A Long-Duration Glacier Change Analysis for the Urumqi River Valley, a Representative Region of Central Asia

: The increasing global warming trend has resulted in the mass loss of most glaciers. The Urumqi Vally, located in the dry and cold zone of China, and its widely dispersed glaciers are significant to the regional ecological environment, oasis economic development, and industrial and agricultural production. This is representative of glaciers in Middle Asia and represents one of the world’s longest observed time series of glaciers, beginning in 1959. The Urumqi Headwater Glacier No. 1 (UHG-1) has a dominant presence in the World Glacier Monitoring Service (WGMS). This paper supplies a comprehensive analysis of past studies and future modeling of glacier changes in the Urumqi Valley. It has received insufficient attention in the past, and the mass balance of UHG-1 was used to verify that the geodetic results and the OGGM model simulation results are convincing. The main conclusions are: The area of 48.68 ± 4.59 km 2 delineated by 150 glaciers in 1958 decreased to 21.61 ± 0.27 km 2 delineated by 108 glaciers in 2022, with a reduction of 0.47 ± 0.04 km 2 · a − 1 (0.96% a − 1 in 1958–2022). The glacier mass balance by geodesy is − 0.69 ± 0.11 m w.e.a − 1 in 2000–2022, which is just deviating from the measured result ( − 0.66 m w.e.a − 1 ), but the geodetic result in this paper can be enough to reflect the glacier changes ( − 0.65 ± 0.11 m w.e.a − 1 ) of the URB in 2000–2022. The future loss rate of area and volume will undergo a rapid and then decelerating process, with the fastest and slowest inflection points occurring around 2035 and 2070, respectively. High temperatures and large precipitation in summer accelerate glacier loss, and the corresponding lag period of glacier change to climate is about 2–3 years.


Introduction
As the largest reservoirs of freshwater resources, glaciers are critical to the hydrological cycle and ecosystems because of their sensitive and synergistic relationship with the global climate [1,2].Glaciers store solid water during cold temperatures and melt during the ablation period to supply water to rivers and downstream populations, especially in cold and dry regions.[3,4].However, the global sea level has been rising at a rate of 0.2-0.4mm per year during the 20th century due to the accelerated shrinkage of the cryosphere; this is as a result of man-made climate change [5,6].A proposal to designate 2025 as the International Year of Glacier Protection, initiated by the Republic of Tajikistan, has been implemented, and a further plan to establish an international fund for glacier protection emphasizes the importance and urgency of glacier change [7].
There are two aspects to the study of systematic refinement of glacier change in a particular region: past systematic analyses and future change trends.The past period of glacier shrinkage is manifested in many aspects, such as reduction in number, area, and volume, while the mass balance is one of the most direct quantitative reflections of glacier

Data and Pre-Processing
The earliest glacial boundary extraction was based on six topographic maps involving this basin that were scanned, aligned, spliced, and corrected based on ENVI 5.3 to ensure that the corrected mean-variance was less than one cell before delineation.All topographic maps have a scale of 1:50,000.Landsat imagery has been widely used to study glacial change [34]; given the difficulty of obtaining early high-resolution imagery, we chose Landsat TM imagery from 1989.The Landsat images were downloaded from the website abbreviated as USGS and have been pre-processed, including radiometric and control point correction.(https://glovis.usgs.gov/).This paper studies glacier changes based on higher-resolution multi-source remote sensing images, such as the Spot 5 in 2005 and Ziyuan No. 3 stereo pairs in 2022.These images had < 10% clouds, occurred within the ablation period, and were used to generate glacial contours.The data used for past glacier changes in the Urumqi Valley are presented in Table 1.Data are available free of charge.SRTM DEM is readily available as open-source data and, being within 1 month of 2000, has a unique advantage for studying glacier elevation change with the 30 m resolution (https://earthexplorer.usgs.gov/).The ALOS satellite from Japan is dedicated to earth

Data and Pre-Processing
The earliest glacial boundary extraction was based on six topographic maps involving this basin that were scanned, aligned, spliced, and corrected based on ENVI 5.3 to ensure that the corrected mean-variance was less than one cell before delineation.All topographic maps have a scale of 1:50,000.Landsat imagery has been widely used to study glacial change [34]; given the difficulty of obtaining early high-resolution imagery, we chose Landsat TM imagery from 1989.The Landsat images were downloaded from the website abbreviated as USGS and have been pre-processed, including radiometric and control point correction.(https://glovis.usgs.gov/).This paper studies glacier changes based on higher-resolution multi-source remote sensing images, such as the Spot 5 in 2005 and Ziyuan No. 3 stereo pairs in 2022.These images had <10% clouds, occurred within the ablation period, and were used to generate glacial contours.The data used for past glacier changes in the Urumqi Valley are presented in Table 1.Data are available free of charge.SRTM DEM is readily available as open-source data and, being within 1 month of 2000, has a unique advantage for studying glacier elevation change with the 30 m resolution (https://earthexplorer.usgs.gov/).The ALOS satellite from Japan is dedicated to earth observation.The PALSAR satellite has the advantage of round-the-clock, all-weather land observation, and was selected for the elevation difference analysis (https://search.asf.alaska.edu/).To measure glacier mass balance changes, this paper takes elevation changes since the 21st century at 10-year intervals.The China Daxigou Meteorological Station, at the source of the Urumqi River, provides meteorological data that can reveal synergistic changes in glaciers and climate.Data are available free of charge.

Delineation of Glaciers and Uncertainty Assessment
Delineation of accurate glacial boundaries is important when glaciers are used to represent glacial changes in a region.Automated edge delineation is determined by many factors such as snow cover, clouds, shadows, and the resolution of images [35].In this study, glacial margins in different periods were delineated to meet the recognized standards [36].In addition to combining the experience of the experts in the field, Google Earth pro 2023 software also contributed [37].Clustered glaciers are separated by ridgelines and referred to glacier cataloging to retain subsequent accuracy.
The error in delineating the glacier can be regarded as caused only by the resolution of the remotely sensed images after the glacial boundaries have been captured by the visual interpretation method, which is the most accurate method available.The uncertainty of delineation EA can be evaluated as follows: where N is the number of pixels and λ is the resolution of different remote sensing images.The uncertainty of changed area (E B ) is expressed as follows [38]: where E 1 and E 2 are uncertainties in glacier area at the beginning and ending years, respectively.

