Snow glacier melt estimation in tropical Andean glaciers using Artificial Neural Networks

Abstract

Abstract. Snow and glacier melt (SGM) estimation plays an important role in water resources management. Although melting process can be modelled by energy balance methods, such studies require detailed data which is rarely available. Hence, new and simpler approaches are needed for SGM estimations. Artificial Neural Networks (ANN) is a modelling paradigm able to reproduce complex non-linear processes without the need of an explicit representation. The present study aims at developing an ANN based technique for estimating SGM rates using available and easy to obtain data such as Temperature and short wave radiation. Several ANN models were developed to represent the SGM process of a tropical glacier in the Bolivian Andes. The main data consisted on short wave radiation and temperature. It was found that accuracy may be increased by considering relative humidity and melting from previous time steps. The model represents the daily pattern showing variation of the melting rates throughout the day, with highest rate at noon. The melting rate in October (1.35 mm h−1) is nearly three times higher than July's melting rate (0.50 mm h−1). Results indicate that the exposure time to melting in October is 12 h, while in July is 9 h. This new methodology allows estimation of SGM at different hours throughout the day, reflecting its daily variation which is very important for tropical glaciers where the daily variation is greater than the yearly one. This methodology will provide useful data for better understanding the glacier retreat process and for analysing future water scenarios.