MODIS-observed spatiotemporal changes in surface albedo of Karakoram glaciers during 2000-2018

spatiotemporal in he role of albedo is very important in modulating the surface energy balance of glaciers. The main objective of this study is to assess the spatiotemporal variability in surface albedo of the Karakoram glaciers in Pakistan during the summer seasons (June, July and August) for the period from 2000-2018. We used Moderate Resolution Imaging Spectroradiometer (MODIS) data to estimate the amount of glacier surface albedo. We combined the MODIS Terra- and Aqua-derived albedo products to reduce the amount of cloud influence and to improve the estimation of glacier surface albedo. Our results indicate that the average annual decrease in albedo is ~0.041% during the summer. The decrease in albedo was relatively high during recent years, with an annual rate of decrease of ~0.45%. The decreasing trend in albedo is towards the north-western part of the Karakoram mountain range. Climate change is the potential cause of albedo variations in the study area. Albedo has a strong negative correlation with temperature (r = -0.811) and a strong positive correlation with precipitation (r = 0.809). The present study concludes that the trend in decreasing albedo is higher during the recent years than the last decade and climate change is playing a vital role in it. Digital Elevation Model (DEM) data. The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) onboard the National Aeronautics and Space Administration (NASA) spacecraft Terra was released in mid-October 2011, covering a land surface between 83°N to 83°S, is capable of collecting in-track stereo images using nadir-and aft-looking near-infrared cameras. Version 2 of the ASTER Global Digital Elevation Model (GDEM) is comprised of 22,702 1° x 1° tiles with an overall accuracy of ~17 m at a confidence level of 95%. In this study Karakoram was divided into 6 elevation zones and later compared it with the MODIS albedo product.

