Geometric Dependency of Tibetan Lakes on Glacial Runoff

Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | , ff ff 900 ∼ Sciences 900 Tibetan erent remote ∼ ff . For example, Discussions ff Earth System Hydrology and , feed Tibetan lakes 2 of glacier. Three case 2 50 000 km ∼ from at least 240 km 730 729 ff can be directly linked to hydrological processes. ff erentiate between lakes with and without outlet. In ad- ff ff . Unfortunately, accurate quantification of glacial changes ff 34 000 glaciers, which occupy an area of ∼ cult over the high relief Tibetan plateau. However, it has been recently shown that ffi This discussion paper is/has been under review for the journal Hydrology and Earth System Sciences (HESS). Please refer to the corresponding final paper in HESS if available. amount of glaciers. It also containslarge more part than of one the thousand water lakes and resources is of the South origin and of East a Asia, the most densely populated results also give the geometric dependencythere of are each 10 lake lakes on with glacialstudies, direct runo glacial including runo one over thedependency well-studied of Nam a Tso, lake on demonstrate glacial how runo the geometric 1 Introduction The Tibetan plateau is the highest and largest plateau of the world, and stores a large lakes. To obtain theselake results, mask we based on combinenetwork the the product MODIS derived so-called from MOD44W SRTM CAREERI water elevationysis, data. product glacier Based all and on mask, drainage a the drainage a links HydroSHEDS networkthat between anal- river 25.3 glaciers % and of lakes the are total glacier determined. area The directly results drains show into one of 244 Tibetan lakes. The is di it is possible toTibetan directly lakes assess greater water than levelsensing one changes products square of to kilometer. a This explicitlyThe create paper significant results links exploits part allow di between of us Tibetandition, the first glaciers, we to lakes di introduce and thedefined rivers. as notion the of ratio between geometricof the the dependency total catchment area of of of the a glaciers lake. draining lake These into dependencies on a are glacial determined lake and for runo all the area The Tibetan plateau isfrom an its essential sourceand of major water Asian for rivers South-Easthas like Asia. its Indus The impact and run-o on Brahmaputra. the Reported run-o glacial shrinkage likely Abstract Hydrol. Earth Syst. Sci. Discuss.,www.hydrol-earth-syst-sci-discuss.net/10/729/2013/ 10, 729–768, 2013 doi:10.5194/hessd-10-729-2013 © Author(s) 2013. CC Attribution 3.0 License. V. H. Phan, R. C. Lindenbergh, and M.Department Menenti of Geoscience and Remote2628 Sensing, CN Delft Delft, University The of Netherlands Technology, Stevinweg 1, Received: 12 December 2012 – Accepted: 11Correspondence January to: 2013 V. – H. Published: Phan 17 ([email protected]) JanuaryPublished 2013 by Copernicus Publications on behalf of the European Geosciences Union. Geometric dependency of Tibetan lakes on glacial runo 5 25 15 20 10 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | and ff per in- 2 . Krause ff er considerably ff 732 731 ect water level changes on the Tibetan plateau and ff erences in glacier status are apparent from region to region, ff . ff caused by loss of ice mass in the Tuotuo River basin. At the moment . The result shows that the simulated lake levels are in good correlation ff ff erence between two digital elevation models between 1999 and 2008. These ff erent parts of the Tibetan plateau. ff In general, water level changes of a Tibetan lake are caused by direct precipitation, In addition to glacier changes, in recent research lake level changes on the Ti- In this section, firstused. Conversions we in introduce data the formatwe origins or define coordinate and geospatial systems characteristics objects aretween of a mentioned such glacier all as as and well. a data a lake, Then glaciers products etc. lake and It lakes. catchment, is Finally also an shown indicatorsare how outlet, for defined to the and a determine computed. dependency connections connection between of be- a Tibetan lake on glaciers Tibetan lakes to glaciers. Thisdency enables on to determine glacial runo for each lake its geometric depen- 2 Data and methods to determine the lake levelglacier dependency runo on temperature, evaporation, precipitationwith and measurements registered since 2005to and 2009. satellite In radar addition, altimetryglacier Zhang data et runo from al. 2000 (2007)these studied types the of relation studies between arecause river necessary only runo measurements possible are for not individualWhat available lakes for is or most part river possible of basins, however the simply as Tibetan be- plateau. demonstrated in this paper is to geometrically link all showed that seasonal variations infor lake di levels and lake level trends di snow melt, glacial melt, moistureet conditions, al. evaporation (2010) and applied rainwater the runo hydrological system model J200g for the Nam Co Lake basin Co Lake in Krause et al. (2010) and Zhang et al. (2011). Moreover, Phan et al. (2012) GOS, 2012) and by in-situ measurements, e.g. Qinghai Lake in Li et al. (2007) and Nam betan plateau were observedet as al. well. (2011), As about describedSat/GLAS 150 LIDAR in campaigns water Phan between level et 2003 trendscated and al. that of 2009 (2011) most were the and estimated. of Tibetan The Zhang plateau the lakes result and lakes along indi- sampled have the by a Himalaya mountain thewater serious range, level ICE- downwards while trend trend most are in of in the theon the inner lakes the southern plateau with during a Tibetan Tibetan the positive plateau observingdescribed were period. in also Some Kropacek large monitored lakes et by al. radar (2012), altimetry, Qinghai e.g. Lake Nam and Co Seilin Lake, Lake, as as reported in (LE- Moreover, Gardelle et al. (2012)ring also at high revealed that rates ice inon the the thinning di central and Karakoram ablation andglacier is the occur- reductions Himalaya will mountain ranges, directlysurroundings. based a Mountains in the north-westported of in Tibet Wang during et thereduced al. period in (2011), of area 910 1950s rapidly glaciers todividual from in 2000s. glacier. 1956 the Also The to Middle glaciers as 2003, Qilian have re- of with also Mountain Yangtze retreated a Region River in mean have in the reduction the Tuotuothe of River inner Mt. basin, 0.10 km plateau, the Qomolangma from source Regioning 1968 of satellite to the laser 2002 Himalayas altimetry (Zhang inthe and et glacial a the al., thinning global last 2008), in elevation 35 model, the and yr Hindu Kaab in (Ye Kush-Karakoram-Himalaya et et region al. al., from (2012) 2009). 2003 quantified to Us- 2008. and surroundings has decreased significantlyal. in (2012), the systematic last di decades.with According the to most Yao intensive et shrinkageterized in by the the Himalayas greatest (excluding reduction theSorg in Karakoram) et charac- glacial length al. and (2012) area presented for the that past glacier 30 yr. Besides, shrinkage has occurred at the Tien Shan regions on earth. Recent studies concluded that the glacial area on the Tibetan plateau 5 5 15 20 10 25 20 10 15 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | , as also shown in Fig. 1. 