Ziyuan No.3 DEMs
Ziyuan No.3 was launched in January 2012 as a forerunner of China's civilian highresolution stereo mapping satellites, with three panchromatic cameras and a multispectral camera, which can acquire stereo image pairs of the same area from three different observation angles.The resolutions are 5.8 m multispectral, 2.1 m panchromatic, and 3.5 m front and rear view.Based on this feature, which can provide rich 3D geometric information, it can be used to generate DEMs, and Resource 3-02 has a higher front-and rear-view resolution of 2.5 m, which can meet the requirements of higher precision stereo mapping.Considering the seasonal snow and cloud cover, we chose four pairs of images from July and September 2022 to extract the DEM.The process of extracting the DEM using stereo image pairs was carried out in the DEM extraction model of the ENVI 5.3 software.After comparing the extraction results, we found that the use of frontal-view and rear-view image pairs to extract the DEM provided better results.The extraction process was carried out by selecting 20 ground control points and 100 correlation points in relatively stable nonglaciated areas, using Landsat 9 OLI images of the same period as reference positioning.All DEMs are georeferenced in the WGS84 coordinate system and stitched together with a spatial resolution of 2.5 m.

DEMs Co-Registration and Error Analyses
In the geodetic method, the elevation difference between DEMs is the value that characterizes the variation of glacier surface elevation; this is fundamental for our assessment of glacier mass balance in the Urumqi Valley.The ALOS DEM in the intermediate periods was selected as the reference DEM to assess the SRTM DEM and ZiyuanNo.3DEM.The 1964 glacier margin was chosen to differentiate non-glacier topography, as the glacier area shrank the most in 2022.The rationale for the alignment is that the topography of the non-glaciated regions remains essentially unchanged, which can be used to test and iteratively adjust for errors in the multi-source DEMs.The relationship between vertical deviation and slope and aspect will tend to sine or cosine if the DEMs in non-glaciated areas are not perfectly aligned with each other [39].The offset difference equation proposed by Nuth is as follows: In order to display the results more intuitively and conveniently, we linearize Equation (3) and then solve Equation (4) to obtain the value of the offset in each direction, thus fulfilling the DEM alignment.dh tan(α) = xsin(φ) + ycos(φ) + z tan(α) x = a × sin(b) In Equations ( 3)-( 7), dh represents the height difference between DEMs of different periods due to offset displacement: a is the magnitude of vertical movement, b is the direction of the movement vector, α is the slope of terrain, φ is the vertical direction of the terrain, and c represents the mean deviation between the different DEMs divided by the mean slope tangent, which can be calculated by dh/tan(α).After linearisation, x represents the offset in EW direction, y represents the offset in the NS direction, and z is the offset in the vertical direction.After aligning the multiphase DEM with the topographic information of the non-glacier area, if elevation residual error satisfies the normal distribution, then the elevation standard deviation residual in the non-glaciated stable area approves of the estimate of the accuracy of the surface elevation change in the glacier area [40].

Geodetic Mass Balance and Uncertainty Assessment
Density should be taken into account when volume changes are converted to glacier mass balance [15,41].This paper applied 850 ± 60 kg•m −3 to provide a parameter for taking the snow and pure ice into account to assess the mass changes by water equivalent (w.e.) [42]: where ρ is ice density transition; S is consequential glacier area; n is pixel numbers; ∆hi is the single pixel height variation; and Si is single pixel area.The uncertainty was calculated as follows: where ∆h is the mean height variation of glacier areas; t is studied period; ∆ρ is the error of ice density, which value is taken as 60 kg•m −3 ; ρ w is water density, which value is taken as 1000 kg•m −3 ; σ is the errors of height variation; and ρ 1 is ice density transition, which value is taken as 850 kg•m −3 .

Glacial Mass Balance and Uncertainty
In 1959, we began to measure the glacier mass balance during ablation (May~August) by using flower sticks to confirm the height variation and snow pits to assess the density [43,44].As illustrated in Figure 2, the stake network spans the whole UHG-1 in order to provide the result of mass balance.For UHG-1, we manually interpolated the measured data between adjacent contours by manually drawing lines of equal mass balance [45].
where Si is the pixel area, Bi is the obtained mass balance of the corresponding pixel glacier, and S is the total area of UHG-1.
by using flower sticks to confirm the height variation and snow pits to assess the density [43,44].As illustrated in Figure 2, the stake network spans the whole UHG-1 in order to provide the result of mass balance.For UHG-1, we manually interpolated the measured data between adjacent contours by manually drawing lines of equal mass balance [45].
where Si is the pixel area, Bi is the obtained mass balance of the corresponding pixel glacier, and S is the total area of UHG-1.

OGGM Model
Open Global Glacier Model gives full consideration to glacial geometry and consists of an explicit ice dynamics module and a calving parametrization.The modular OGGM supports being redefined, remixed, and repeated, with reliance on publicly available data for calibration and validation.In the model, glacial boundaries and DEM data are used as model base input data.It is driven by climate scenarios to forecast future area, volume changes, etc.The model is globally consistent in predicting ice thickness and glacier mass loss.To maintain the purity of the results, the observed glacier mass balance is the best choice for checking the accuracy of OGGM.Note that model accuracy is verified by predicting glacier mass balance using an extended temperature index model [21] () =   • ()  −  •  (() −   , ) +  (11) where characters with (h) qualify the elevation, M is monthly glacier mas balance, Pcf is precipitation correction factor, Psolid is monthly solid precipitation, and µ is updated calibration and correction factor, which represents the glacier sensitivity parameter and

OGGM Model
Open Global Glacier Model gives full consideration to glacial geometry and consists of an explicit ice dynamics module and a calving parametrization.The modular OGGM supports being redefined, remixed, and repeated, with reliance on publicly available data for calibration and validation.In the model, glacial boundaries and DEM data are used as model base input data.It is driven by climate scenarios to forecast future area, volume changes, etc.The model is globally consistent in predicting ice thickness and glacier mass loss.To maintain the purity of the results, the observed glacier mass balance is the best choice for checking the accuracy of OGGM.Note that model accuracy is verified by predicting glacier mass balance using an extended temperature index model [21] where characters with (h) qualify the elevation, M is monthly glacier mas balance, P cf is precipitation correction factor, P solid is monthly solid precipitation, and µ is updated calibration and correction factor, which represents the glacier sensitivity parameter and where a particular glacier can be set to agree with observations [20].T is monthly temperature, Tmelt represents monthly mean air temperature, and ε is deviation correction.

