Consolidating the Randolph Glacier Inventory and the Glacier 1 Inventory of China over the Qinghai-Tibetan Plateau and 2 Investigating Glacier Changes Since the mid-20 th Century 3

Glacier retreat in the Qinghai-Tibetan Plateau (QTP), the ‘third pole of the world’, has attracted the 11 attention of researchers worldwide. Glacier inventories in the 1970s and the 2000s provide valuable information 12 to infer changes in individual glaciers. However, individual glacier volumes are either missing, incomplete or have 13 large errors in these inventories, and thus, the use of these datasets to investigate changes in glaciers in QTP in the 14 past few decades has become a challenge, particularly in the context of climate change. In this study, individual 15 glacier volume data in the Randolph Glacier Inventory version 4.0 (RGI 4.0, 1970s) and the second Glacier 16 Inventory of China (GIC-II, 2000s) are recalculated and consolidated using a slope-dependent algorithm based on 17 elevation datasets for the QTP. The two consolidated inventories (The data are available under 18 https://doi.org/10.11888/Glacio.tpdc.270390 (Liu, 2020). For the time of review, the data will be accessible 19 through the following review link https://data.tpdc.ac.cn/en/data/4b88e394-0eb4-44c4-aa38-32aeb614daff/.) are 20 validated by comparing the observed and estimated glacier data reported in the literature. The two consolidated 21 glacier inventories are then compared for different mountains over the QTP to detect changes in glacier areas, 22 volumes, fragmentation status, etc. during the past 3-4 decades. Based on the results, the slope-dependent 23 algorithm performed well in computing individual glacier volumes and other elements, compared with the widely 24 used volume-area scaling which often leads to overestimation in the interior Plateau and underestimation in other 25 areas of the QTP in both RGI 4.0 and GIC-II. The comparison of the two inventories reveals a total area of glaciers 26 in the QTP of approximately 59026.5 km in the RGI 4.0 and 44301.2 km in the GIC-II. The total glacier volume 27 is 4045.9 km in the GIC-II compared with 4716.7 km in the RGI 4.0. The results suggest a significant retreat 28 and melting of glaciers in the QTP. However, variations are observed in different glaciers. The Karakoram 29 Mountains contain the largest number of surged glaciers, while the highest level of retreat is observed in the 30 Gandise Mountains. An increase in the fragmentation index is observed in the northern mountains, particularly the 31 Pamir Plateau, which displays the highest trends of glacier movement and deformation. The glacier volumes 32 decrease mainly on south-westward aspects and increase to various extents on the other aspects of most mountains. 33 The consolidation of the glacier inventories and the findings of the analysis performed in this study provide 34 important databases for future glacier-related studies, particularly for investigating the effects of climate change 35 on glaciers in the past and projecting future effects. 36 37

volumes, while the data in GIC-II contain some overestimations/underestimations compared with the observed 117 data. Meanwhile, the mean thickness of the glaciers is not provided in either inventory. These gaps must be filled 118 and the existing data in the two inventories must be verified to provide robust databases for glacier-related 119 studies.

120
This study has two aims. The first is to recover the individual glacier volumes over the QTP based on the 121 existing glacier information in RGI 4.0 and GIC-Ⅱ. A slope-dependent algorithm (the specific description is 122 provided in Section 4.1) is applied for the calculation. The recalculated glacier volumes will be validated with 123 the data from published studies and field observations. The second aim is to investigate the effects of climate 124 change impacts by comparing the two glacier inventories, which represent the statuses at different periods. The 125 results can provide a basis for understanding the glacier evolution in the QTP in the context of climate change.

126
Moreover, the comparison would be helpful to capture the association between glacial retreat or advance with 127 different atmospheric circulation patterns, which will enable a re-tracking the signal of historical climate change 128 and project the changes into the future. Himalayas even exceeds 8000 m a.s.l. In general, the average elevation over the entire QTP with total area of 136 approximately 2.5 million km 2 is greater than 4000 m a.s.l. Thus, the QTP has two nicknames: "the Third Pole" 137 of the earth and "The Water Tower of Asia".