2 : a larger limit would decrease the ff , as displayed in Fig. 1. erent ways whether an algorithm decided 2 ff 734 733 information, is directly derived from the drainage ff ers a suite of geo-referenced data sets (vector and ff , while the drainage basin data describes the catchment areas or the watershed area of the glaciers is approximately 53 236 km The HydroSHEDS data: ThedroSHEDS river (USGS, and 2012). drainage HydroSHEDS basin (HydrologicalSHuttle data data Elevation and is Derivatives maps at distributed based multipletion by on Scales) in Hy- provides a hydrographic consistent informa- plications. and HydroSHEDS comprehensive o format forraster), regional including and stream global-scaleand networks, ap- ancillary watershed data boundaries, layersogy such drainage information. as directions, It flow is accumulations, derivedMission distances, from (SRTM) elevation and at data river 3 of arc-second topol- the resolutionat Shuttle (a Radar the grid Topography equator). cell size Newlyfilling, of developed filtering, approximately algorithms 90 stream m have burning, beentions and were applied, made up-scaling including where techniques void- necessary while as manual described correc- in Lehner et al. (2006). glacier inventory was based onas observed reported data of in the Shia glaciers et from new 1978 al. version to (2009). of2007. 2002, Besides, The the original Shi glacier data et was inventoryEnvironmental al. collected and in and Engineering (2009) digitized a Research also by Institute, 5 the Chinese(CAREERI). mentioned yr Academy Cold The that project of and data Science Arid has accuracy Regions integration been is with started other not data, since mentioned we projectgraphic in Coordinate the its System glacier by coverage metadata using onto file. a theIn map WGS84 To summary, conversion Geo- enable the tool in glacier the mask ArcGISare contains Toolbox. 34 representative 676 glaciers, for stored the as glacier polygons which boundaries on the Tibetan plateau. The total jected Coordinate System based onprojection, the as detailed Krasovsky in spheroid its andglacier metadata the file. in Albers Its square attributes map meter, consist perimeter of the in area meter, of and each glacier identification codes. The The MODIS land-water mask: Thewas water produced mask, using called MODIS aover MOD44 6 combination yr W 250 of of m, AquaFor over each MODIS 8 250 spectral yr m pixel, data of it and isthat Terra indicated the SRTM MODIS in pixel elevation di represents spectral data water. (GLCF,visible. data, Except 2012). The for the lakes lakes, from also theLandsat some mask parts TM are of images. checked rivers In with are are Google this removed. Earth way, Then parts and the of appropriate lakesside rivers over limit and one of empty kilometer one depressions square kilometernumber are or square of selected. holes is The lakes a down- in trade-o possibilities the of analysis, the while 250 m applyingsults MODIS in a 891 land-water smaller lakes mask limit over too onea would kilometer total much. stretch square area This the on of procedure the approximately Tibetan re- 38 plateau 800 km whichThe occupy CAREERI glacier mask:Center for The Glaciology glacier and maskdistributed Geocryology, in is Lanzhou GIS delivered file (WDC, format by 2012). as the ArcInfo This coverage World product data. It Data is is referenced to the Pro- ff – – – Only rivers with upstream drainageare areas exceeding selected. a The threshold of river 100 data upstream cells is in line vector format where each line is formed by a The river data, providing surface runo direction layers. The flow accumulationriver layer data is used is for built selection at and attribution. 15 arc-second The resolution (approximately 500 m at the equator). runo boundaries on the Tibetantions plateau. between The lakes river and network linksdata between is are glaciers used projected and onto to lakes. the analyze To WGS84 integrate the Geographic them, connec- Coordinate all System. these Main data sources usedREERI in glacier mask, this and papermask the include determines HydroSHEDS the the river locations of MODIS and theof drainage land-water Tibetan the lakes. basin mask, The Tibetan data. glacier the glaciers. mask The gives The CA- water the river outlines data provides information on the direction of surface 2.1 Data 5 5 25 15 20 10 25 15 20 10 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | , + er- ff . 2 736 735 in the example catchment converges to Kekexili Lake and the purple ff A lake sub-catchment is combined from smaller catchments draining to the lake. Catchments drain into other catchments in a hierarchical pattern, with smaller catch- As a matter of fact, there exist discrepancies between notably the CAREERI glacier The drainage basin data, describing the catchment areas or the watershed bound- river segment is representative forin the Fig. origin 5. of A the node glacier-melt of drainage, the as river illustrated network is used to be representative for the outlet of a lake catchment is the point wheresequently there the exists a boundary river of segment the leaving Yinma the lake Yinma catchment lake. is2.2.2 Sub- formed, as shown Determining in connections Fig. between 4b. glaciers andBased lakes on theanother river can network, be an determined.and Accordingly, oriented determining a route lake a corresponds of connectionthe to between river river finding a segments network a glacier and route from the from one outlet an node of origin a where to lake the catchment. glacier In drains most into cases, a from-node of a (USGS, 2012). Each rivera segment river drains segment to of an anthe outlet number adjacent which of catchment. can accumulated In also upstreamgrid this grid be analysis, cells cells, an each of involving origin river the the segmentriver of number green stores segments of catchment from the upstream adjacent itself catchments and draining the to number it. of The accumulated outlet grid of cells the of Yinma lake there is one rivercatchment that is leaves part a of certaini.e. a lake, the bigger the Yinma catchment. lake lake An isalso catchment example not shown for in is a Fig. this sink. a 2. case Therefore sub-catchment is the of the lake the Yinma Kekexili lake, As lake shown catchment, in as Fig.100 4a, upstream cells each draining green totension catchment a of is river ArcGIS, representative segment based as for on determined the an using HydroSHEDS area the DEM of ArcHydro data at ex- at least 15 arc-second resolution boundary describes its closedis catchment, surrounded i.e. by Kekexili Lake a is geographical barrier a such sink. as Each a catchment mountainments, also ridge. called sub-catchments, combining toapplication, larger a catchments. Depending sub-catchment on the can be determined accordingly. In classifying lakes, if all surface runo described in (DeBarry, 2004).sink, In can be a a closed lake catchment, or a the point single where point, water is also lost called underground. a As illustrated in Fig. 4, lakes and the HydroSHEDS rivers and catchments match the2.2 Landsat image. Methods 2.2.1 Determining the catchment of aA Tibetan catchment, lake also known as drainagethe basin surface or watershed, water is defined fromoutlet, as where rain the area the and where water melting joins snow another water or body ice such converges as to a a lake, river, single ocean, point etc, or as MOD44W lakes, the HydroSHEDSposed river over segments a and true drainagecollected color basins on are image 26-08-2002. superim- composed It iscompatible from with evident bands that their 1, the position 2, CAREERI onences and glacier the are positions Landsat 3 in are image. the of In not average Landsat-ETM order this fully of example, maximum approximately 2 di km. On the other hand, the MODIS aries, is also builtvectors, as at shown 15 in Fig. arc-secondSystem. 