Meteorological Data
In this study, three SSPs scenarios were selected to drive the OGGM model.In order to reduce the error of the results, for each SSPs pathway, the results of 13 climate models were selected, which were BBC, CAMS, CESM2-WACCM, CESM, EC-Earth3, EC-Earth3-Veg, FGOALS, GFDL, INM-CM4-8, INM-CM5-0, MPI-ESM1-2-HR, MPI-ESM2-0, and NorESM2-MM.The final results of glacial area and volume in the URB are obtained by averaging the standard errors of the 13 climate models; ensembles are taken as the errors of the area and volume simulations results [31].Given the assumptions that climate change arises only over large areas and the associations of climate variables remain consistent in the base period, interpolating or inserting data where none exists with thin-plate spline eliminates errors.

Delienation and Uncertainty of Glaciers
In 2022, 108 glaciers were mapped in the entire URB, a decrease of nearly one third compared to 150 in 1964.The number of glaciers throughout the basin has been trending downward over time.Moreover, the whole basin can be divided into four sub-basins, coded as 5Y730A, 5Y730B, 5Y730C, and 5Y730D.We discerned and statistically calculated the contours of the glacier in detail and combined the number of pixels occupied by the glacier and the image resolution to obtain the uncertainty in the delineated and vanished glacier area.Specific information is presented in Tables 2 and 3.Because the region belongs to multiple field observations, the error of manual visual interpretation is ignored in this paper.Higher resolution images are the basis and direction for the development of glacier remote sensing for the future.The areas of glaciers delineated in different years of this paper and the errors are shown below:

Area Changes and Analysis in Glaciers
The glacier area in the basin decayed rapidly, from 48.68 ± 4.59 km 2 in 1964 to 21.61 ± 0.26 km 2 , a shrinkage rate of 0.47 ± 0.04 km 2 •a −1 and an overall relative rate of change in glacier area of 55.6% (0.96%•a −1 ).Because the glacier scale in this region is inherently not very large, it is believed that the glacier will be completely ablated.In this paper, the rate of glacier area change is divided into three periods, and the ablation rate is −0.31 in the periods of 1964-1990, 1990-2005, and 2005-2022, respectively.The years from 1990-2005 had the fastest rate of glacial ablation, and the ablation rate for the latest period has remained above the average rate for the last 58 years.Moreover, four sub-basins, coded as 5Y730A, 5Y730B, 5Y730C, and 5Y730D, involved 67.75%, 54.96%, 55.32%, and 50.65% of the ablation area from 1964 to 2022, respectively.The detailed ablation shape is shown in Figure 3.
The Urumqi Valley is on northern Tianshan, so most glacier numbers (106) and glacier area (20.79 ± 0.24 km 2 ) belong to the N, NE, and NW facings.The glacier area oriented towards N had the most ablation, accounting for 58% of the total.The NE and NW oriented glacier areas are about equally ablated, accounting for about 54%.The two glaciers in other orientations had an the area of only 0.81 ± 0.02 km 2 , and were much smaller in 2022.(Figure 4).The mean elevation for the glaciers inventoried ranged from 3667 to 4272 m, with the mean being 3948 m in 2000; they ranged from 3671 m to 4270 m with the mean 3926 m in 2022.About 80% of the glacier area was distributed within 3800-4000 m.Over 50% of the glacier area was between 3900-4000 m.The area of glaciers is roughly normally distributed at different elevation scales.The most serious ablation of glacier area is in the altitude zone below than 3700 m, accounting for 76.55%, followed by 4000-4100 m, accounting for 68.47%, while the smallest rate of ablation is in the most aggregated altitude zone of 3900-4000 m, accounting for 52.5%, which reflects the intense ablation of glacier area in the region.The Urumqi Valley is on northern Tianshan, so most glacier numbers (106) and glacier area (20.79 ± 0.24 km 2 ) belong to the N, NE, and NW facings.The glacier area oriented towards N had the most ablation, accounting for 58% of the total.The NE and NW oriented glacier areas are about equally ablated, accounting for about 54%.The two glaciers in other orientations had an the area of only 0.81 ± 0.02 km 2 , and were much smaller in 2022.(Figure 4).The mean elevation for the glaciers inventoried ranged from 3667 to 4272 m, with the mean being 3948 m in 2000; they ranged from 3671 m to 4270 m with the mean 3926 m in 2022.About 80% of the glacier area was distributed within 3800-4000 m.Over 50% of the glacier area was between 3900-4000 m.The area of glaciers is roughly normally distributed at different elevation scales.The most serious ablation of glacier area is in the altitude zone below than 3700 m, accounting for 76.55%, followed by 4000−4100 m, accounting for 68.47%, while the smallest rate of ablation is in the most aggregated altitude zone of 3900−4000 m, accounting for 52.5%, which reflects the intense ablation of glacier area in the region.

