 2010;Liu et al., 2008;Liu et al., 2009;Qin et al., 2009;Yang et al., 2014;Yao et al., 2012a;You et al., 2010).Different trend of average annual temperature results from different data in the southeastern TP.Based on temperature data from meteorological stations, the southeastern TP present the lowest warming rate (Duan et al., 2015), while the most warming rate was found from the MODIS land surface temperature (MODIS LST) in the southeastern TP (Yang et al., 2014), even decreasing trend of average annual temperature was found from the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) Reanalysis data (You et al., 2010).The changes in air temperature were accompanied by the changes in precipitation due to variations in monsoonal activity.Based on the Global Precipitation Climatology Project (GPCP) data, precipitation decreased in the southeastern TP from 1979 to 2010 (Yao et al., 2012a).While positive trend of annual precipitation was found from Chinese meteorological stations data in the southeastern TP, precipitation amount increased and the frequency of severely dry events decreased significantly (Li et al., 2010).Due to the ambiguous of climate change, the presented glacier changes and mass balance cannot be explained by the above summarized climate variations directly.
For the rainfall-variation law, a slightly increasing trend was present in the detailed study SRTM C-band and X-band indicate that the penetration depth of the C-band averaged 1.24 m in the study area.The mean value of the SRTM C-band penetration over glaciers in Kangri Karpo Mountains is in agreement withGardelle et al. (2013), who found penetration of 1.7 m in eastern Nyainqentanglha Mountains (named Hengduan Shan inGardelle et al. (2013)).
Mountains was -17.46 ± 0.54 m from 1980 -2014.The glaciers, with an area of 788.28 km 2 , have experienced a mean thinning of 0.51 ± 0.09 m a -1 , or mean mass loss of 0.46 ± 0.08 m w.e. a -1 , equivalent to an overall mass change of -13.76 ± 0.43 Gt for the case study area during 1980 -2014.Thinning of these glaciers has speeded up during the periods of1980 -2000  and 2000  -2014 .From 1980  -2000, glaciers have thinned by 5.30 ± 0.77 m on average, and experienced a mass loss of 0.24 ± 0.16 m w.e. a -1 .Then a surface lowering of 11.04 ± 0.43 m was found from 2000 -2014, with a mass loss of 0.71 ± 0.10 m w.e. a -1 (Fig.6 and Table 7).Heterogeneous glacier mass balance was presented in the detailed study area of Kangri Karpo Mountains during 1980 -2014.Glaciers, with an area of 471.05 ± 3.03 km 2 in the drainage basin of 5O282B, experienced greater mass deficit of 0.51 ± 0.22 m w.e. a -1 from 1980 -2014, and mean mass deficit of 0.30 ± 0.14 m w.e. a -1 and 0.76 ± 0.22 m w.e. a -1 during the periods of 1980 -2000 and 2000 -2014, respectively.Mean mass deficit of 0.39 ± 0.11 m w.e. a -1 in the drainage basin of 5O291B was smaller than those in 5O282B drainage basin during 1980 -2014.Glaciers with an area of 317.22 ± 4.27 km 2 in 5O291B drainage basin experienced accelerating mass deficit during the periods of 1980 -2000 and 2000 -2014, with mean mass deficit of 0.13 ± 0.16 m w.e. a -1 and 0.63 ± 0.04 m w.e. a -1 .
. Compared the length change of Azha Glacier in different periods, -56.1 m a -1 from 1973 -2005 (Yao et al., 2012a), -65 m a -1 from 1980 -2006 (Yang et al., 2010) and -70 m a -1 from 1980 -2015 (this study), we can found that Azha Glacier have experienced greater retreat after 2000s than that before 2000s.The length change of Yalong Glacier from 1980 -2000 in this study is similar with Liu et al. (2006), who found retreat of 73 m a -1 of Yalong Glacier between 1980 and 2001.And then the average rate of retreat decreased significantly for Yalong Glacier after 2000.
. Based on ICESat and SRTM, Kä ä b et al. (2015), Neckel et al. (2014) and Gardner et al. (2013) acquired different results over the Kangri Karpo Mountains, with glacier thickness loss of 1.34 ± 0.29 m a -1 , 0.81 ± 0.32 m a -1 and 0.30 ± 0.13 m a -1 during 2003 to 2009, respectively.Using SRTM DEM and TerraSAR-X/TanDEM-X acquisitions (18 February 2014 and 13 March 2014), glaciers experienced a mean thinning of 0.79 ± 0.11 m a -1 in Kangri Karpo Mountains.At a first glance, the result of glacier thickness loss in this study is in agreement with Neckel et al. (2014), and has significant differences with Kä ä b et al.