2. resolution. It Catchments This is are also product attributed referencedLake is to with catchment the formatted an occupies WGS84 area as Geographic an Coordinate in area polygon of square 2636.5 kilometer, km e.g. themask Kekexili and the otherdroSHEDS data, data. for This instance is the illustrated MODIS in MOD44W Fig. water 3, mask where and the the CAREERI Hy- glacier outlines, the The river network isriver referenced reach to or the segment WGS84a has Geographic number a Coordinate of pointer System. grid to Each the cells. its Kekexili For corresponding Lake example, flow catchment, the accumulationindicated. and inset given in the as Fig. flow 2 route describes from the Yinma river Lake network to in Kekexili Lake is from-node (or a starting-point), a list of vertices and a to-node (or an ending point). 5 5 25 15 20 10 25 15 20 10 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | , to 2B 1 d , . How- 2A to each ff d 1 glacier-melt drainage, 2 . 1 so from-node E is representative for is the minimum distance of 3D 2B d d 738 737 is the minimum distance from a glacier G 1A d may also drain its glacier-melt water via from-node 2 is smaller than 3E d glacier- melt drainage. 3 are assumed to belong to catchment Cat 2 , so from-node B is considered the origin of the G glacier- melt drainage. Similarly, 2C 1 d and G 1 nation: Each river segmentnetwork, is either no an or orientedThus one vector the river and oriented segment route at leaves fromof the each a a node, lake node source catchment as as of down illustrated to aning the a in origin the destination river Fig. of as following another 6. procedure. glacier-melting outlet Firstly or canwith the an be the river determined source outlet belongs segment us- to whose thecides from-node route. with coincides Then the if from-node the of to-node anis of adjacent collected the river to river segment, form segment the the coin- route. adjacentsegment This river coincides process segment is with repeated the until destination. theroute to-node For from of example source the in river E Fig.route to 6a, includes destination when two A, segments finding point ED the andconsists D DA. of is Similarly, the two the route segments point from FD F of and to conjunction, DH. H and in the Fig. 6c C. In the other case, the origin of the G Identifying the outlet of athe lake lake catchment: region. The outlet If of riveris each segments a lake all sink has stream to of to beof a the inside a closed outlet closed catchment. inside catchment, In the thereforeleaves Fig. point lake, the A 6a the lake and or lake and drains point c, toand H point another c, is lake A point a or C or sink. river, or the point If point lake one H is F river is not isIndicating segment the a the the sink. outlet outlet In of oriented Fig. the 6b route sub-catchment of of the river lake. segments from a source down to a desti- and although in fact the glacier G Determining which catchment a glacieracteristics belongs of to: catchment due boundaries, to each the glacier geographical only char- belongs to one catchment. that contains the largest partG of the glacier. For example, in Fig. 5Estimating the two the glaciers originglacier of directly drains the to glacier-melt theever, outlet drainage: in of in one the catchment reality glacier throughthe surface mask melt-water melt runo each from water glacier from a route is the of digitized glacier river as drains segments.the to In an from-node this the undivided of study, outlet, polygon, thefor a only so source each river following of catchment, one segment the the oriented from-node, which route distances e.g. is is from nodes assumed nearest a A, toa glacier to B be polygon as the in and a glacial Fig. a polygon,polygon. 5, outline. point e.g. The are Thus is from-node G computed, the which where has minimumof minimum a distance the distance distance from drainage to the between route. the Apotential point glacier distance from-nodes. is to In the threshold a source Fig. from vertex 5, thepotential of glacier from-nodes the A is and used B, tothe so G pre-select the from-node A is assumed to be the origin of However, there exist discrepanciesthe between HydroSHEDS the catchments, as CAREERI mentionedsects glacier in Fig. with outlines 3. more and Therefore than if a one glacier catchment, inter- it is assumed to belong to that catchment 3. 4. 2. 1. to another: each route with its attributes of the from-lake code and the to-lake code. The module resultspresents in an shapefiles oriented in routeidentification GIS from a codes polyline source of vector to thecases format a source to where destination. and Its determine: each attributes the (i)tributes polyline consist destination. a of of The connection the the module from glacial is a code applied glacier and in to two a the lake: lake each code, route and with (ii) its a at- connection from one lake an oriented routecatchment. from To determine the these outlet glaciermodule of – in lake the a ArcGIS and lake environment lake that catchment – includes lake to the connections, four the we procedures build below. outlet a of another lake catchment, as illustrated in Fig. 6. Similarly, a connection between lakes is considered 5 5 25 15 20 10 25 15 20 10 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | (1) . is earth’s 2 R lat equals (lat2–lat1) as ∆ , referenced to the Projected 2 ) , while C } A 2) / lon ∆ ( 2 740 739 sin ∗ (lat1, lon1), (lat2, lon2) cos(lat2) { ∗ ) 35.62 degree, and therefore the grid cell size is estimated at a 6371 km). ∼ − = 1 lon equals (lon2–lon1) as a cell width in radians, and cos(lat1) . Alternatively, the Yinma Lake catchment can also be obtained from ∆ √ 2 , + a 2) √ / lat is the shortest distance over the earth’s surface – giving an “as-the-crow-flies” 0.6433 km. ∆ tan2( ( c × d 2 a · · the Kekexili Lake catchment is reported toAs occupy an the area of product 2636.5cells. km of Similar to the the grid computation of cell the size area with of the the Yinma total Lake number sub-catchment of upstream grid Directly from the attributes of the HydroSHEDS drainage basin data. For example, Calculating the