with meteorological data to explain their relationship and it was the first attempt to its nature monitoring the albedo of whole glacier throughout the ablation season. A research about the glacial albedo change was proposed by Fugazza, et al. [20] using Landsat data available from 1984 to 2011. The study was conducted on the fifteen selected glaciers in Central Alps region to determine the magnitude of change in glaciers albedo. The study confirmed that there was decrease in the albedo on 14 glaciers out of all 15 glaciers significant at 95% confidence level. The decrease in the rate of albedo ranged between 0.001 and 0.006 per year with an average decrease rate of 0.003 per year.
A research was conducted at Johnsons Glacier to investigate the albedo trend and seasonality by using MODIS daily snow albedo product (MOD10A1). MODIS data from the month of December 2006 to February 2015 was used in the study for all sky conditions. The overall amount of albedo increase was observed 2% for in-situ measurements and 6% for MODIS-derived albedo measurement during the period 2006-2015 [21]. The spatial and temporal changes over snow surface area can be determined with the help of accurate measurements of spectral snow albedo obtained from the satellite data. To measure the spectral snow albedo, a study was conducted in the northern areas of Pakistan by using Landsat TM and ETM+ spatiotemporal imagery datasets. The results showed a decreasing albedo amount of about 0.058 in the recent year (2000) as compared to the older surface albedo data (1992) [22].
There are some primary and secondary objectives of the research conducted to measure surface albedo variability of the Karakoram region. The primary goal of this study is to understand the behavior of spatial and temporal variation in the glacial albedo and its response to climate change in the Karakoram region. In recent times, there have been a lot of researches on the albedo change on different glaciers around the world. It has been noticed since the past few decades that the variation in the albedo have caused a huge impact on the global climate change. The main objectives of the present study are:  To assess the spatial and temporal variability of surface albedo on the glaciers in the Karakoram region of Pakistan using MODIS-derived (MOD10A1) data  To analyze the surface albedo variation on every single glacier of each year from 2000-2018 during the summer season (June, July, August)  Comparing the surface albedo variation with elevation using ASTER digital elevation model (DEM) data by dividing study area in six elevation zones  Discussing the factors playing an important role in controlling surface albedo variation in the study area Material and Methods. Investigation site. The Karakoram range of mountains is located at approximately 36°N and 74°E, spanning parts of Pakistan (Gilgit-Baltistan), China (Xinjiang), India (Ladakh) and containing the largest concentration of glaciers outside the polar regions [23]. This system of mountains extending approximately 500 km in length between Central and South Asia is a complex range of mountains that includes the Himalayas to the southeast, the Kunlun Mountains to the northeast, the Pamir Mountains to the northwest and the Hindu Kush to the west.  Note: CenLon and CenLat are the longitude and latitude (in degrees), respectively, of a single point representing the location of the glacier; the area (Area) of the glacier is in km 2 , calculated in Cartesian coordinates on a cylindrical equal-area projection of the authalic sphere of the WGS84 ellipsoid, or, for nominal glaciers, accepted from the source inventory; Zmin and Zmax are the minimum and maximum elevations (meters above sea level) of the glacier, obtained in most cases directly from a DEM covering the glacier; Zmed is the median elevation (m) of the glacier, chosen by sorting the elevations of the DEM cells covering the glacier and recording the 50th percentile of their cumulative frequency distribution; Slope is the mean slope of the glacier surface (deg), obtained by averaging single-cell slopes from the DEM; and Lmax is the length (m) of the longest surface flowline of the glacier. The length is measured according to the algorithm proposed by Machguth and Huss [25].
In this study, we analyzed the part of Karakoram located in Pakistan ( Figure 1; Table  1). Siachen Glacier is the largest glacier in the region, with an area of approximately 1078 km 2 , and is the largest glacier outside the Arctic region. The Karakoram exhibits heavy glaciation because of the great elevation, especially on the more humid southern slopes. The inventory suggests that the total number of glaciers in the Karakoram area may exceed 7000, which consists of approximately half of the glacierized area surveyed on earth. The Indus River in Pakistan and the Yarkand River in Xinjiang Province, China, are fed by the glaciers of the Karakoram as a watershed, variously estimated at between 18,000 and 20,000 km 2 [26]. The Karakoram glacial region is the most glaciated area outside the polar region consisting of some longest glaciers in the world (i.e. Siachen glacier 76km -World second longest glacier outside the polar region, Biafo glacier 67km -third longest glacier in the World outside the polar region) [27]. MODIS Daily Snow Product (MOD10A1/MYD10A1). MOD10A1 is a daily MODIS (Terra-derived) snow product containing NDSI snow cover and daily albedo products from 1999 to the present. The National Snow and Ice Data Center (NSIDC) processes and distributes the MODIS (MOD10A1) Snow Cover and Snow Albedo Daily L3 Global 500 m SIN Grid, which is available at https://nsidc.org/data/mod10a1 [28]. The MOD10A1 product provides an estimation of daily blue-sky albedo, corresponding to broadband albedo at a resolution of 500 m [29]. This albedo product is stored in the form of integer values ranging between 0 to 100 (percent) and the accompanying the quality control flags [30]. MYD10A1 is daily MODIS (Aqua-derived) snow product containing NDSI snow cover and daily snow albedo products from 2002 to the present. The Earth Observation System satellite aqua is also known as EOS PM-1, which passes in the afternoon from south to north over the equator completing an orbit around the Earth every 1-2 days. Climate data. The study area lies in the Gilgit-Baltistan territory of Pakistan situated in the northern part of Pakistan. The weather station data of four weather stations (Astore, Gilgit, Hunza and Skardu) in Gilgit-Baltistan were obtained from the Pakistan Meteorological Department (PMD) ( Table 2). The weather station data were later corrected with the remote sensing data obtained from MODIS to generate a relationship between the ground-based weather station (meteorological) data and remote sensing data. MOD10A1 and MYD10A1 are tiles of daily level-3 global snow cover product at a spatial resolution of 500m selected from a best observation obtained from the multiple observations of MODIS MOD10_L2G and MYD10_L2G product. M (O, Y) D10A1 contains the best observation of the day for NDSI daily snow cover product and snow albedo product. The study area is located in the MODIS tile h24v05, only the MODIS daily snow albedo was extracted from the M (O, Y) D10A1 product for further analysis. The study was conducted on 27 glaciers in the study area. The research area was analysed as a whole as well the analysis was performed on every single glacier in the Karakoram glacier region of Pakistan. Only the summer months (June, July, August) data was used in the research to analyse the spatial and temporal variation of glacial albedo in the study area. MODIS Reprojection Tool (MRT), MATLAB and ArcGIS version 10.8 for spatiotemporal analysis of albedo in the study region and SPSS and Origin software were used to perform statistical analysis.

