[[[78.0, 27.0], [92.0, 27.0], [92.0, 35.0], [78.0, 35.0], [78.0, 27.0]]]

• Khumbu Glacier (Everest): [[[86.7, 27.9], [87.1, 27.9], [87.1, 28.15], [86.7, 28.15], [86.7, 27.9]]]
• Gangotri Glacier: [[[78.9, 30.7], [79.3, 30.7], [79.3, 31.0], [78.9, 31.0], [78.9, 30.7]]]
• Siachen Glacier: [[[76.7, 35.3], [77.3, 35.3], [77.3, 35.8], [76.7, 35.8], [76.7, 35.3]]]

The year 2016 stands out with a dramatic [38.3% drop in snow cover](calculated from MODIS annual means: (7.88-4.86)/7.88×100) coinciding with one of the strongest El Niño events on record, which suppressed winter precipitation across the Himalayas. Similarly, 2024 shows another significant decline to [4.94](MODIS snow cover index), representing one of the lowest values in the decade-long record. The [R² value of 0.072](linear regression statistics) indicates high inter-annual variability—a characteristic of mountain climate systems—but the overall trajectory remains unambiguously downward.

• 2015 Glacier Area: [449.87 km²](Landsat 8 NDSI analysis, 30m resolution, Khumbu region 86.7°E-87.1°E, 27.9°N-28.15°N)
• 2024 Glacier Area: [351.00 km²](Landsat 9 NDSI analysis, 30m resolution, same region)
• Area Lost: [98.87 km²](difference calculation: 449.87 - 351.00)
• Percentage Retreat: [21.98%](calculated as change/baseline × 100) This retreat rate of nearly 22% over a decade translates to approximately [9.89 km² of ice loss per year](annual rate: 98.87 km² / 10 years)—an area equivalent to roughly 1,400 soccer fields of ice surface disappearing annually from this single glacier system. The imagery analysis reveals that the terminus has retreated significantly, while the lower ablation zone shows increasing exposed rock and debris cover. The false color composite imagery (SWIR-NIR-Red bands 6,5,4) enhances the visualization of snow and ice, with glacier surfaces appearing as bright cyan/blue against the brown rock exposures. The change detection algorithm identifies pixels that transitioned from glacier (NDSI > 0.4 in 2015) to bare ground (NDSI < 0.4 in 2024), painting a stark visual picture of the retreat.
  
  Figure 1: Khumbu Glacier region, 2015 baseline. True color composite (Landsat 8 Bands 4,3,2) showing extensive snow and ice coverage across the Everest region. Note the continuous white/bright surfaces indicating healthy glacier coverage.
  
  Figure 2: Khumbu Glacier region, 2024 current state. True color composite (Landsat 9 Bands 4,3,2) showing visibly reduced ice coverage, increased rock exposure, and terminus retreat compared to the 2015 baseline.
  
  Figure 3: Change detection map showing glacier retreat. Red pixels indicate areas where ice was present in 2015 but absent in 2024 (glacier loss), gray indicates no change, and green (rare) would indicate glacier gain. The predominance of red, particularly at lower elevations and glacier margins, confirms substantial retreat. The methodology employed for this analysis uses the Normalized Difference Snow Index (NDSI), a well-established remote sensing technique:

# NDSI calculation for Landsat 8/9

# NDSI = (Green - SWIR) / (Green + SWIR)

# For Landsat: NDSI = (SR_B3 - SR_B6) / (SR_B3 + SR_B6)
def calculate_ndsi(image):
    green = image.select('SR_B3')  # Green band
    swir = image.select('SR_B6')   # SWIR1 band
    ndsi = green.subtract(swir).divide(green.add(swir)).rename('NDSI')
    return ndsi

# Threshold application: NDSI > 0.4 = glacier/snow
glacier_mask = ndsi_image.gt(0.4)

# Area calculation using pixel area summation
glacier_area = glacier_mask.multiply(ee.Image.pixelArea())
                          .reduceRegion(reducer=ee.Reducer.sum(),
                                       geometry=khumbu_region,
                                       scale=30)