gridheight, cell varies size regularly in dependingmated meters: using on the the its “haversine” grid formula latitude. (Sinnott, cell 1984) The below: size, grid including cell width size and is approxi- Obtaining the number ofshould be total determined, based upstream on grid performing the cells: module identifying First outlets of for lakes each lake an outlet cells flowing into sink Aand is into outlet 1000 cells, F into isbe sink null. In H calculated the is manually case 500 of using cells,below. outlet into ArcHydro, In F, outlet the practice, as the area C mentioned number of is of in thiscatchment, 850, total lake as the grid catchment illustrated discussion cells can in representative section Fig. for the 4b, is Yinma Lake 3775 grid cells. as mentioned above. Then theof total the number numbers of of upstream the gridto upstream cells directly grid the derived cells outlet from is of river segments the the draining sum lake. For example in Fig. 6, the total number of upstream grid R 2 sin = According to the information above, the area of the Yinma Lake catchment is approx- Obtaining the area of a closed lake catchment: the area of a closed Tibetan lake Following Sinnott’s formula above, the grid cell size depends on the latitudes of the Computing the area of a lake sub-catchment: fortunately, the HydoSHEDS river data 1. 2. 2. 1. = = c a d imately 658.7 km its geospatial boundary, as illustratedthe in area Fig. of 4b. the AccordingCoordinate Yinma to System Lake its WGS84 catchment geospatial UTM. is boundary, also 658.7 km catchment can be determined by three methods. sentative for the whole lakedata sub-catchment. As is mentioned built in atlocated the data at 15 arc-second section latitude the resolution, of river 0.3766 the outlet of the Yinma Lake catchment is Where distance between the points a cell height and radius (mean radius grid cell. The Tibetanof lake the catchments outlet may actually of occupy the large lake areas. sub-catchment The is latitude used to compute the grid cell size, repre- Fig. 6. Subsequently the areaof of each the lake grid catchment can cellriver be size segments calculated with as converging to the the product the totalfollowing outlet number steps of have of the to the lake. be That upstream performed. is grid for cells each sub-catchment obtained the from all ments inside the inner Tibetan plateauto are closed the catchments. Lake closed catchments catchments belong Ganges, or Indus, one Irrawaddy, of Mekong,drainage the Salween, catchments basin Yangtze, or of data Yellow the River. onlyriver major Because catchments, distributes rivers in the shapes Brahmaputra, this andcalculated study, areas as areas follows. of of the closed-catchments lake sub-catchments or need big toprovides be explicitly the number of upstream grid cells of each river segment, as illustrated in between glaciers and lakes at the Kekexili catchment. 2.2.3 Calculating the area of aBased lake on catchment the ( HydroSHEDS drainage basin data, it is concluded that most of the catch- Figure 7 shows the result of the module to determine connections between lakes, and 5 5 25 20 15 10 15 10 25 20 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 2 (4) (5) (2) (3) ) and GD A ff is the total j GD erence. Therefore in ff indicator at Yinma Lake. As is the ratio between the area D ff R indicator for Kekexili Lake is 0.035. U R -th upstream lake flowing to the lake. j 742 741 ) draining to it. , referenced to the Projected Coordinate System GU indicator at Yinma Lake equals 0.064, while the 2 A D indicator equals the , as shown in Fig. 7. Because Yinma Lake is the only R 2 U R -th glacier directly draining to the lake, and A i presents the geometric dependency of a lake on upstream U R j GD A 1 . = m 2 j X 0.6433 km. The total number of grid cells in the Kekexili Lake catchment . erences in catchment areas are caused by using the representative grid 2 + indicates the geometric dependency of a lake on glaciers directly draining × ff i i D A A is the area of the R i 1 1 C C n n = = GU GD i i X X A A A A Calculated from its geospatial boundary. Foroccupies example the an Kekexili Lake area catchment of 2636.8 km above, the grid0.3767 cell sizeis 15 100 representative grid cells. Thus for2635.3 the km area the of the Kekexili Kekexili Lake catchment Lake is approximately catchment is WGS84 UTM. = = indicator for Kekexili Lake equals 0.019. Because there isn’t any glaciers-fed lake = = Subsequently the total area of glaciers directly draining into Yinma Lake is 41.9 km 3. U D D GD GU R Following the results above,R the draining into Yinma Lake, the Yinma Lake is upstream of Kekexili Lake, the glaciers. R An indicator for the dependency ofin a the lake on catchment glacier occupied runo equals by zero, glaciers the lake and catchment thenot doesn’t fed lake contain by catchment any glaciers glaciers, areafully at meaning itself. covered all. that by If the If glaciers. the lake According the to ratio is indicator indicator the is total area close of to glaciersto one, draining it the to while the lake the lake, catchment the indicator is almost and into Kekexili Lake is 50.1upstream km lake of Kekexiliequals Lake, 92 km the total area of2.2.5 upstream glaciers of Determining Kekexili the geometric Lake dependency of a lake on glacier runo Where A area of directly contributing glaciers of the the total area of upstream glaciers ( A A Based on the distributionpart of of the the Tibetan glacier glaciers melt-watercollect as flows glacier-melt downward shown water to in directly some Fig. from offor glaciers 1, the each or Tibetan it lake, indirectly lakes. is A we from obvious lake upstream distinguish that can lakes. the Thus total area of directly contributing glaciers ( The small di cell size. The bigger thethe area following, the of the area catchment,the of drainage the each basin larger Tibetan data. the lake di closed-catchment is directly2.2.4 derived from Computing the total area of glaciers draining into a lake 5 5 15 20 10 20 15 10 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | . In 2 geometric . D ff 2 R and the total area of indicators are also de- GD geometric dependency are potential to drain to U indicator values occur for U 2 R R U R and 9 sinks, which are indirectly indicators are symbolized by red ff D R indicator values because these lakes are indicator of under 0.005. Based on the dif- D U 744 743 R R drain to streams which are origins of Mekong River, . There are 150 lakes with an area of over 50 km 2 2 . They are obviously located near glaciers and spread along indicator under 0.005, corresponding to 75 % of lakes with 2 indicator, the geometric dependency of a lake on direct glacial D , there are 13 lakes with runo presenting the dependency of a Tibetan lake on glaciers have R D R U GU R A and value of over 0.5. These eight lakes are all relatively small, occupying D D R R , is determined for all Tibetan lakes. The ff ered by an upstream lake. Moreover, case studies studying the glacial dependency ff Subsequently the the same lakes havingsources the of highest the eight river network. In addition toglaciers