Abstract
Because New Zealand relies heavily on water for electricity generation, it requires strong and reliable information about its water supplies as well as better knowledge about the processes that affect them. Situated in the Southern Alps, the Waitaki basin is the most important hydro catchment in New Zealand. Three alpine sub-catchments, namely Ohau, Pukaki, and Tekapo, provide most of the discharge to the Waitaki River. In this alpine region a large part of the water resource is temporarily stored as seasonal snow cover. To utilize better the value of water in hydro lakes, improving knowledge of the timing and supply of water from seasonal snow is a priority.
It has been long established that satellite remote sensing is a powerful tool to monitor snow cover in remote and inaccessible areas. In New Zealand, this technology has received only scant consideration. In addressing the remote sensing of the seasonal snow cover in the alpine catchments of the Waitaki basin, this thesis aims at filling a considerable void. This is achieved through the implementation of routine monitoring of the snow cover dynamics with the MODerate Imaging Spectro-radiometer (MODIS). Towards this goal, several advanced remote sensing techniques that are novel to MODIS are integrated in a single and operational algorithm.
This research demonstrates the desirable performance of an image fusion algorithm. The algorithm enables the mapping of snow with MODIS at 250m spatial resolution instead of the 500m spatial resolution imagery currently available. Furthermore, MODIS images are standardized by means of a physically-based atmospheric and topographic correction (ATOPCOR) approach. Finally, the radiometric normalization of the time series permits the design of a robust spectral unmixing technique. This allows further enhancement of the spatial details of the snow maps through the determination of sub-pixel snow fractions at 250m spatial resolution. Together, the combination of these techniques forms a processing chain, well suited to the mountainous environment, to map snow with the highest possible amount of spatial detail. A careful assessment of the quality of the maps of snow fractions is conducted by means of comparison with high resolution reference snow maps obtained from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER).
The processing of seven years of MODIS observations covering the 2000--2006 hydrological years permitted the creation of a new dataset that depicts the spatial distribution of snow. Based on this dataset, this thesis demonstrates that the current modelling approach of the snowpack by the model SnowSim tends to propagate errors that increase to an important level. Every year by the end of the ablation season SnowSim models nearly a quarter of the total water storage in locations that are free of snow according to observations from MODIS. Finally, the hydrological modelling approach enabled by the Snowmelt Runoff Model (SRM) is revisited. Daily meteorological data (i.e., temperature and precipitation) and the frequent observations of the snowpack provided by MODIS enable the daily discharge to be simulated. Unprecedented performance in the simulation of daily inflows are obtained for the three largest water reservoirs in New Zealand. This sheds new light on the relative contribution of seasonal snowmelt and ice melt to the discharge.
In revealing the large daily variability of the snowmelt, new estimates of its contribution to the lake inflows are obtained. Over the study period, snowmelt accounted for 37%, 40%, and 31% of the discharge in the Lakes Ohau, Pukaki, and Tekapo catchments, respectively. Finally, this research documents the severe drought of 2005. It strongly suggests that inflows were largely mitigated by ice melt from glaciers in the Pukaki basin. A contribution of glacier melt much larger than usual is believed to have sustained the discharge to within 17% of the mean annual flow, although precipitation was reduced by 34%. This mitigating factor was less marked in Tekapo and not observed in the Ohau basin, in accordance with the relative proportion of glacierized areas in the catchments. This potentially provides a striking example of the contribution of long term storage to inflows during dry periods.