• 2015 Glacier Area: [616.47 km²](Landsat 8 NDSI analysis, 30m resolution, Gangotri region 78.9°E-79.3°E, 30.7°N-31.0°N)
• 2024 Glacier Area: [527.46 km²](Landsat 9 NDSI analysis, 30m resolution, same region)
• Area Lost: [89.01 km²](difference calculation)
• Percentage Retreat: [14.44%](calculated as change/baseline × 100) The Gangotri's retreat, while percentage-wise lower than Khumbu, represents a significant absolute ice loss given the glacier's larger initial extent. The [8.9 km² annual loss rate](annual rate calculation) translates directly into altered seasonal flow regimes for the Bhagirathi River and, subsequently, the Ganges.
  
  Figure 4: Side-by-side comparison of Gangotri Glacier between 2015 and 2024. False color enhancement (SWIR-NIR-Red) reveals the reduction in bright cyan glacier areas, with increased brown/tan rock exposure particularly visible along glacier margins and at lower elevations.
  
  Figure 5: Gangotri Glacier 2015, false color composite highlighting snow/ice coverage. The extensive cyan coloration indicates healthy glacier extent with continuous ice cover across the catchment.
  
  Figure 6: Gangotri Glacier 2024, false color composite. Reduced cyan/blue areas indicate glacier retreat, with more rock exposure visible, particularly at glacier margins and in areas previously covered by seasonal snow. Historical records from the Geological Survey of India indicate that Gangotri has been retreating since at least the mid-19th century, but the rate has accelerated dramatically since the 1970s. Between 1971 and 2004, the glacier retreated approximately 22 meters per year. The current analysis suggests this trend continues unabated.

• 2015 Glacier Area: [2,380.57 km²](Landsat 8 NDSI analysis, 30m resolution, Siachen region 76.7°E-77.3°E, 35.3°N-35.8°N)
• 2024 Glacier Area: [1,908.26 km²](Landsat 9 NDSI analysis, 30m resolution, same region)
• Area Lost: [472.31 km²](difference calculation)
• Percentage Retreat: [19.84%](calculated as change/baseline × 100) The Siachen's retreat of nearly 20% represents the largest absolute ice loss among the three focus glaciers—[472.31 km² of ice surface vanished over the decade](Landsat change analysis). This is an area larger than the city of Vienna. The annual loss rate of [47.23 km² per year](annual calculation) poses serious concerns for the Nubra and Shyok river systems that feed into the Indus.
  
  Figure 7: Siachen Glacier 2015, true color composite showing the extensive ice mass of the world's second-longest non-polar glacier. The continuous white coverage indicates the glacier's robust extent at the beginning of the analysis period.
  
  Figure 8: Siachen Glacier 2024, true color composite revealing visible reduction in ice coverage. The emergence of rock outcrops and the narrowing of glacier tongues are clearly visible when compared to the 2015 baseline.

• 2015 Regional Estimate: [33,347.66 km²](Landsat NDSI regional analysis, 2000m resolution, NDSI > 0.4)
• 2024 Regional Estimate: [15,041.93 km²](Landsat NDSI regional analysis, same methodology)
• Change: [-18,305.73 km²](difference calculation)
• Percentage Change: [-54.89%](calculated as change/baseline × 100) Critical Caveat: This regional figure of 54.89% reduction includes both permanent glacier ice and seasonal snow cover. The true glacier loss is better represented by the focus glacier statistics (14-22%), which use consistent annual composites to minimize seasonal snow variability. However, the reduction in seasonal snow itself is significant, as it affects water storage in the form of snowpack and influences timing of spring meltwater release.
  
  Figure 9: Decadal trend in regional snow cover index (2015-2024). The scatter plot with trend line demonstrates the overall declining trajectory despite significant inter-annual variability. Note the particularly low values in 2016 (El Niño year) and 2024.
  
  Figure 10: Regional snow cover map, 2015. White/bright areas indicate high snow cover (NDSI values approaching 1), while brown areas indicate no snow. Note the extensive coverage across the high-altitude spine of the Himalaya.
  
