The importance of observed gradients of air temperature and precipitation for modeling runoff from a glacierized watershed in the Nepalese Himalayas

Abstract

The performance of glaciohydrological models which simulate catchment response to climate variability depends to a large degree on the data used to force the models. The forcing data become increasingly important in high-elevation, glacierized catchments where the interplay between extreme topography, climate, and the cryosphere is complex. It is challenging to generate a reliable forcing data set that captures this spatial heterogeneity. In this paper, we analyze the results of a 1 year field campaign focusing on air temperature and precipitation observations in the Langtang valley in the Nepalese Himalayas. We use the observed time series to characterize both temperature lapse rates (LRs) and precipitation gradients (PGs). We study their spatial and temporal variability, and we attempt to identify possible controlling factors. We show that very clear LRs exist in the valley and that there are strong seasonal differences related to the water vapor content in the atmosphere. Results also show that the LRs are generally shallower than the commonly used environmental lapse rates. The analysis of the precipitation observations reveals that there is great variability in precipitation over short horizontal distances. A uniform valley wide PG cannot be established, and several scale-dependent mechanisms may explain our observations. We complete our analysis by showing the impact of the observed LRs and PGs on the outputs of the TOPKAPI-ETH glaciohydrological model. We conclude that LRs and PGs have a very large impact on the water balance composition and that short-term monitoring campaigns have the potential to improve model quality considerably. Key Points Precipitation is variable and uniform precipitation gradients cannot be derived Temperature lapse rates are not constant throughout the year and shallow Temperature lapse rates and precipitation gradients are key inputs in modeling