being through directly other fed upstream lakes. bytermined Accordingly, to the glaciers, present 266 a the geometric TibetanThe dependency result lake of of can Tibetan grouping lakes beon the on upstream Tibetan fed upstream lakes glaciers glaciers. indirectly according is by one to shown upstream their in glacier Fig. correspond 9 toference as between an well. the total About area 75 of %upstream directly glaciers of contributing glaciers the A lakesfed with by at glaciers least but not directly. Moreover the highest eight mountain ranges in the southerna and list western of Tibetan plateau. the Besides, top Table 10 3 lakes shows ranked by3.1.3 total area of Geometric directly dependency contributing of glaciers. Tibetan lakes on upstream glaciers runo disks in Fig. 8. The resultdependency of on the directly grouping contributing the glaciers Tibetanof is lakes the by also level shown lakes of in the haveat Fig. an 9. least Accordingly, one most glacierhave draining an directly into it.each approximately The 2 km result also indicates that eight lakes ment of the Brahmaputra River,33 11.1 % lakes of while its the glacierBrahmaputra rest area River, of directly passing glaciers drains through of intoglaciers China, approximate one of 14 India of 000 approximate km its and 316 Bangladesh. km supporting Similarly, fresh most water for of China, Myanmar, Laos, Thailand, Cambodia and Vietnam. lakes, mostly situated in the north and the northwest of the inner plateau. For the catch- directly drains to one of111 the sinks. 244 In lakes. the These inner lakes plateau, consist 37.4 of % 133 of upstream the lakes glacier and area directly drains to one of 160 two third of the Tibetanalso lake water four is to contained five in timesPlateau sinks. keeps as On 86 average, big the sinks as sink of the lakes a are lakes total with of3.1.2 96 outlet. sinks The having region an Geometric defined area dependency as of of over the Tibetan 50 Inner km lakesBased on on direct the glacier spatial distribution runo ofment glaciers is and catchments, computed the as glacierTibetan shown area per plateau in catch- are Table 2. considered. In According this to case Table 2, only 25.3 the % major of catchments the of total the glacier area The Tibetan plateau containsarea 891 of lakes approximate over 38 one 800Table km 1 square all kilometer, Tibetan occupying lakes a areas divided total a in sink sinks and if lakessink, it with a is outlet. lake A the has lake termination an is outlet point considered that of drains water water flow into another in lake a or closed a catchment. river. Apparently, If over it is not a bu of three lakes are included: thethe Aksai Chin Nam Lake in Tso the Lake North-Westernthe near Kunlun south Mountain, 100 of km Tibet. north of Lhasa and3.1 the 72 km Lake long dependency Yamdrok on Lake glaciers in for3.1.1 the full Tibetan plateau Classification of lakes with or without outlet After defining the geospatial objectsindicators and coding the proceduresbeen introduced computed above, for the the wholeare Tibetan directly plateau. fed The by result glaciers shows while that 266 244 lakes Tibetan have lakes at least one upstream glacier, possibly 3 Results 5 5 25 15 20 10 25 20 15 10 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | from (Phan ff 1 indicator − during the 0.096. We D 1 R = − U yr R . It is located at . 3 2 ff 23 cm yr + is mostly direct, i.e. ff of which approximately the total area of glaciers 2 , as derived from the Hy- 2 GD A value equals U . ff . As there is only one small lake R for the period between November 2 , as derived from the HydroSHED 3 2 of glaciers upstream Aksai Chin Lake for its highest geometric dependency D GU R A . Nam Tso is a sink at the foot of the Nyain- 2 746 745 . Second the Nam Tso Lake closed catchment 0.084 while its indicator value of 0.031, which indicates that over D = D R D R R (Zhang et al., 2011). In addition, based on analysis of op- 1 − . So the dependency of Aksai Lake on glaciers is determined. (1991, 2001, 2005, 2009), Bolch et al. (2010) report that the 2 . Geospatial properties of this lake are characteristic for Tibetan + value of 0.816, presenting the dependency of a lake on direct ff value gives D R 25 cm yr D + R , occurs for a relatively small lake, occupying only 2 km . It is one of the top 10 lakes directly fed by glaciers, as shown in Table 2. 2 ff is covered by glaciers. Thus this lake is almost fully fed by glacial melt water. 2 . Its sub-catchment occupies an area of about 118 km ff indicator value of over 0.5. According to Krause et al. (2010), the sum of all water inflow to Nam Tso Lake re- The maximum D and Landsat TM/ETM sulted in an increase of1961 the and lake October volume 2010. by Krause 33.5inflow km et of al. water (2010) from also indicated glaciersobserving that into period, the Nam and mean Tso total that waswater annual this computed volume. glacial as Moreover, melt 7.12 based km watertween on is 2003 the analysis and largest of 2009, contributoret satellite Nam to al., Tso laser the 2011), has altimetry or lake a datatical positive data from lake from be- level Hexagon trend KH-9 of and Landsat MSS (both 1976), Metric Camera (1984), basin data, while theas total 334.5 km area of directThis high glaciers value draining is into reflected3 by Nam % an Tso of the is Nam calculated value Tso also catchment confirms is the covered high by dependency glaciers. of Subsequently, Nam this Tso on glacial runo 3.2.2 Nam Tso closed catchment Nam Lake, also calledplateau. Nam The Tso lake or isoccupies Nam a located surface Co, area at is of (30.718qentanglha about N, the Mountains 1960 km and 90.646 largest E) is salt mostly atTso fed lake an by has on glaciers elevation in the two of theseclosed Tibetan small mountains. 4718 catchment Besides, m, upstream Nam occupies and lakes, an but area no of glacier 10 741 km drains into them. The Nam Tso runo 96 km The geographic properties of thisR lake are representative for the top eight lakes with an shown in Fig. 10. This lake is the only lake draining into Aksai Chin Lake that has glacial glacier runo (35.293 N, 80.572 E) at a height of approximately 5500 m in the Kunlun Mountains, as droSHED drainage basin data. Thedirectly results draining indicate to that Aksaiupstream Chin of Aksai Lake Chin equals Lake, the 673is km total approximately area, 769 km Accordingly its conclude that the dependency ofalmost Aksai not Chin tempered Lake by on intermediate glacial lakes. runo Aksai Chin(35.208 Lake N, 79.828 is E) in avast the high-altitude sink desert south at of onis an fed the by average the Aksai Kunlun elevation River of Mountains.Lake Aksai and closed 5500 many Aksai m. catchment Chin other Chin In streams, occupies as fact, plateau. an is illustrated Aksai area in largely The of Chin Fig. a about 10. Lake lake 8000 The km Aksai is Chin located at lakes having top indicatoris included values as itance. has Nam been Tso a Lake mostly pilot dependstanglha for on many Mountains. directly studies contributing Finally glaciers in the on lakecapped Yamdrok the level mountains, Lake Nyainqen- change but