Geodetic Uncertainty Analysis
After correcting the offset bias among SRTM, ALOS, and ZiyuanNo.3DEM data with a trigonometric correction (Figures 5 and 6

Geodetic Uncertainty Analysis
After correcting the offset bias among SRTM, ALOS, and ZiyuanNo.3DEM data with a trigonometric correction (Figures 5 and 6), the elevation difference residuals in the non-glaciated areas can be used to evaluate the errors among the DEMs and to calculate the accuracy of the estimation results for the glacier volume change and the mass balance.Because the spatial resolution of the DEM data used is not uniform, the spatial autocorrelation distances were chosen to be 600 m, 250 m, and 50 m for the SRTM, ALOS, and ZiyuanNo.3resolutions of 30 m, 12.5 m, and 2.5 m, respectively (Bolch T. et al., 2011).After error correction, Mean Elevation Difference (MED) tends to be close to 0; the relative error between the DEM data is significantly less than 1 m (Table 4).

Geodetic Uncertainty Analysis
After correcting the offset bias among SRTM, ALOS, and ZiyuanNo.3DEM data with a trigonometric correction (Figures 5 and 6), the elevation difference residuals in the nonglaciated areas can be used to evaluate the errors among the DEMs and to calculate the accuracy of the estimation results for the glacier volume change and the mass balance.Because the spatial resolution of the DEM data used is not uniform, the spatial autocorrelation distances were chosen to be 600 m, 250 m, and 50 m for the SRTM, ALOS, and Zi-yuanNo.3resolutions of 30 m, 12.5 m, and 2.5 m, respectively (Bolch T. et al., 2011).After error correction, Mean Elevation Difference (MED) tends to be close to 0; the relative error between the DEM data is significantly less than 1 m (Table 4).Theoretically, the difference embodied in each raster of the DEMs can respond to a single point of glacier mass balance.Given the corrected results, the glacier elevation in the URB declined significantly in various periods, with an accelerated trend of glacier retreat.The following figure is a visualization of the difference between the DEMs of the

Geodetic Glacier Mass Loss
Theoretically, the difference embodied in each raster of the DEMs can respond to a single point of glacier mass balance.Given the corrected results, the glacier elevation in the URB declined significantly in various periods, with an accelerated trend of glacier retreat.The following figure is a visualization of the difference between the DEMs of the two periods processed by the Arcmap 10.3.
After discounting obviously erroneous extremes from the results of the subtraction of the DEM, most of the variables are tightly clustered around the average value, with only a few at the extremes (Figure 7).

Predicted Future Glacier Mass Loss in Urumqi River Basin
The simulation and analysis of glacier area and reserves in the Urumqi River basin under different climate scenarios of CMIP6, including SSP1-2.6,SSP2-4.5, and SSP5-8.5, found that the trends of glacier area and reserves changes under different emission scenarios coincide with each other.In 2020-2080, the glacier area and reserves in the Urumqi River Basin of the Tianshan Mountains still show an overall trend of retreat and decrease (Figure 8).Reserves as a whole still show a trend of retreat and decrease, and both basically disappear completely by the end of this century.Under the three climate scenarios, the initial and final values of SSP1-2.6,SSP2-4.Taking the mid-century (2050) as the cut-off point, the rate of glacier area decrease in the first half of this century (2020-2050) is higher than that in the second half of this century (2050-2080) under the emission scenarios of SSP1-2.6,SSP2-4.5, and SSP5-8.5 (0.482 km 2 •a −1 , 0.485 km 2 •a −1 , and 0.505 km 2 a −1 , respectively); the rate of decrease in glacier area decrease is 0.353 km 2 •a −1 and the rate of decrease in glacier storage/volume decrease is 5.29×10 6 m 3 •a −1 .The rate of decrease of glacier reserves in the first half of this century (2020-2050) is higher than that in the second half of this century (2050-2080) by 0.226 According to the results of the glacier height variations converted to glacier loss, the volume of glaciers in the Urumqi Valley has decreased by 0.24 km 3 over the past 22 years, with an annual mean decrease of 0.01 km 3 •a −1 .The annual glacier mass balance of Urumqi Valley was −0.65 ± 0.11 m w.e.a −1 during 2000-2022.Of this, the mean glacier mass balance was −0.67 ± 0.12 m w.e.a −1 before 2010 and was −0.63 ± 0.11 m w.e.a −1 after 2010.As a special mention, we also calculated the individual glacier mass balance separately for the key monitoring glacier (the UHG-1).The average annual glacier mass balance of the UHG-1 was −0.69 ± 0.12 m w.e.a −1 during 2000-2022.The average glacier mass balance was −0.71 ± 0.12 m w.e.a −1 in 2000-2010 and was −0.67 ± 0.12 m w.e.a −1 in 2010-2022.

Predicted Future Glacier Mass Loss in Urumqi River Basin
The simulation and analysis of glacier area and reserves in the Urumqi River basin under different climate scenarios of CMIP6, including SSP1-2.6,SSP2-4.5, and SSP5-8.5, found that the trends of glacier area and reserves changes under different emission scenarios coincide with each other.In 2020-2080, the glacier area and reserves in the Urumqi River Basin of the Tianshan Mountains still show an overall trend of retreat and decrease (Figure 8).Reserves as a whole still show a trend of retreat and decrease, and both basically disappear completely by the end of this century.Under the three climate scenarios, the initial and final values of SSP1-2.6,SSP2-4.5, and SSP5-8.5 are approximately the same, but the intermediate trends are very different.For glacier area, the SSP1-2.6 scenario has the slowest rate of glacier retreat, with an average annual decrease in change of 0.351 km 2 •a −1 and an average annual decrease in glacier volume/volume of 5.21 × 10 6 m 3 •a −1 .Next is the SSP2-4.5 scenario, with an average annual decrease in change of 0.352 km 2 •a −1 and an average annual decrease in glacier volume/volume of 5. 23 × 10 6 m 3 •a −1 , and the SSP5-8.5 scenario, with a decrease in glacier area/volume of 0.352 km 2 a −1 and an average annual decrease in glacier volume/volume of 5.23 × 10 6 m 3 •a −1 .SSP5-8.5 scenario had the largest glacier area and reserve melt.The glacier area change decreases by 0.353 km 2 •a −1 and the glacier reserve/volume decreases by 5.29 × 10 6 m 3 a −1 .
Remote Sens. 2024, 16, x FOR PEER REVIEW 12 of 20 area retreat, followed by thickness reduction.This pattern is even greater in the middleand late-century.The reason for this is that the glaciers in the Urumqi River Basin are inherently thinner than the glacier properties in other regions.

