Abstract
This thesis examines the impact of predicted global climate change on alpine catchment discharge from the eastern Southern Alps of Canterbury. The reasons for this assessment are twofold. Firstly, the water resources of Canterbury and other drier parts of New Zealand are becoming limited due to demand for irrigation, hydroelectric generation and recreation. Secondly, studies overseas have shown water resources of alpine catchments to be highly sensitive to predicted global climate change, but as yet this subject has received little attention in New Zealand.
The research objectives are: (1) To investigate and choose a methodology to assess the impact of climate change on discharge from alpine catchments. (2) To predict the effects of potential climate change on alpine catchment discharge from the eastern Southern Alps of Canterbury.
The HBV3-ETH9 conceptual runoff model is chosen for seven reasons: easily met data input requirements, appropriate spatial and temporal scale, distributed snow and glacier representation, simple calibration and verification technique, accurate simulations of discharge in previous studies in several different environments, model availability and previous use for similar climate change studies. It is applied to four catchments with differing characteristics ( J ollie, Hooker, Rangitata, and Rakaia). The model is calibrated and verified using daily inputs of precipitation, mean temperature and discharge. Climate change scenarios are generated from downscaled GCM results.
Model performance indicates the HBV3-ETH9 model can be applied successfully to alpine catchments in New Zealand under the current climate. Verification period model efficiency (R2) values range from, 0.70- 0.78 (Jollie), 0.58- 0.81 (Hooker), 0.61 - 0.85 (Rangitata) and 0.45- 0.83 (Rakaia). Model performance is largely related to representation of areal precipitation.
By the 2080's higher volumes of discharge are predicted in all seasons by all scenarios (except under the CSIRO scenario for summer). For example, mean seasonal discharge from the Jollie catchment using the HadCM2 scenario for the 2080's increases by 1.5 m3 /s in autumn, 5 m3 /s in winter, 3 m3/s in spring and 1.5 m3/s in summer. The largest predicted increase in discharge is during late winter and spring, when substantially larger floods are likely. For example, late winter and spring floods double in peak volume under the 2080's HadCM2 scenario for the Rangitata. An increased proportion of discharge will occur in winter and early spring. For example, mean monthly discharge as a proportion of mean annual discharge from the Jollie catchment during August jumps from 0.6 to 1.1, and during January drops from 1.7 to 1.5 using the HadCM2 scenario for the 2080's. This is partially caused by a reduced proportion of precipitation falling as snow. For example, currently 40 per cent of precipitation falls as snow in the Hooker catchment, this reduces to 27 per cent using the CSIRO scenario for the 2080's.
Future studies need to focus on further assessing the validity of climatic transferability of conceptual models. Physically based models driven by dynamically downscaled scenarios could reduce doubt surrounding climatic transferability. Applying such models to the alpine environment in New Zealand will require a long-term project to gather quality input data with high spatial and temporal resolution.