Cloud Elimination
A simple technique to eliminate cloud cover from a dataset is by combining MODIS Terra-and Aqua-derived products. Xie, et al. [31] generated the combined product of MOD10A1 and MYD10A1 to eliminate the cloud effect and to improve the quality of the results by decreasing the amount of cloud cover by ~10% compared to the MODIS Terra/Aqua products alone.  29.88% clouds at the Baltoro Glacier, and the data obtained from the MODIS Aqua satellite (MYD10A1) contained 24.61% clouds at the Baltoro Glacier. When we combined both the Aqua and Terra products, the cloud cover was reduced to 13.29%. By using this simple technique, approximately 55.50% of the clouds were removed. The number of pixels containing snow was also improved by combining the MOD10A1 and MYD10A1 images. Similarly, on August 9, 2008, the data obtained from Terra contained 22.60% clouds, and data obtained from Aqua contained 29.30% cloud cover. The cloud cover was reduced to 14.25% by combining the Aqua-and Terra-derived products (Table 3). MODIS Terra and Aqua albedo product The snow area was 72.87% in the MODIS Terra product and 75.34% in the MODIS Aqua product before combining the MODIS Terra and Aqua products. The quality of the snow area improved to 81.98% by combining both products. In total, 10.07% and 6.67% of the areas in the MODIS Terra and Aqua products, respectively, contained clouds, which improved to just 2.41% by combining both products (Figure 3). Results and discussion. Glacier surface albedo variations. To analyse the variation in the glacier surface albedo, we calculated the means of glacial albedo over the Karakoram mountain range during the summer seasons and observed the variation trends over 2000-2018 ( Figure 4). The average albedo during June, July and August (the summer months) was calculated from 2000 to 2018. Least squares regression was used to estimate the trends to determine the best linear fit for the data. There was a very slight decreasing trend at an average rate of 0.41% per decade during the summer seasons of 2000 to 2018 ( Figure 5).    Trend of the glacier surface albedo variation. The trend in the average albedo from 2000 to 2018 tended to decrease as we move from eastern part towards the north-western part of the Karakoram in Pakistan (Figure 8), as the study of Rankl, Kienholz and Braun [23] revealed that the glaciers are retreating towards the north-western part of the Karakoram. This decrease in albedo is due to the decrease in altitude towards the Hindu Kush region and is influenced by the reduced precipitation in the northern part of Karakoram compared to the central part of the Karakoram [32]. The highest albedo in the study area was observed on the Rimo Glacier (53.75%), and the trend decreased towards the Hindu Kush region. The lowest average albedo was observed at the Minapin Glacier (40.96%), which is the last glacier in the study area moving towards the Hindu Kush mountain region.   . The amount of albedo increases with increasing elevation and as the temperature decreases, causing less glacial melt, which alternately increases the albedo moving towards areas with higher elevations. The variations in albedo is very closely related to meteorological factors (i.e., precipitation and temperature) during the summer season. The weather station data were obtained for the Gilgit-Baltistan territory and then correlated with the MODIS-observed albedo data. Pearson correlation coefficient were used to quantify the correlations between the MODIS albedo data and the ground-based weather station data (Table 4). Table 4. Pearson correlation coefficients between albedo and the meteorological factors Temperature Precipitation Albedo Pearson Correlation -.811 ** .809 ** Sig. (2-tailed) .002 .003 **. Correlation is significant at the 0.01 level (2-tailed).
A strong negative correlation was found between the temperature and MODIS albedo product. The relationship shows that the MODIS-observed albedo value during the summer season increases with decreasing temperature. The value of Pearson's r is -0.811, which indicates that the relationship between albedo and temperature at the Karakoram glaciers in the summer season is very strong and very significant at the 0.01 level (Figure 4). Under warmer conditions, the rate of metamorphism of snow grains ultimately increases, and due to snow melting, the rate of albedo retrieval decreases. Similarly, albedo is higher during cold climate conditions. The melting of snow grains leaves impurities behind, which absorb solar radiation, causing a decrease in albedo.

Figure 12.
Correlation of surface albedo with temperature and precipitation. Albedo have a significant negative and significant positive correlation with temperature and precipitation respectively. The relationship between albedo and precipitation was found to be very strong in the positive direction. The correlation shows that the value of albedo increases with increasing precipitation. The value of Pearson's r is 0.809, exhibiting a strong and significant correlation at the 0.01 level (Figure 12). The precipitation at the high altitudes in the Karakoram region is in the form of snowfall during the winter as well as the summer, which helps to buffer the glaciers against warming [33]. Gul, et al. [34] observed that the retreating glaciers in the Karakoram mountains in the northern part of Pakistan have a shown a significant negative correlation with the changing temperature in the region. The correlations between albedo and temperature and between albedo and precipitation during the summer season are expressed in Table 4.