Geodetic Glacier Mass Balance Validation
The mass balance obtained from the UHG-1 based on traditional glaciological measurements is shown in  Taking the mid-century (2050) as the cut-off point, the rate of glacier area decrease in the first half of this century (2020-2050) is higher than that in the second half of this century (2050-2080) under the emission scenarios of SSP1-2.6,SSP2-4.5, and SSP5-8.5 (0.482 km 2 • a −1 , 0.485 km 2 •a −1 , and 0.505 km 2 a −1 , respectively); the rate of decrease in glacier area decrease is 0.353 km 2 •a −1 and the rate of decrease in glacier storage/volume decrease is 5.29 × 10 6 m 3 •a −1 .The rate of decrease of glacier reserves in the first half of this century (2020-2050) is higher than that in the second half of this century (2050-2080) by 0.226 km 3 •a −1 , 0.224 km 3 •a −1 , 0.232 km 3 •a −1 , and the percentage of decrease is higher than that in the second half of this century (2050-2080) by 31%, 30%, and 31%, respectively.Comparing the glacier area and glacier reserve/volume, the percentage reduction of glacier area is about 10% higher than the percentage reduction of glacier reserve in the first part of this century (2020-2050), and the percentage reduction of glacier area is about 40% higher than the percentage reduction of glacier reserve in the second part of this century (2050-2080), indicating that the glaciers in the Urumqi River Basin are mainly characterized by glacier area retreat, followed by thickness reduction.This pattern is even greater in the middle-and late-century.The reason for this is that the glaciers in the Urumqi River Basin are inherently thinner than the glacier properties in other regions.

Geodetic Glacier Mass Balance Validation
The mass balance obtained from the UHG-1 based on traditional glaciological measurements is shown in Figure 9

OGGM Model Validation and Forecasting
To prove the accuracy of the future changes in glacier area and volume, we used the OGGM model to acquire the glacier mass balance from 2000 to 2020, which is displayed in Figure 10.The OGGM model and the observed values show a near-identical trend, with a correlation coefficient of 0.87, which has a strong significance.In addition, the difference between the linear fit lines of OGGM simulated and the observed values have a tendency to widen; the error range of 52-124 mm w.e.a −1 . is acceptable.After confirming that the results of glacier area and volume were credible, we calculated the glacier mass balance based on the formula and combined it with the snow-ice density transformation to get the future glacier mass balance value of the URB.The results show that the glacier mass balance is the most negative in the SSP5.8-5pattern, followed by SSP2.4-5, and is the least negative in SSP1.2-6.The variability in glacier mass balance among the three models is

OGGM Model Validation and Forecasting
To prove the accuracy of the future changes in glacier area and volume, we used the OGGM model to acquire the glacier mass balance from 2000 to 2020, which is displayed in Figure 10.The OGGM model and the observed values show a near-identical trend, with a correlation coefficient of 0.87, which has a strong significance.In addition, the difference between the linear fit lines of OGGM simulated and the observed values have a tendency to widen; the error range of 52-124 mm w.e.a −1 . is acceptable.After confirming that the results of glacier area and volume were credible, we calculated the glacier mass balance based on the formula and combined it with the snow-ice density transformation to get the future glacier mass balance value of the URB.The results show that the glacier mass balance is the most negative in the SSP5.8-5pattern, followed by SSP2.4-5, and is the least negative in SSP1.2-6.The variability in glacier mass balance among the three models is present and in good agreement with future changes in area and volume, with an accelerated rate of change in 2035 and little fluctuation after 2075.

Glacier Loss Comparison in Typical Regions
For the entire Tianshan mountain range, the glacier mass deficit within the URB is relatively dramatic, with much higher values of negative mass balance than many typical basins, as well as the mass balance of the monitored glaciers.The factors influencing the ablation and mass balance of glaciers are sophisticated (Table 5).First, the degree and speed of glacier ablation are different when the size and type of glacier are different.The larger the size and the more concentrated the distribution of the glacier, the less intense the ablation.Large debris-covered glaciers are less prone to ablation, and such glaciers are most strongly developed in the Tomur region of the Tianshan Mountains.The Urumqi Valley belongs to the typical glacier in a small area and scattered distribution, where fast ablation is inevitable.Second, the combination of aspect and slope affects the development of glaciers mainly by changing the amount of solar radiation received.The mass balance is exacerbated by a negative balance in the N/N-direction, which is usually favored by small slopes for the accumulation of glaciers.The vast majority of glaciers in the Urumqi Valley are orientated N and have little topographic relief, and there are no high mountain systems around them that can block solar radiation.Third, the altitude zone in which the glacier is located is also a key factor affecting the ablation of glaciers, as glaciers are not widely developed at lower altitudes, and the ablation rate is generally faster in the region below 4500 m.The glacier studied in this paper belongs to this situation.Fourth, there is a change of water and heat because of the significant impact of westerly winds from the Atlantic Ocean.In this paper, the effects of temperature and precipitation are discussed in Section 5.4 as a separate section.

