Mesure et modélisation d'équivalents en eau de la neige dans un contexte de prévision hydrologique

Abstract

Snow water equivalent (SWE) is an important hydrological parameter in Nordic regions which has traditionally been measured manually by core sampling. In order to replace this manual survey by an automated solution, Rio Tinto Aluminum (RTA) has recently acquired nine GMON, sensors allowing real-time SWE acquisition from attenuation of natural soil radioactivity and installed them at several weather stations in the Lac St-Jean watershed. In this context, the first objective of this study is to evaluate the quality of GMON data concerning its precision and reliability by comparing it against other types of snow measurements such as manual snow survey and snow height derived from sonic sensors. The second objective is to investigate the possibility of utilizing the weather data in order to generate SWE for stations non-equipped with GMON. To this end, usable data inventory is created in order to select the most appropriate snow models. Our results show that the GMON data measurements are not performed continuously and a non-negligible number of outliers, periods with excessive noise and missing data are observed. All the usable GMON series (16 out of 21) show a systematic delay in the detection of the disappearance of snowpack as compared to the reference (measure of the snow height), with its median of 6 days. Moreover, in comparison to the manually acquired SWE data, at least one of the GMON shows a bias of more than 15%. Regarding the snowpack modeling, the lack of homogeneity among the stations in the types of measured parameters and the data quality results in limiting the number of snow models. Solely simple models based on the empirical relations can be tested and it was thus challenging to evaluate the extent to which numerical simulations can generate SWE data for non-GMON equipped stations. The following series of operational recommendations are suggested in order to improve the measurements and the numerical simulations of SWE. Firstly, despite the valuable information that can be derived from GMON, it is strongly recommended to implement a monitoring system that verifies the GMON data prior to further deployments of GMON. Secondly, the potential of numerical simulations can be further extended by developing a program that controls the quality of the collected data and by uniformly installing the instruments such as anemometers and relative humidity sensors covering the whole watershed. Finally, this study also sheds light on future research directions. For instance, numerical simulations of SWE could be improved when one takes into consideration of additional factors such as catch efficiency of snow and interception by the forest canopy. Moreover, alternative ways of utilizing the soil radioactivity for SWE measurement could be explored.

Date	4 Dec 2019
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Michel Baraër (Supervisor) & Richard Arsenault (Co-supervisor)