138
The unique geomorphology of the QTP has largely resulted in the boundary discrepancy, with high mountains 139 located at the southwestern border and deep cuts located at the eastern margin. Due to the block of high

154
Glacier changes in the QTP are largely attributed to the changing regional water vapour and energy conditions 155 (Deng and Zhang, 2018;Qiu, 2008). The sources of water vapour over this region mainly include the Indian 156 Summer Monsoon, westerlies and East Asia Monsoon (Moor and Stoffel, 2013). In the context of global climate 157 change, these climate systems are altered, causing changes in the glaciers located in the QTP. Due to its 158 complicated topography and geomorphology, and monsoon-surrounded atmospheric circulation conditions, 159 regional warming over the QTP is quite substantial and three times higher than other areas in China (Qiu, 2008;   . The distribution of annual precipitation in the QTP shows a 165 decreasing trend from southeastern to northwestern areas (Qi et al., 2013). In general, the climate in the QTP 166 presents a pattern of warm-wet in the southeast and dry-cold in the northwest (Wang et al., 2002).           Table 1.
272 Table 1 General information about the data collections where VⅠ and VⅡ represent the glacier volume in GIC-Ⅰ and GIC-Ⅱ, respectively. The units of A and V are km 2 and 282 km 3 , respectively. VⅠ(Ⅱ)/A is applied to calculate the average thickness of a glacier.

283
As mentioned above, data are missing from the two inventories and problems of over and underestimations of aspect maps over in the QTP were applied to recover the missing data, determine ice thickness, and recalculate 286 glacier volumes in the two inventories as a method to improve the accuracy of the calculations.

287
In addition, there is an assumption that the grid with the maximum surface elevation in an individual glacier is 288 assumed to remain unchanged during two studied periods (Erasov, 1968;Gardelle et al., 2013;Frey et al., 2014).

289
The following specific procedures were used:  2) Use the following slope-dependent algorithm shown below (Fig. 2) to obtain the surface elevation map 293 based on the identified maximum elevation location (grid) identified in step 1.

294
① Identify the adjacent pixels of the grid with the maximum elevation recognized in step 1 (the distance 295 between centres of two grids' at a spatial resolution of 1km is equal to or less than 1.45 km (the largest centre 296 distance between two neighbouring pixels). These pixels are labelled as i, i＝1 2, ..., n1, with a surface elevation 297 Z1,i, and location (x1,i, y1,i)), and then calculate the surface elevation for these pixels.

306
where ZB, j is the bedrock elevation corresponding to Z2, j.
307 4) Based on the grid-based ice thickness, the individual glacier volume was computed using the following 308 equation: where ̅ is the pixel-averaged ice thickness.

310
In this process, the maximum surface elevation grid is synchronous to the grid cell of the maximum bedrock

329
A higher fragmentation index indicates that more surfaces are exposed to sunlight, which might result in more 330 energy accepted by glaciers to produce more meltwater. Meanwhile, the shear stress would also increase and 331 basal sliding would accelerate, which is the key interpretation of how the glacier movement and deformation will 332 develop.

333
In addition, the ratio of disintegrated glaciers (RDG) is computed as follows.

338
The following equation was used to calculate the average number of glaciers in the GIC-Ⅱ that disintegrated from 339 a glacier in the RGI 4.0.
where DGN presents the glacier number that disintegrated from the RGI 4.0 to GIC-Ⅱ.       Table 2. An underestimation is observed in the results obtained with the volume-area scaling.

374
In particular, the approximately 45.6%~58.4% rate of change in the total glacier volume has been underestimated

384
The calculated results are also compared with relevant studies in the QTP presented in the literature (Table 3). All    (Table 4).

441
The glacier information in the calculated RGI 4.0 and GIC-Ⅱ is compared to detect the disappeared and surged 442 glaciers with the aid of the ArcGIS toolbox (Fig. 4). The disappeared glaciers refer to glaciers that were included   the impact of uncertainty in glacier area induced by boundary pixels on the calculated glacier volume (Table 5).

580
The Qiangtang Plateau and Tangula

617
(1) The glacier volumes calculated using the slope-dependent algorithm perform better than the traditional

627
(4) An obvious offset of glacier volumes between different aspects is observed in most mountains. In general, 628 the glaciers on the western and southern aspects displayed a greater reduction in volume in the studied period.

629
Glaciers with increased volumes are mainly located on the northern and northeastern aspects in the northern 630 mountains, while the southern mountains have surging glacier volumes on the eastern and southeastern aspects.