  Figure 11: Regional snow cover map, 2024. Comparison with the 2015 map reveals reduced white/bright areas, indicating diminished snow coverage across the region.
  
  Figure 12: Snow cover change detection map (2024 minus 2015). Blue areas indicate increased snow cover, red areas indicate decreased coverage, and yellow/white indicates no change. The predominance of red tones, particularly at lower elevations and southern exposures, confirms the overall decline.

Figure 13: Winter snow coverage 2015 (December-February composite). Winter represents the accumulation season when glaciers gain mass through snowfall. Extensive coverage indicates healthy accumulation patterns.

Figure 14: Winter snow coverage 2024 (December-February composite). While winter coverage remains substantial, the overall extent appears reduced compared to 2015, suggesting decreased accumulation-season snowfall. Summer Snow Cover (June-August):

Figure 15: Summer snow coverage 2015 (June-August composite). Summer represents the ablation season when glaciers lose mass. Snow remaining during summer indicates permanent ice and late-melting seasonal snow.

Figure 16: Summer snow coverage 2024 (June-August composite). The comparison reveals reduced persistent snow/ice coverage during the melt season, consistent with accelerated ablation. High-Altitude Analysis (>5000m):

The analysis specifically examined high-altitude zones where permanent glaciers predominantly exist:

Figure 17: Snow coverage at elevations above 5000m, 2015. These high-altitude zones should maintain permanent snow and glacier coverage year-round.

Figure 18: Snow coverage at elevations above 5000m, 2024. Even at these extreme altitudes where temperatures traditionally remain below freezing year-round, visible reduction in coverage is apparent.

• 2015 Mean LST: [16.04°C](MODIS MOD11A2, annual regional mean)
• 2024 Mean LST: [14.27°C](MODIS MOD11A2, annual regional mean)
• Total Change: [-1.77°C](difference calculation)
• Trend Slope: [-0.141°C per year](linear regression analysis)
• R² Value: [0.158](goodness of fit) Interestingly, the land surface temperature data shows a slight cooling trend over the decade. This apparent paradox—cooling temperatures accompanying glacier retreat—requires careful interpretation:

1. Elevation bias: LST measurements are area-weighted, and the loss of low-altitude snow (which has higher reflectivity) can actually reduce apparent surface temperature in thermal imagery
2. Albedo feedback: As glaciers retreat, they expose darker rock surfaces that absorb more solar radiation, but the reduced ice area means the regional average may still appear cooler
3. Atmospheric circulation changes: Regional climate patterns in the Himalayas are influenced by the monsoon system, and variations in cloud cover and precipitation can affect LST independently of global warming trends
   
   Figure 19: Land Surface Temperature trend (2015-2024). While the trend appears slightly negative, the high inter-annual variability (R² = 0.158) indicates complex thermal dynamics.
   
   Figure 20: Land Surface Temperature map, 2015. Blue tones indicate cooler surfaces (high altitude, snow-covered), while red/orange indicates warmer surfaces (lower elevations, bare ground).
   
   Figure 21: Land Surface Temperature map, 2024. The thermal pattern shows the elevation-dependent temperature distribution characteristic of mountain environments.
   
   Figure 22: LST change detection map (2024 minus 2015). Blue indicates cooling, red indicates warming. The complex pattern reflects local variations in land cover change and atmospheric conditions.

1. Debris-covered glaciers: Many Himalayan glaciers are heavily debris-covered at lower elevations. This insulating debris layer can slow melt rates even as temperatures rise, creating localized cooling effects as the dark debris absorbs heat
2. Monsoon influence: The South Asian monsoon delivers the majority of precipitation to the Himalayas. Changes in monsoon patterns—timing, intensity, and duration—can overwhelm temperature signals in determining glacier mass balance
3. Elevation-dependent warming: Climate models predict enhanced warming at higher elevations (elevation-dependent warming), which may not be captured in regional LST averages dominated by lower-elevation pixels The critical insight is that glacier retreat depends not solely on temperature but on the balance between accumulation (winter snowfall) and ablation (summer melt). The dramatic snow cover declines documented in this analysis suggest that reduced accumulation is a major driver of retreat, potentially even more significant than increased melt.

