Contrasting climate change impact on river flows from high-altitude catchments in the Himalayan and Andes Mountains

Abstract

Mountain ranges are the world's natural water towers and provide water resources for millions of people. However, their hydrological balance and possible future changes in river flow remain poorly understood because of high meteorological variability, physical inaccessibility, and the complex interplay between climate, cryosphere, and hydrological processes. Here, we use a state-of-the art glacio-hydrological model informed by data from high-altitude observations and the latest climate change scenarios to quantify the climate change impact on water resources of two contrasting catchments vulnerable to changes in the cryosphere. The two study catchments are located in the Central Andes of Chile and in the Nepalese Himalaya in close vicinity of densely populated areas. Although both sites reveal a strong decrease in glacier area, they show a remarkably different hydrological response to projected climate change. In the Juncal catchment in Chile, runoff is likely to sharply decrease in the future and the runoff seasonality is sensitive to projected climatic changes. In the Langtang catchment in Nepal, future water availability is on the rise for decades to come with limited shifts between seasons. Owing to the high spatiotemporal resolution of the simulations and process complexity included in the modeling, the response times and the mechanisms underlying the variations in glacier area and river flow can be well constrained. The projections indicate that climate change adaptation in Central Chile should focus on dealing with a reduction in water availability, whereas in Nepal preparedness for flood extremes should be the policy priority.

Keywords: climate change; glaciers; high-altitude water cycle; hydrological modeling; river flow.

Fig. 1.

Study catchments and glacier and debris-covered area at the beginning of the century. (A and B) The boxes showing the 25th, 50th, and 75th percentile of the GCM ensemble projections indicate relative changes in projected decadal mean annual precipitation and temperature with respect to the reference period (2001–2010). (i–iv) Ensemble-median projected glacier area (RCP45 and RCP85) by the year 2100.

Fig. 2.

Future catchment runoff and composition of total water input. The figure shows the median of all simulations for each decade, RCP, and study area (A and B: Upper Langtang catchment; C and D: Juncal catchment). The error bars represent the 80% confidence interval about the climate model ensemble.

Fig. 3.

Future evolution of ice melt contribution to runoff, future glacier area, and glacier area contributing to ice melt. The lines show the median of all simulations for each decade, RCP, and study area (A–C: Upper Langtang catchment; D–F: Juncal catchment). Subdebris ice melt (A) is a component of total ice melt in the Upper Langtang catchment. The debris-covered glacier area and non–debris-covered glacier area contributing to ice melt are shown separately (B and C). The bars represent the 80% confidence interval about the climate model ensemble (also shown for debris area contributing to ice melt but too close to the line to see).

Fig. 4.

Simulated seasonal runoff cycles and water input composition. The median results of all simulations are represented. Error bars represent the 80% confidence interval of the climate model uncertainty. The composition of water inputs is shown for the reference period 2001–2010 (A and D), for the period 2091–2100 and RCP45 projections (B and E), and for the period 2091–2100 and RCP85 projections (C and F). Solid lines represent the median of all simulations for the reference period (2001–2010) and two future periods (2051–2060 and 2091–2100).

Fig. 5.

Simulated annual maximum daily runoff corresponding to average recurrence intervals of 2 y (ARI 2), 10 y (ARI 10), and 100 y (ARI 100) for both RCPs (RCP45 and RCP85) and study areas (A: Upper Langtang catchment; B: Juncal catchment). The figure shows the median outputs of the stochastic runs for the reference period (2001–2010) and for two future decades (2051–2060 and 2091–2100).