Glacier Loss Comparison in Typical Regions
For the entire Tianshan mountain range, the glacier mass deficit within the URB is relatively dramatic, with much higher values of negative mass balance than many typical basins, as well as the mass balance of the monitored glaciers.The factors influencing the ablation and mass balance of glaciers are sophisticated (Table 5).First, the degree and speed of glacier ablation are different when the size and type of glacier are different.The larger the size and the more concentrated the distribution of the glacier, the less intense the ablation.Large debris-covered glaciers are less prone to ablation, and such glaciers are most strongly developed in the Tomur region of the Tianshan Mountains.The Urumqi Valley belongs to the typical glacier in a small area and scattered distribution, where fast ablation is inevitable.Second, the combination of aspect and slope affects the development of glaciers mainly by changing the amount of solar radiation received.The mass balance is exacerbated by a negative balance in the N/N-direction, which is usually favored by small slopes for the accumulation of glaciers.The vast majority of glaciers in the Urumqi Valley are orientated N and have little topographic relief, and there are no high mountain systems around them that can block solar radiation.Third, the altitude zone in which the glacier is located is also a key factor affecting the ablation of glaciers, as glaciers are not widely developed at lower altitudes, and the ablation rate is generally faster in the region below 4500 m.The glacier studied in this paper belongs to this situation.Fourth, there is a change of water and heat because of the significant impact of westerly winds from the Atlantic Ocean.In this paper, the effects of temperature and precipitation are discussed in Section 5.4 as a separate section.

Contribution of Meteorological Conditions to Glacier Ablation
Climate is the basic factor in assessing glacier change.It is shown that temperature is the leading causes of the glacier change [54,55].The amount of glacial melt needs to increase by 40~50% of precipitation to compensate for every 1 • C increase in the mean summer temperature [56].Figure 11

Contribution of Meteorological Conditions to Glacier Ablation
Climate is the basic factor in assessing glacier change.It is shown that temperat the leading causes of the glacier change [54,55].The amount of glacial melt needs crease by 40~50% of precipitation to compensate for every 1 °C increase in the mean mer temperature [56].Figure 11   In order to further explore the dominant factors and approximate time lag of tem ature and precipitation on past glacier changes in the URB, we calculated the corre coefficients between glacier mass balance delayed by 1-6 years and climate factors, re tively (Figure 12).The highest correlation between summer temperatures and the m tude of glacier loss can be seen, which means that the higher the summer tempera the more glacier loss.In addition, because the annual precipitation accumulation In order to further explore the dominant factors and approximate time lag of temperature and precipitation on past glacier changes in the URB, we calculated the correlation coefficients between glacier mass balance delayed by 1-6 years and climate factors, respectively (Figure 12).The highest correlation between summer temperatures and the magnitude of glacier loss can be seen, which means that the higher the summer temperatures, the more glacier loss.In addition, because the annual precipitation accumulation in the region is mainly contributed by summer precipitation, the correlation coefficients obtained for annual and summer precipitation with glacier mass balance do not differ much.In terms of the delay period, there is the highest correlation coefficient between glacier mass balance with a three-year lag and summer temperature.The highest correlation coefficient between glacier mass balance and annual precipitation was found in the two-year lag, and the correlation coefficient between glacier mass balance and winter precipitation was negative at that time, which proved that winter precipitation not only did not cause the glacier loss, but contributed to the accumulation of the glacier, but only with a minor contribution.Therefore, the response of glaciers to climate in the URB has a lag of 2-3 years.region is mainly contributed by summer precipitation, the correlation coefficients obtained for annual and summer precipitation with glacier mass balance do not differ much.
In terms of the delay period, there is the highest correlation coefficient between glacier mass balance with a three-year lag and summer temperature.The highest correlation coefficient between glacier mass balance and annual precipitation was found in the two-year lag, and the correlation coefficient between glacier mass balance and winter precipitation was negative at that time, which proved that winter precipitation not only did not cause the glacier loss, but contributed to the accumulation of the glacier, but only with a minor contribution.Therefore, the response of glaciers to climate in the URB has a lag of 2-3 years.

Conclusions
This study aimed to quantify the glacier changes in the Urumqi Valley in the Tianshan Mountains, and the area changes of glaciers were assessed by multi-source remote sensing imagery and analyzed in conjunction with topographic factors.In addition, the mass balance of the glaciers in the basin was estimated using geodetic methods and validated with the aid of traditional glaciological methods, and the following main conclusions were obtained: (1) In 2022, 108 glaciers were mapped in the entire Urumqi Valley, a decrease of nearly one-third compared to 150 in 1964.The glacier area in the basin decayed rapidly from 48.68 ± 4.59 km 2 in 1964 to 21.61 ± 0.26 km 2 in 2022, with a shrinkage rate of 0.47 ± 0.04 km 2 •a −1 and an overall relative rate of change in glacier area of 55.6% (0.96%•a −1 ).Glacier area shrinkage has increased since 1990 and is higher than the average annual change rate (0.96%•a −1 ) over the past 58 years.The majority of the glaciers are orientated towards N (N\NE\NW) and are situated in the elevation zone between 3800-4000 m.The glacier area oriented N had the most ablation, accounting for 58% of the total area, and the largest percentage of glacier area (over 50%) was between 3900-4000 m.Different combinations of topographic factors are associated with differential glacier area changes