The comprehensive dashboard brings together all key indicators in a unified view:

Figure 23: Integrated analysis dashboard showing four key metrics: (top-left) Snow cover trend with declining trajectory, (top-right) LST trend with slight cooling, (bottom-left) Glacier area comparison showing dramatic reduction, (bottom-right) Year-over-year snow cover changes with green bars indicating gains and red bars indicating losses. The dashboard provides a holistic view of Himalayan cryospheric health. The year-over-year analysis reveals several notable patterns:

• 2016: Dramatic [38.3%](year-over-year calculation) decline coinciding with strong El Niño
• 2019: Strong [42.2%](year-over-year calculation) recovery, highest snow cover of the decade
• 2024: Second-largest decline at [33.5%](year-over-year calculation) below 2023 levels This variability underscores the importance of long-term trend analysis rather than single-year comparisons.
  
  Figure 24: Normalized heatmap of climate indicators across the decade. Green tones indicate conditions favorable for glaciers (higher snow, lower temperature), while red indicates unfavorable conditions. The pattern reveals the temporal evolution of climate stress on the Himalayan cryosphere.
  
  Figure 25: Bar chart comparing glacier areas between 2015 and 2024 across the three focus glaciers. All three show substantial reductions, with Siachen showing the largest absolute loss and Khumbu showing the highest percentage decline.

# Complete NDSI-based glacier mapping workflow
import ee

# Define study region
khumbu_region = ee.Geometry.Rectangle([86.7, 27.9, 87.1, 28.15])

# Load and filter Landsat collection
landsat_2015 = (ee.ImageCollection('LANDSAT/LC08/C02/T1_L2')
    .filterBounds(khumbu_region)
    .filterDate('2015-01-01', '2015-12-31')
    .filter(ee.Filter.lt('CLOUD_COVER', 20)))

# Create median composite to minimize cloud effects
composite_2015 = landsat_2015.median()

# Calculate NDSI
green = composite_2015.select('SR_B3')
swir = composite_2015.select('SR_B6')
ndsi = green.subtract(swir).divide(green.add(swir))

# Apply threshold for glacier classification
glacier_mask = ndsi.gt(0.4)

# Calculate glacier area in square kilometers
area = glacier_mask.multiply(ee.Image.pixelArea()).divide(1e6)
glacier_area_km2 = area.reduceRegion(
    reducer=ee.Reducer.sum(),
    geometry=khumbu_region,
    scale=30,
    maxPixels=1e13
).get('NDSI').getInfo()

1. Cloud filtering: Only scenes with less than 20-30% cloud cover were included
2. Median compositing: Annual median composites reduce the impact of remaining cloud contamination
3. Multi-sensor integration: Landsat 8 (2013-present) and Landsat 9 (2021-present) provide consistent spectral measurements
4. Scale-appropriate analysis: 30m for focus glaciers, 2000m for regional estimates to balance precision and computational efficiency

• The IPCC Special Report on the Ocean and Cryosphere documented accelerating glacier loss in High Mountain Asia
• Bolch et al. (2012) reported average mass loss rates of 0.21 m water equivalent per year for the Himalayas
• The ICIMOD Hindu Kush Himalaya Assessment projected one-third glacier loss even under 1.5°C warming scenarios The 14-22% area loss documented here over 2015-2024 is consistent with these findings, representing an acceleration of previously documented trends.

The SRTM Digital Elevation Model at 30-meter resolution provides critical topographic context for understanding glacier distribution and retreat patterns.

Figure 26: Digital Elevation Model of the Central Himalaya study region. Brighter tones indicate higher elevations, revealing the dramatic topographic relief of the world's highest mountain range. Glaciers predominantly occupy elevations above 4500m.