sub-catchment and Yamdrok is Lake watersurrounding highly surrounded bal- lakes depends by rather on snow- than upstream from glacial direct runo glacial runo 3.2.1 Aksai Chin Lake closed catchment Case studies present specificLake catchments and of Yamdrok three Lake.small lakes lake First Aksai with the Chin the Aksai maximumon Lake, indicator direct Nam Chin value glacial Tso Lake runo closed catchment contains the 3.2 Case studies 5 5 25 15 20 10 25 15 20 10 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | D . R ff . The while 2 and is 2 2 . It is located at 2 , as derived from the 2 indicator, while results D indicator value of 0.004. R 0.026. It means that Yam- U U R R value equaling 0.125. from Gongma Lake and Phuma indicator, the geometric depen- . Subsequently the dependency 2 ff D D R R , only equals 0.002. However, the total ff is encoded by our 748 747 ff . Thus the 2 passing first through a nearby small lake with the ff from upstream lakes more than on direct glacial runo ff . It is located at a height of 4500 m near the western end of 2 . Therefore Phuma Lake also highly depends on direct glacial 2 of the sub-catchment area is covered by glaciers. Therefore the dependency 2 . ff Phuma Lake, also called Puma Yumco, is a big upstream lake draining to Yamdrok Gongmo Lake is one of the lakes upstream of Yamdrok Lake. The lake occupies The results also indicate that three close by lakes Bagyu, Gongmo and Phuma flow It should be noted that Nam Tso is an exception among the many lakes on the catchments from a digital elevation model. In Sect. 2.2.3, the areacell of size a with lake the sub-catchment total is numberoutlet of computed of accumulated as the grid the sub-catchment. cells product Due drainingwork, of into to however, the the the the grid representative limited outlet resolution ofof of a a the river lake network, HydroSHED sub-catchment e.g. river can outletcumulated net- F grid be in cells just Fig. cannot one 6c. beNevertheless, In of determined this the several particular automatically sources area case, from the of thegeometric total HydroSHEDS a shape. number data. of lake The ac- ArcHydro sub-catchment extension can of also ArcGIS be supports the calculated manual based outlining on of its 4 Discussion 4.1 Calculating the area of a lake sub-catchment using ArcHydro manually of Gongmo Lake on direct glaciers is high, with the Lake, as shown inlocated Fig. at a 12. height Phumamountains. of The Lake 5030 area m. occupies of The an the lakeresult Phuma area is indicates Lake directly of sub-catchment that fed equals about the around byLake 285 1815 total melt is km km water area about from of 153 surrounding km runo directly contributing glaciers draining to Phuma Lake. an area of aboutYamdrok 40 Lake. km The Gongmo Lake77.7 km sub-catchment occupies an area of 620 km Yamdrok Lake has a high dependency on glacial runo indicator value of 0.004. That is, Bagyu Lake has the feed Yamdrok Lake occupy onlydency 21 of km Yamdrok Lake onarea direct of glaciers glacial upstream runo ofof Yamdrok Yamdrok Lake Lake is on 255 upstream km drok glaciers Lake is depends relatively on high, runo into Yamdrok Lake, asLake shown depends in on Fig. glacial 12. runo Although no glacier directly feeds it, Bagyu (28.979 N, 90.717 E) at anest/Qomolangma. elevation The of lake is about fed 4440 m byYamdrok Lake numerous on sub-catchment small the is streams. at north The the side outletThe far of stream Yamdrok western Lake of Mount end sub-catchment the Ever- of occupies theHydoSHED an lake, drainage as area shown basin of in data 9940 Fig. km River. and 12. Although belongs surrounded to by many the snow-capped major mountains, the catchment glaciers of that Brahmaputra directly from other papers indicateThe possible a significance link of these betweenlakes links glacial however. should shrinkage be studied and further lake for a level large increase. number3.2.3 of Yamdrok Lake sub-catchment Yamdrok Lake, also called Yamzhoplateau. Yumco, The is lake one is of fan-shaped and the occupies largest an lakes area of on about the 640 km Tibetan shrink during the period 2001–2009. Tibetan plateau, in the sensefor that Nam it Tso indicate is the relativelydependency potential well of of studied. Nam the In Tso approach our on of opinion this glacial the paper. runo results Indeed, the geometric glaciers from the Nyainqentanglha Mountains draining into the Nam Tso catchment 5 5 20 25 15 10 25 15 20 10 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | is calculated for ff . Based on drainage ff and therefore indicate which lakes are . Still, our results clearly list which lakes ff ff 750 749 ected by the predicted further shrinkage of the glaciers on ff This work was jointly supported by the EU-FP7 project CEOP-AEGIS computations Schneider, C.: A glacier inventory for the western Nyainqentanglha range and the Nam Co Based on the HydroSHED DEM data at 15 arc-second resolution, the geometric Bolch, T., Yao, T., Kang, S., Buchroithner, M. F., Scherer, D., Maussion, F., Huintjes, E., and expected to be stronglythe a Tibetan plateau. Acknowledgements. (grant number 212921) and by the Vietnam Ministry of Education and Training. References network analysis, geometric connectionsThen between the glaciers total and area of lakes directlydraining are contributing glaciers determined. into or a the total lake areaactual of is dependency upstream of glaciers computed. a Thisare lake geometric more on dependency glacial or is runo less just dependent a on proxy glacial for runo the 5 Conclusions In this paper, the geometricthe dependency complete of each Tibetan lake plateau. oncontributing The glacial glaciers runo results and indicate 266 thatthe lakes 244 total depend area lakes on of depend upstreambe glaciers on glaciers. representative draining directly The for into ratio the a between lake dependency and of the a area lake of on its glacial catchment runo is used to several virtual outletsnetwork in analysis. the Then river thesimilar network module to to to make connections reduce make between connectionsbetween the lakes, has between glaciers time been the and applied. of virtual lakes Finally,from the the in outlets, glaciers connections performing to each outlets major the and catchment between are outlets. found by combining routes run for several times manually. Moreover, for each major river catchment we created the plateau into sub-areas,means grouping that some the module closed to catchments determine in connections between the glaciers inner and lakes plateau. has It been or on a laptopprocess with the Core data for 2Although the Duo the whole CPU river Tibetan 2.2 network plateau. GHz isis Sometimes and organized the requested per 4 GB processing catchment, to got RAM, a findcatchment large stack. it of a amount took Brahmaputra route of River 4 PC of occupiesout memory or a river large 5 densely. segments area, Besides, days for and most to thedon’t each of river drain glacier. network the Especially, into spreads glaciers the thedirectly major inside outlet distributing the of glaciers Brahmaputra draining any major into lake