Field
measurement of mass balance is the best indicator of glacier change.A monitoring program has been carried out on Parlung No. 4 Glacier (5O282B0004/G096920E29228N) and Parlung No. 10 Glacier (5O282B0010/G096904E29286N), both located on the northern slope of the Kangri Karpo Mountains.Large ice deficit were found on the two monitored glaciers at rates of -0.71 m w.e. a -1 from May 2006 to May 2007 and -0.78 m w.e. a -1 during 2005 -2009, respectively horizontal distributions of surface temperature change and precipitation change from May to September derived from the CMFD data was shown in Fig. 9.It is clear that warming is a dominant phenomenon in the southeastern TP during recent decades.The warming rate on the northern slope of the Kangri Karpo Mountains is larger than that on the southern slope slightly.The law of precipitation change was inconsistent that an increasing trend was present in much of Kangri Karpo Mountains, but a decreasing trend in the eastern Kangri Karpo Mountains.The changes of air temperature and precipitation are confirmed by three nearest meteorological stations datasets, area of Kangri Karpo Mountains during 1980 -2012.The increase of precipitation results in more glacier accumulation, while glaciers have experienced an intense mass deficit in the detailed study area of Kangri Karpo Mountains.It can be concluded that other factors are playing a more important role in glacier mass deficit.For the change of temperature, warming was present in the detailed study area of Kangri Karpo Mountains during 1980 -2012.And the warming rate on the northern slope of the Kangri Karpo Mountains is larger than that on the southern slope slightly.A small warming rate was present from 1980 -2000, and increased to large warming rate thereafter.This is consistent with the tendency of glaciers change.Glaciers have experienced an intense area reduction and mass deficit in the Kangri Karpo Mountains, and mean mass deficit in the drainage basin of 5O282B (located on the northern slope of detailed study area) was larger than that in the drainage basin of 5O291B (located on the southern slope of detailed study area) during 1980 -2014.Meanwhile, the rate of glaciers shrinkage and mass loss from 1980 -2000 were lower than those from 2000 -2015.Hence, glaciers change in the Kangri Karpo Mountains can be attributed to climate warming.7 Conslusions This study estimated glacier area, glacier length, surface elevation and mass balance of the Kangri Karpo Mountains for the period of 1980 -2015 based on Topographic Maps, Landsat images, SRTM and TerraSAR-X/TanDEM-X acquisitions.Our results show that the Kangri Karpo Mountains contain 1166 glaciers, with an area of 2048.50 ± 48.65 km 2 in 2015.Ice cover in the Kangri Karpo Mountains diminished by 679.51 ± 59.49 km 2 (24.9% ± 2.2%) or 0.71% ± 0.06% a -1 from 1980 -2015.Comparing the terminal of all glaciers, only nine glaciers advanced while others retreated in the Kangri Karpo Mountains from 1980 -2015.Compared with the retreat of mountain glaciers in the western China, glaciers in Kangri Karpo Mountains have experienced extremely strong glacial retreat.The average elevation change of the entire glacier surfaces in the detailed study area of Kangri Karpo Mountains was -0.51 ± 0.09 m a -1 , or mean mass deficit of 0.46 ± 0.08 m w.e. a -1 from 1980-2014.Heterogeneous glacier mass balance was presented in the Kangri Karpo Mountains during 1980-2014.The comparison between filed measurements of mass balance and the result of this study indicated that there was high consistency of glacier mass loss about Parlung No. 4 Glacier and Parlung No. 10 Glacier.Geodetic mass balance measurements in the detailed study area of Kangri Karpo Mountains revealed that the debris-covered regions exhibit higher thinning rates than the clean-ice region on average obviously, with an average of -0.99 ± 0.09 m a -1 (-0.89 ± 0.08 m w.e. a -1 ) from 1980 -2014.There was a slight tendency that the rate of glaciers shrinkage and mass loss from 1980 -2000 were lower than those from 2000 -2015 in the detailed study area of Kangri Karpo Mountains.

Figure 1 .
Figure 1.Overview of study area and glacier distribution, including the locations of the detailed study area and meteorological stations.96 glaciers were selected to generate centerlines and calculate length change, and then be distinguished into advanced glaciers and retreated glaciers.

Figure 2 .
Figure 2. Location of the detailed study area, and distribution of TOPO DEMs, TSX/TDX acquisitions and ICESat footprints.Numbers indicate specific sample glaciers.5O282B and 5O291B indicate the drainage basins, where located on north slope and south slope of the detailed study area.

Figure 3 .Figure 4 .Figure 5 .2
Figure 3. Example of glacier outlines derived from imagery collected in 1980, 2000s and 2015.The background image is Landsat OLI image (6 October 2015).(A) Examples of glacier retreat.(B) An example of glacier advance.

Table 1 .
Overview of satellite images and data sources.5

Table 2 .
Statistics of vertical errors between the TOPO, SRTM and TSX/TDX.MED is mean elevation difference, STDV is standard deviation, N is the number of considered pixels, SE is standard error and σ is the overall error of the derived surface elevation change.

Table 3 .
Glacier area changes in the Kangri KarpoMountains from 1980Mountains from   -2015

Table 4 .
Area changes for ten sample glaciers in the detailed study area of Kangri Karpo Mountains.

Table 5 .
The length change of advanced glaciers in the Kangri Karpo Mountains.The uncertainty of glacier length in 1980 and 2015 are 6 m and 7.5 m, and the uncertainty of length change is 0.27 m a -1 .