Figure 27: High-resolution DEM of the Khumbu region, showing the extreme topography around Mount Everest. The elevation range from valley floors at approximately 4000m to summit peaks exceeding 8000m creates the conditions for glacier formation and persistence.

Figure 28: Hillshade visualization of Khumbu terrain, enhancing the visualization of valleys, ridges, and glacier-carved landscapes. The U-shaped valleys characteristic of glacial erosion are clearly visible. Glaciers in the Himalaya occupy a specific elevation band—typically between [4500m and 6500m](elevation analysis), with the Equilibrium Line Altitude (ELA) where accumulation equals ablation typically around [5200-5500m](estimated from regional studies). As climate warms, the ELA rises, reducing the accumulation zone and accelerating retreat.

The power of satellite remote sensing lies in its ability to provide objective, repeatable measurements across time. The comparison images provide compelling visual evidence of change:

Figure 29: Direct comparison of Khumbu Glacier between 2015 (left) and 2024 (right). The panels allow visual assessment of changes in ice extent, terminus position, and overall glacier health.

Figure 30: Glacier extent comparison across all three focus regions, showing the spatial distribution of ice loss between the baseline and recent periods.

Figure 31: Regional snow cover comparison between 2015 and 2024, illustrating the broader cryospheric changes beyond individual glaciers.

Figure 32: Seasonal snow pattern comparison, revealing changes in both winter accumulation and summer persistence of snow cover.

Figure 33: Land Surface Temperature comparison between 2015 and 2024, providing thermal context for the observed cryospheric changes.

1. Short-term (5-15 years): Accelerated melt may actually increase dry-season river flows, potentially masking the long-term crisis. This phase presents opportunities for hydroelectric generation but also increased flood risks.
2. Medium-term (15-50 years): As glacier mass diminishes, the "buffer" function weakens. Dry-season flows will begin declining, stressing agricultural and urban water supplies during the critical pre-monsoon period.
3. Long-term (50+ years): Severely diminished glaciers will transform Himalayan rivers from glacier-fed to primarily rain-fed systems, with dramatically reduced dry-season flows and increased seasonality.

• India: Over 30 major hydroelectric projects in Uttarakhand, Himachal Pradesh, and Sikkim
• Nepal: Hydroelectric exports represent a major revenue source, with numerous projects planned
• Bhutan: National GDP heavily dependent on hydroelectric exports to India
• Pakistan: The Tarbela and Mangla dams depend on Indus flows originating from Himalayan glaciers

• Wheat production in Punjab and Haryana requires irrigation during the dry rabi season
• Rice paddies across Bangladesh depend on Brahmaputra flows
• Timing shifts in meltwater release can misalign with critical crop growth stages

• India-Pakistan: The Indus Waters Treaty, governing shared river resources, faces increased stress as absolute water volumes decline
• India-China: Brahmaputra/Yarlung Tsangpo management disputes intensify with changing flow regimes
• Nepal-India: Hydroelectric development negotiations become more contentious as resource constraints tighten

1. Cloud contamination: The monsoon season (June-September) presents persistent cloud cover over the Himalayas, limiting optical satellite observations during the ablation season
2. Debris-covered glaciers: The NDSI method may underestimate glacier extent where thick debris cover masks the underlying ice. Studies suggest 10-30% of Himalayan glacier area is debris-covered
3. Seasonal vs. permanent ice: The analysis includes seasonal snow within the glacier classification. Focus glacier statistics (14-22% loss) better represent true glacier change than regional estimates (54.89%)
4. Spatial resolution trade-offs: Regional analysis at 2000m scale sacrifices detail for computational efficiency; fine-scale 30m analysis is limited to focus regions

1. Threshold sensitivity: The NDSI threshold of 0.4 is well-validated but represents a binary classification of a continuous phenomenon. Small threshold changes can affect area estimates by several percent
2. Temporal alignment: Comparing annual composites reduces but does not eliminate seasonal differences between comparison years
3. Atmospheric effects: Despite Level 2 surface reflectance products, residual atmospheric effects may introduce noise

1. Establish comprehensive monitoring programs: Invest in ground-based validation stations to complement satellite observations. Priority locations should include the three focus glaciers analyzed here.
2. Develop adaptive water management frameworks: Transition from supply-side to demand-side water management in downstream regions. Implement efficiency measures before crisis conditions emerge.
3. International cooperation mechanisms: Strengthen transboundary water governance through updated treaties reflecting changing resource availability. The Indus Waters Treaty provides a foundation but requires modernization.
4. Climate adaptation investments: Prioritize infrastructure resilience, drought-resistant agriculture, and alternative water sources (groundwater, recycling) in glacier-dependent regions.