each catchment catchment. lake Thus on to the calculate whole the plateau, total we area divided of 4.2 Dividing the Tibetan plateau into smaller parts for speedingThe up the Tibetan plateau is anetwork large analysis region, so module, itexample determining takes when connections so the long between time module glaciers to was perform and run the lakes. on drainage For a desktop with CPU 3.2 GHz and 2 GB RAM this purpose, a thresholdsegment, of to 30 improve upstream the drainagecatchments level grid with of an cells detail area is oflake of used at the sub-catchment to least river is define 30 network. a mergedillustrated grid It river from in cells also Fig. will the 4. means be small Subsequently,shape. built. that its catchments For Finally small area draining example, for can the each into be geometricthe lake its obtained south shapes the directly outlet, of of from the as the Palku its three Lake geometric closed small catchment lake are sub-catchments determined in as shown in Fig. 13. shape of lake sub-catchments in theing case into the total the number lake ofArcHydro upstream cannot grid tools. be cells In drain- determined practice,the automatically there terrain are exist data created 19 for manuallyondly such the using the small 19 the lakes catchments lake directly are regions created fed in are by steps clipped following glaciers. the from Firstly ArcHydro the user guide. HydroSHED For DEM. Sec- 5 5 20 15 10 25 15 20 10 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | inventory. variation in the ff , 2012. (last access: Novem- erent glacier status with atmo- ff . http://www.legos.obs-mip.fr/produits/soa/ (last access: November 2012), 2012. doi:10.1038/NGEO1450 751 752 ¨ ugel, W.-A., Kang, S., and Gao, T.: Hydrological 50 years, J. Earth Sci., 22, 539–548, 2011. ∼ (last access: November 2012) 2012. , 2010. http://landcover.org/data/watermask el, M., Solomina, O., and Beniston, M.: Climate change impacts on ff in Tien Shan (Central Asia), Nat. Clim. Change, 2, 725–731, 2012. http://wdcdgg.westgis.ac.cn/chinese/DATABASE/Glacier/glacier ff Carte-Asie.html of (last access: June 2012) 2012. as observed by ICESat laserRemote altimetry, ISPRS Sensing – and International Spatial Annals Information of Sciences, the 1, Photogrammetry, 237–242, 2012. glaciers and runo lakes from space inChina, Journal Mt, of Qomolangma Mountain Region and Science, of 6, the 211–220, 2009. Himalayas on the Tibetan plateau in Lake Qinghai, West China during2007. recent decades, Int. Ser. Prog. Wat. Res., 21, 1505–1516, lakes between 2003 and 2009, Int. J. Appl. Earth Obs., 17, 12–22, 2011. B., Pu, J., Lu, A.,spheric Xiang, circulations Y., Kattel, in D. B., Tibetan2012. and Plateau Joswiak, and D.: Di surroundings, Nat. Clim. Change, 2, 663–667, Tuotuo River basin, the source region59–68, of 2008. Yangtze River in western China, Environ. Geol., 56, and phased uplift of the62–66, Himalaya 2011. – Tibetan plateau since Late Miocene times, Nature, 411, hydrologie-radiometrie/hydrologie-continentale-par-altimetrie-spatiale/Base-Hydroweb/ Cartes/copy ology, Lanzhou, water storage at Nam Co Lake, central Tibetan Plateau,on J. Hydrol., the 405, Tibetan 161–170, Plateau 2011. 115, using 1733–1742, 2011. ICESat altimetry data (2003–2009), Remote Sens. Environ., Int. J. Appl. Earth Obs., 17, 3–11, 2012. tion data, Eos, Transactions, AGU, 89, 93–94, 2008. of Northwest China during the last system analysis anddoi:10.5194/adgeo-27-29-2010 modelling of the Nam Cochanges basin in Nam in Co in Tibet, central Tibet Adv. utilizing synergistic Geosci., satellite altimetry 27, and 29–36, optical imagery, http://gisdata.usgs.gov/website/HydroSHEDS/ htm ber 2012), 2012. twenty-first-century glacier mass change in the Himalayas, Nature, 488, 495–498, 2012. twenty-first century, Nat. Geosci., 5, 322–325, & Sons, Inc., , 2004. Basin, Tibet, and glacier changeshttp://www.the-cryosphere-discuss.net/4/419/2010/ 1976–2009, The Cryosphere, 4, 419–433, 2010, Sorg, A., Bolch, T., Sto Phan, V. H., Lindenbergh, R. C., and Menenti, M.: Seasonal trends in Tibetan lake level changes Shi, Y., Liu C., andSinnott, Kang, R. E.: W.: The Virtues glacier of Inventory the of Haversine, China, Sky Ann. and Glaciol., Telescope, 50, 68, 1–4, 159, 2009. 1984. Ye, Q., Zhong, Z., Kang, S., Stein, A., Wei, Q., and Liu, J.: Monitoring glacier and supra-glacier Phan, V. H., Lindenbergh, R. C., and Menenti, M.: ICESat derived elevation changes of Tibetan Li, X., Xu, H., Sun, Y., Zhang, D., and Yang, Z.: Lake-level change and water balance analysis at Yao, T., Thompson, L., Yang, W., Yu, W., Gao, Y., Gou, X., Yang, X., Duan, K., Zhao, H., Xu, Zhang, Y., Liu, S., Xu, J., and Shangguan, D.: Glacier change and glacier runo Zhisheng, A., Kutzbach, J. E., Prell, W. L., and Porter, S. C.: Evolution of Asian monsoons Zhang, G. Q., Xie, H. J., Kang, S. C., Yi, D. H., and Ackley, S. F.: Monitoring lake level changes LEGOS, Hydrology from space: Zhang, B., Wu, Y., Zhu, L., Wang, J., Li, J., and Chen, D.: Estimation and trend detection of Lehner, B., Verdin, K., and Jarvis, A.: New global hydrography derived from spaceborne eleva- WDC, Chinese Glacier Inventory – World Data Center for Glaciology and Geocry- Wang, P., Li, Z., and Gao, W.: Rapid shrinking of glaciers in the Middle Qilian Mountain region Kropacek, J., Brauna, A,. Kang, S., Feng, C., Yeb, Q., and Hochschild, V.: Analysis of lake level Krause, P., Biskop, S., Helmschrot, J., Fl USGS, Hydrological data and maps based on Shuttle Elevation Derivatives at multiple Scales, Kaab, A., Berthier, E., Nuth, C., Gardelle, J., and Arnaud, Y.: Contrasting patterns of early GLCF: MODIS Water Mask, Gardelle, J., Berthier, E., and Arnaud, Y.: Slight mass gain of Karakoram glaciers in the early DeBarry, P. A.: Watersheds: Processes, Assessment and Management, Chapter 1, John Wiley 5 5 30 25 25 20 15 20 10 15 10 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | (%) Total ) R 2 ) 2 (km Total A is the total area of glaciers Total ) sinks (km 2 A ) glacier-fed lakes 2 754 753 is the ratio between the total area of glaciers with direct ) area (km Total 2 R lakes upstream lakes (km in a lake and ff Total 605 12296.0 286 26502.8 Glacier area per catchment on the Tibetan plateau. Tibetan lakes with and without outlet. name area (km Total 2 531 240 53 236 244 13 492.8 25.3 No. Catchment1 Upstream2 Total area of Brahmaputra3 Ganges 784 Indus5 Irrawaddy6 14 Mekong Sinks7 1535.3 Salween Total area8 0 of 28 Yangtze9 3 Yellow River 78.5 16 Inner plateau 56 87 323 0 1333.5 15.3 3 253.4 2165.6 5949.0 965.4 53.6 2 5 330.1 0 0 1 212.7 2 260 13 0 20560.7 0 17.7 4170.1 1157.9 in a lake and the catchment area. ff 12 Brahmaputra3 Ganges4 Indus 3445 528 Irrawaddy6 Mekong7 Salween8 15 677 39 772 Yangtze9 101 Yellow 428 River 4227 Inner plateau 86 392 108 266 3 636 263 928 1 098 484 382 317 2430 33 32 1893 327 26 512 2432 297 1748.2 10 11.1 14 355.5 0 160 4 2 727.9 18 3 9909.7 9.8 30.0 0.0 53.4 11.0 520.0 167.1 37.4 21.4 0.0 2.8 3.4 56.4 No. Catchment Catchment Total glacier No. of directly Table 2. runo with direct runo Table 1. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