Conclusions
This study aimed to quantify the glacier changes in the Urumqi Valley in the Tianshan Mountains, and the area changes of glaciers were assessed by multi-source remote sensing imagery and analyzed in conjunction with topographic factors.In addition, the mass balance of the glaciers in the basin was estimated using geodetic methods and validated with the aid of traditional glaciological methods, and the following main conclusions were obtained: (1) In 2022, 108 glaciers were mapped in the entire Urumqi Valley, a decrease of nearly one-third compared to 150 in 1964.The glacier area in the basin decayed rapidly from 48.68 ± 4.59 km 2 in 1964 to 21.61 ± 0.26 km 2 in 2022, with a shrinkage rate of 0.47 ± 0.04 km 2 •a −1 and an overall relative rate of change in glacier area of 55.6% (0.96%•a −1 ).Glacier area shrinkage has increased since 1990 and is higher than the average annual change rate (0.96%•a −1 ) over the past 58 years.The majority of the glaciers are orientated towards N (N\NE\NW) and are situated in the elevation zone between 3800-4000 m.The glacier area oriented N had the most ablation, accounting for 58% of the total area, and the largest percentage of glacier area (over 50%) was between 3900-4000 m.Different combinations of topographic factors are associated with differential glacier area changes Of this, the average glacier mass balance was −0.67 ± 0.12 m w.e.a −1 before 2010 and was −0.63 ± 0.11 m w.e.a −1 after 2010.The average geodetic mass balance of the monitored UHG-1 (−0.69 ± 0.11 m w.e.a −1 ) somewhat deviates from the observed result (−0.65 m w.e.a −1 ), but the geodetic method result in this paper can be used to reflect the changes of glaciers in the region; (3) Based on the OGGM model simulations, SSP5.8-5 model has the fastest glacier area and volume losses and the rate of glacier area and volume loss in 2020-2100 undergoes a fast and then slow process.The fastest rate of glacier area loss occurs between 2030 and 2035; the rate of glacier area loss slows between 2042 and 2050; and the lowest rate of glacier area loss occurs between 2070 and 2080, when glacier area will no longer be decreasing.The highest and lowest inflection points for glacier volume loss are in 2035 and 2070, respectively.The glacier mass balance based on the OGGM model can correspond well with the observed values, so it confirms that the results of this paper's prediction of glacier changes in the Urumqi Valley are credible.Moreover, Different SSP scenarios show different changes in glacier mass balance, with the fastest negative mass balance in the SSP5.8-5model, followed by SSP2.4-5, and the slowest negative mass balance in SSP1.2-6.
(4) Although the glacier area change and mass balance are spatially heterogeneous, glacier melting in the basin has shown an accelerated shrinkage trend in recent years compared to other regions.Higher temperatures are the dominant factor contributing to accelerated glacier loss, and increased (summer) precipitation can also make glacier mass balance values more negative.Winter precipitation will contribute almost no amount to glacier accumulation.There is a lag of roughly 2-3 years in the response of glaciers to climate in the Urumqi Valley.
Remote Sens. 2024, 16, x FOR PEER REVIEW 3 of 20 −1.5 °C in January.The precipitation reaches 466 mm, and summer contributes the most precipitation.

Figure 1 .
Figure 1.Location of the Urumqi Valley and glacier distribution.(a) is a diagram of the entire Urumqi River basin; (b) is a diagram of all the glaciers involved in the study area.

Figure 1 .
Figure 1.Location of the Urumqi Valley and glacier distribution.(a) is a diagram of the entire Urumqi River basin; (b) is a diagram of all the glaciers involved in the study area.

Figure 2 .
Figure 2. Schematic diagram of flower pole deployment under traditional glaciological methods.

Figure 2 .
Figure 2. Schematic diagram of flower pole deployment under traditional glaciological methods.

Figure 3 .
Figure 3. Distribution and delineation of glaciers in different periods of the Urumqi River basin.

Figure 3 .
Figure 3. Distribution and delineation of glaciers in different periods of the Urumqi River basin.

Figure 4 .
Figure 4. Schematic illustration of the distribution and variation of glacier outlines over time.The left figure shows glacier orientation over time.The right figure shows the changes in glacier area at different altitudes.
), the elevation difference residuals in the nonglaciated areas can be used to evaluate the errors among the DEMs and to calculate the accuracy of the estimation results for the glacier volume change and the mass balance.Because the spatial resolution of the DEM data used is not uniform, the spatial autocorrelation distances were chosen to be 600 m, 250 m, and 50 m for the SRTM, ALOS, and Zi-yuanNo.3resolutions of 30 m, 12.5 m, and 2.5 m, respectively (Bolch T. et al., 2011).After

Figure 4 .
Figure 4. Schematic illustration of the distribution and variation of glacier outlines over time.The left figure shows glacier orientation over time.The right figure shows the changes in glacier area at different altitudes.

20 Figure 7 .
Figure 7. Box plots of annual height variations for multi-period DEMs.
5, and SSP5-8.5 are approximately the same, but the intermediate trends are very different.For glacier area, the SSP1-2.6 scenario has the slowest rate of glacier retreat, with an average annual decrease in change of 0.351 km 2 •a −1 and an average annual decrease in glacier volume/volume of 5. 21 × 10 6 m 3 •a −1 .Next is the SSP2-4.5 scenario, with an average annual decrease in change of 0.352 km 2 •a −1 and an average annual decrease in glacier volume/volume of 5. 23 × 10 6 m 3 •a −1 , and the SSP5-8.5 scenario, with a decrease in glacier area/volume of 0.352 km 2 a −1 and an average annual decrease in glacier volume/volume of 5.23 × 10 6 m 3 •a −1 .SSP5-8.5 scenario had the largest glacier area and reserve melt.The glacier area change decreases by 0.353 km 2 •a −1 and the glacier reserve/volume decreases by 5.29×10 6 m 3 a −1 .

Figure 7 .
Figure 7. Box plots of annual height variations for multi-period DEMs.

Figure 8 .
Figure 8.The area and volume change of glaciers in the Urumqi River Basin from 2020 to 2100.Where (a,b) shows the area and volume trends of the 13 climate models under different climate models, respectively; (c,d) presents the mean values of the simulation results of the 13 climate models after processing under the scenarios of SSP1-2.6,SSP2-4.5, and SSP5-8.5.The 13 model patterns described in the legend and whose simulation standard errors are displayed as shaded areas in (c,d).I, II, III represent the different stages of the trend change.