Discussion.
The albedo value decreases during the summer season in the Karakoram glacial, which is situated in the Gilgit-Baltistan territory of Pakistan. There may be many reasons for this decrease in albedo. Global warming can be the main cause. According to the research conducted by Kapnick, et al. [35], global warming is not particularly effective in that region because of the unique seasonal cycle dominated by non-monsoonal precipitation, which protects it from the glacial reduction. Thus, our findings indicate that the large amount of precipitation in the study area shows a greater albedo response in the study area.
Bolch, et al. [36] studied that the behaviour in the Karakoram region is different from the other regions, as there is stability or gain in the mass compared to the Himalayas, where there is a rate of decrease in the glaciers similar to that in other parts of the world. Gardelle, Berthier and Arnaud conducted a study based on two different digital elevation models to calculate the mass balance of central Karakoram between 1999 and 2008 and found that there was a slight gain in the glacier mass balance in the early twenty-first century.
Climatic conditions are the potential sources affecting albedo retrieval. According to recent research conducted by Kumar, et al. [37], the rise in temperature and decline in the precipitation had a massive impact on the glaciers in the Karakoram, resulting in the form of mass loss supporting our results of albedo decline. The mass loss was very low during 2000-2009; our results are quite similar in that the change in albedo retrieval (decline) was nominal during 2000-2009 compared to recent years. The trend in the darkening surface is due to the reduction in snow from enhanced glacier melting, which is likely reinforced by climatic factors, i.e., increasing temperature and decreasing precipitation [38]. The decrease in albedo appears to be the consequences of the changes in the climate, including reduced precipitation and an increase in the air temperature of glaciers in the Ortles-Cevedale group, Central Italian Alps [20] and Dongkemadi Glacier in China [18].

Conclusion.
The surface energy and mass balance of glaciers strongly depends on the amount of solar radiation absorbed at the surface, which is mainly controlled by glacier surface albedo. The present research was conducted using MODIS-derived daily snow albedo data to observe the spatial and temporal variations of surface albedo of glaciers in the Karakoram region of Pakistan. The research was focused on the spatiotemporal variations during the summer season (June, July and August) from year 2000 to 2018. After reprojection and resampling of data, MODIS aqua satellite data and MODIS terra satellite data was combined to eliminate the clouds effecting the surface albedo of glaciers. Study time period was divided into two spans i.e. Past decade (2000-2009) and recent years (2010-2018). The main purpose of this division was to elaborate the glacier change in recent years is quite significant than the past decade. Ground-based meteorological weather station data was collected from Pakistan Meteorological Department (PMD). Climate data was correlated with the satellite data to find out the relationship of glacier surface albedo with temperature and precipitation. The impact of climate change and LAI on glaciers darkening is discussed in details. Here are the main findings of the study:  .  The study area was divided into six elevation zones (2000-3000 m, 3000-4000 m, 4000-5000 m, 5000-6000 m, 6000-7000 m, and 7000-8000 m) and the albedo change in each zone was discussed separately. The lowest amount of albedo was observed 6.67% at lowest elevation zone (2000-3000m) and highest amount of average albedo was observed 63.22% at highest elevation zone (7000-8000m). The increase in the amount of albedo with increasing elevation is due to the low temperature and high precipitation at higher elevations.  The direction of change in albedo was observed decreasing towards north-western direction towards Hindu-Kush mountain range. This decrease in the amount of albedo towards NW direction is because of decreasing altitude following low precipitation. Highest amount of albedo was observed at Rimo glacier which is at highest elevation among all the glaciers in the study region. While the lowest albedo was observed at Baultar glacier and Minapin glacier are at lowest elevation among all the glaciers in Karakoram study region.  The albedo change in the study region is closely related to the climatic factors (precipitation and temperature). A strong positive correlation was observed between precipitation and surface albedo at 99% confidence level. While a strong negative correlation was observed between temperature and surface albedo at 99% confidence level.

Recommendations.
According to our findings, the decrease in albedo was very slight in the first decade, but the rate of decrease was higher during recent years throughout the whole Karakoram glacial region of Pakistan. The overall trend of albedo decrease was not very significant in Karakoram glacial region of Pakistan. As per our study based on MODIS data, there was a slight decrease or no change (for some glaciers) in the Karakoram region in Pakistan during the first decade of the 21st century, but later, the rate of the decrease in glacier albedo is higher as the amount of albedo observed is decreasing relatively fast during recent years. Although the overall average amount of decrease in albedo is very small, but it is not negligible. However, more extensive studies using high-resolution satellite data as well as ground-based station data considering the anthropogenic activities causing albedo changes are required.