D R is the D R ) 2 GD is the area of A C A ) (Km C 2 A ) (Km 2 . ﬀ area (Km

756 755 is the total area of glaciers directly draining into the lake, and GD A Top 10 lakes ordered by total area of directly contributing glaciers.

Glaciers and lakes on the Tibetan plateau. No. Lake name12 Dongtaiji’nai’er3 Lake Aksai Latitude4 37.496 Chin Kul Ligmen Longitude5 Tso 93.935 Ngagong6 Tso Ayakum Kul7 35.208 Nam8 Tso Lake 79.828 Draksum9 Tso 35.028 29.413 Nganglaring10 Tso 223 81.082 Achik 96.817 Kul 37.546 Dabsan Nor 34 148 31.540 89.373 30.026 691.5 30.718 166 83.101 0.020 93.997 90.646 249 7993 36.978 37.067 9 672.8 95.205 631 2727 88.431 0.084 518.7 500 1290 24 1967 147 26 0.190 484.6 383.7 12 464 10 741 0.376 0.016 291.2 1722 334.5 294 355 0.023 307.2 0.031 109 629 0.178 13 263 242.7 280.8 0.002 0.021

Glaciers and lakes on the Tibetan plateau the Tibetan on lakes and Figure 1. Glaciers Figures Fig. 1. the lake catchment, Table 3. geometric dependency of the lake on direct glacial runo 608 609 610 607 606 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

y a . Notably a discrep- + ETM+. Notabl ETM+. - water mask data and and data mask water - 758 757

between the CAREERI glacier outlines and the location of the glaciers on the on glaciers the of location the and glacier outlines CAREERI the between Tibetan catchments derived Tibetan from HydroSHEDS catchments and the river at network the CAREERI glacier mask data, MODIS MOD44W land-water mask data and Tibetan catchments derived from HydroSHEDS and the river network at the Kekexili

CAREERI glacier mask data, MODIS MOD44W land MOD44W MODIS data, mask glacier Figure 3. CAREERI riversHydroSHEDS basinLandsat superimposed over and outlines discrepancy observed. be can image Landsat 618 619 620 621 616 617

Figure 2. catchment Kekexili HydroSHEDS rivers and basin outlines superimposed over Landsat-ETM Fig. 3. ancy between the CAREERIimage glacier can be outlines observed. and the location of the glaciers on the Landsat Fig. 2. catchment. 612 613 614 615 611 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

nodes A, B and E and E nodes B A, the -

3 (b)

and G and 2 , and from 1 , G 1

. 3 second resolution, and b) The and b) second resolution, , and from-nodes A, B and E corre- - 1 and G elt drainage G elt drainage 2 ,G 1 760 759 belonging to catchment Cat belonging to catchment 2 and G belonging to catchment Cat 1 2 catchment as part of the Kekexili Lake closed catchment closed Lake the Kekexili of as part catchment - and G 1 Catchments representative for an area of at least 100 upstream cells draining to a a) Catchments representative for an area of at least 100 upstream cells draining to a cells draining upstream 100 at least of area an for representative FigureCatchments 4. a) roSHEDS DEM Hyd datathe basedriversegment on at 15arc sub Lake Yinma

Glaciers G 628 629 630 624 625 626 627 622 623

Glaciers G Figure 5. Glaciers the of glacier tocorresponding origins m sponding to origins of the glacier melt drainage G Fig. 5. Yinma Lake sub-catchment as part of the Kekexili Lake closed catchment. Fig. 4. (a) river segment based on the HydroSHEDS DEM data at 15 arc-second resolution, and 633 634 635 631 632 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

catchment 762 761 The sink A or H of a closed catchment and the outlet C or F of a sub-catchment. Determining the glaciers inside the Kekexili catchment, the connections from glaciers to

a closed a of sub or- H of catchment and sink the C F Figure or A outlet 6. The

from connections the catchment, Kekexili the inside glaciers the Figure 7. Determining Lake to Kekexili Lake Yinma route from the and Lake, Kekexili and Lake to Yinma glaciers Yinma Lake and Kekexili Lake, and the route from Yinma Lake to Kekexili Lake. Fig. 7. Fig. 6. 636 637 638 639 640 645 646 642 643 644 641 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

on U R and D R .

ﬀ

764 763 of of Tibetan lakesdirect runoff glacial on D . ﬀ Grouping Tibetan lakes by level (%) of their geometric dependencies The geometric dependency RD of Tibetan lakes on direct glacial runo