Figure 9 .
It is obvious that the period of maximum positive accumulation of mass balance is between 1960 and 1980.Since 2000, the positive values of glacier mass balance have almost disappeared and the negative accumulation trend of glacier mass balance has intensified, with the highest negative accumulation value in 2010.The annual average values of glacier mass balance of UG-1 for the periods of 2000-2010, 2010-2022, and 2000-2022 are 664 mm w.e.a −1 , 666 mm w.e.a −1 , and 662 mm w.e.a −1 , respectively.Overall, the annual mean mass balance for the period 2000-2022 is stable and shows little fluctuation; it is still suitable as a reference for the same period of time for obtaining the mass balance based on remote sensing.Compared with the annual average mass balance of a single glacier obtained by geodetic methods during the same period of this study, the values are-0.67 m w.e.a −1 , −0.71 m w.e.a −1 , and −0.69 m w.e.a −1 , respectively, and it can be concluded that the mass balances of geodetic and glaciological measurements are in agreement with each other.Taking the glaciological mass balance values as benchmarks, the relative differences between 2000-2010, 2010-2022, and 2000-2022 are only 1.5%, 7.5%,

Figure 8 .
Figure 8.The area and volume change of glaciers in the Urumqi River Basin from 2020 to 2100.Where (a,b) shows the area and volume trends of the 13 climate models under different climate models, respectively; (c,d) presents the mean values of the simulation results of the 13 climate models after processing under the scenarios of SSP1-2.6,SSP2-4.5, and SSP5-8.5.The 13 model patterns described in the legend and whose simulation standard errors are displayed as shaded areas in (c,d).I, II, III represent the different stages of the trend change.
. It is obvious that the period of maximum positive accumulation of mass balance is between 1960 and 1980.Since 2000, the positive values of glacier mass balance have almost disappeared and the negative accumulation trend of glacier mass balance has intensified, with the highest negative accumulation value in 2010.The annual average values of glacier mass balance of UG-1 for the periods of 2000-2010, 2010-2022, and 2000-2022 are 664 mm w.e.a −1 , 666 mm w.e.a −1 , and 662 mm w.e.a −1 , respectively.Overall, the annual mean mass balance for the period 2000-2022 is stable and shows little fluctuation; it is still suitable as a reference for the same period of time for obtaining the mass balance based on remote sensing.Compared with the annual average mass balance of a single glacier obtained by geodetic methods during the same period of this study, the values are −0.67 m w.e.a −1 , −0.71 m w.e.a −1 , and −0.69 m w.e.a −1 , respectively, and it can be concluded that the mass balances of geodetic and glaciological measurements are in agreement with each other.Taking the glaciological mass balance values as benchmarks, the relative differences between 2000-2010, 2010-2022, and 2000-2022 are only 1.5%, 7.5%, and 4.5%, respectively.Moreover, a linear fit to the mass balance trends obtained by traditional glaciological methods after 2000 showed a stronger negative accumulation trend.The interval of fitted values is from −0.83 m w.e.a −1 to −0.67 m w.e.a −1 and the results obtained in this study fall perfectly in that interval.

and 4 .
5%, respectively.Moreover, a linear fit to the mass balance trends obtained by traditional glaciological methods after 2000 showed a stronger negative accumulation trend.The interval of fitted values is from −0.83 m w.e.a −1 to −0.67 m w.e.a −1 and the results obtained in this study fall perfectly in that interval.

20 Figure 10 .
Figure 10.Validation and prediction of glacier mass balance for the OGGM model.(1) Comparison of OGGM-based and observed glacier mass balance for 2000-2020.(2) Future glacier mass balance values based on glacier area and volume from the OGGM model.

Figure 10 .
Figure 10.Validation and prediction of glacier mass balance for the OGGM model.(1) Comparison of OGGM-based and observed glacier mass balance for 2000-2020.(2) Future glacier mass balance values based on glacier area and volume from the OGGM model.
illustrates the temperatures and precipitation obtained from the Daxigou meteorological station for the years 1958-2022, showing that the Urumqi Valley has experienced warming and wetting potential during the last half-century.The average temperature from 1959 to 2022 is −4.75 • C in this region.The average temperature in 1959-2000 and 2000-2022 is 5.02 • C and 4.23 • C, which is an increase of 0.20 • C/10a and 0.07 • C/10a, respectively.The average annual precipitation from 1960 to 2021 is 473.80 mm in this region.The average precipitation in 1959-2000 and 2000-2022 is 444.64 mm and 528.13 mm, which has a increasing rate of 19.38 mm/10a and 29.95 mm/10a, respectively.
illustrates the temperatures and precipitation obt from the Daxigou meteorological station for the years 1958-2022, showing tha Urumqi Valley has experienced warming and wetting potential during the last hal tury.The average temperature from 1959 to 2022 is −4.75 °C in this region.The av temperature in 1959-2000 and 2000-2022 is 5.02 °C and 4.23 °C, which is an incre 0.20 °C/10a and 0.07 °C/10a, respectively.The average annual precipitation from 19 2021 is 473.80 mm in this region.The average precipitation in 1959-2000 and 2000-2 444.64 mm and 528.13 mm, which has a increasing rate of 19.38 mm/10a and 29.95 mm respectively.

Figure 11 .
Figure 11.Variations in annual mean temperature and precipitation observed at the Daxigou orological Station from 1959 to 2022.

Figure 11 .
Figure 11.Variations in annual mean temperature and precipitation observed at the Daxigou Meteorological Station from 1959 to 2022.

Figure 12 .
Figure 12.Correlation analysis of different lags of glacial mass balance with air temperature and precipitation.

Figure 12 .
Figure 12.Correlation analysis of different lags of glacial mass balance with air temperature and precipitation.

( 2 )
The glacier elevation in the Urumqi Valley declined by 44.64 ± 0.69 m during 2000-2022, with an average annual decline of 2.31 ± 0.05 m•a −1 in 2000-2010 and 1.80 ± 0.01 m•a −1 in 2010-2022.The volume of glaciers in the Urumqi Valley has decreased by 0.24 km 3 over the past 22 years.The average annual glacier mass balance of URB is −0.65 ± 0.11 m w.e.a −1 during 2000-2022.

Table 1 .
Attributes and applications of the data used in the past glacier change.

Table 1 .
Attributes and applications of the data used in the past glacier change.

Table 2 .
Glacier delineation and uncertainty in each sub-basin of the Urumqi River.

Table 3 .
Metrics of glacier change and uncertainty in different periods.

Table 4 .
Original and adjusted errors between muti-DEMs.

Table 4 .
Original and adjusted errors between muti-DEMs.

Table 5 .
Comparison of glacier mass balance of monitored glaciers and typical basins in western China.

Table 5 .
Comparison of glacier mass balance of monitored glaciers and typical basins in western China.