1. Re-evaluate long-term project viability: Hydroelectric investments with 50+ year horizons face substantial climate risk. Conduct detailed climate-adjusted feasibility studies.
2. Diversify energy portfolios: Balance hydroelectric investments with solar and wind assets less vulnerable to hydrological variability.
3. Price water risk appropriately: Financial models should incorporate declining glacier contribution to river flows over project lifetimes.

1. Extend temporal coverage: Continue satellite monitoring through 2030 and beyond to capture evolving trends.
2. Improve debris-covered glacier detection: Develop algorithms combining optical, thermal, and radar data to better characterize debris-covered ice.
3. Integrate mass balance measurements: Satellite area measurements should be complemented with gravimetric (GRACE) and altimetric (ICESat-2) mass balance data.
4. Downscale climate projections: Develop high-resolution regional climate models to improve projections for specific watersheds and glacier systems.

• The Khumbu Glacier lost [21.98%](Landsat NDSI analysis) of its ice-covered area—nearly one-quarter of this iconic glacier vanished in ten years
• The Gangotri Glacier, source of the Ganges, contracted by [14.44%](Landsat NDSI analysis), threatening the water security of hundreds of millions
• The Siachen Glacier retreated by [19.84%](Landsat NDSI analysis), with [472 km² of ice loss](area calculation)—an area larger than the city of Vienna
• Regional snow cover declined by [37.3%](MODIS trend analysis) over the decade, indicating reduced accumulation that compounds glacier stress These are not abstract environmental statistics. They represent the transformation of water storage systems that humanity has depended upon for millennia. The glaciers of the Himalayas are the water towers of Asia, and they are emptying. The analysis confirms that the rate of retreat observed in recent decades is not slowing but accelerating. The window for preventing catastrophic glacier loss has likely passed; the focus must now shift to adaptation—managing the transition from glacier-fed to rain-fed river systems while minimizing human suffering and geopolitical conflict. The satellite imagery tells an unambiguous story: the Third Pole is melting. The question is not whether we will face a transformed Himalayan hydrology, but how well we prepare for its inevitable arrival.

• USGS Landsat Collection 2 Level 2 Science Products
• NASA MODIS Snow Cover (MOD10A1)
• NASA MODIS Land Surface Temperature (MOD11A2)
• Google Earth Engine Data Catalog

• ICIMOD Hindu Kush Himalaya Assessment
• IPCC Special Report on the Ocean and Cryosphere
• Nature: The State of the Cryosphere
• NASA Global Temperature Records

• World Bank Ganges Basin Overview
• World Bank Indus Waters Treaty Fact Sheet
• Asian Development Bank Climate Economics

• Bounding Box: 78°E to 92°E, 27°N to 35°N
• Approximate Area: 1.36 million km²
• Countries: Nepal, India (Uttarakhand, Sikkim), Bhutan, Tibet (China)

1. Image Collection: Filter Landsat 8/9 and MODIS collections by date, location, and quality
2. Compositing: Create annual median composites to reduce cloud contamination
3. Index Calculation: Compute NDSI for glacier/snow classification
4. Classification: Apply NDSI > 0.4 threshold for glacier identification
5. Area Calculation: Sum pixel areas within glacier masks
6. Change Detection: Subtract recent from baseline classifications
7. Trend Analysis: Apply linear regression to time series data

• Cloud cover filtering (< 20-30%)
• Median compositing for noise reduction
• Native resolution analysis for focus glaciers (30m)
• Consistent methodology across all time periods