Abstract
This research aims to investigate the disconnect between methods of drought identification. Drought is a natural hazard but is becoming increasingly aggravated by human actions. Meteorological indices such as the SPI and SPEI are often used to monitor drought as they are readily available and easy to model compared to hydrological data. Therefore meaning they are widely used, but generally these do not directly include the human influence on drought occurrence or magnitude. Site comparisons between a natural and a human-impacted site can be used to infer the human influence on a catchment. As such, the focus for this research comes from the difficulty in quantifying the disconnect that exists between drought and human activities. The study location is the Ōmakō/Lindis River in the Clutha catchment near Wānaka.
River flow data for the relatively natural upper Lindis (Lindis Peak gauging station) and substantially impacts lower catchment (Ardgour Road) were obtained for the 2006-2019 period), alongside rainfall and air temperature data from the nearest weather station (Wānaka Airport). The meteorological indices explored were the Standardised Precipitation Index (SPI) and Standardised Precipitation-Evaporation Index (SPEI). A hydrological index, the Standardised Streamflow Index (SSI) was used to represent hydrological drought. The threshold level method approach was also used to identify drought events in the discharge record, with the 95th percentile used to determine the top low river flow events. Comparisons were subsequently made between the characteristics of dry periods identified by the standardised indices and observed river flow records in the upper and lower Ōmakō River.
The 95th percentile of mean daily river flow at the (upper) Lindis Peak is 1.105 m-3s-1, compared to 0.205 m-3 s-1 at the (lower) Ardgour Road gauging station. River flow variability at the two study sites correlate strongly in general, but when assessing extreme events there are stark differences in magnitude and length. Ardgour Road experiences fewer but more prolonged drought events compared to Lindis Peak. Lindis Peak experienced 29 events with an average length of 10 days, while Ardgour Rd experienced 17 events with an average length of 17 days. Generally, these were more likely to occur during the summer months (November- April) and matched the long term average trends for monthly river flow. Therefore, between the natural and the human-influenced location a difference was found between when standardised indices and the 95th percentile identified periods of drought.
Drought occurrence represented by SPI and SPEI values <-1 generally correspond well with low river flow for the Lindis Peak record. However at Ardgour Rd, results indicate that there are substantial differences in the timing and magnitude of low flow events for meteorological indices versus the discharge record. As such, these results suggest that caution is required when using meteorological drought indices as proxies for hydrological drought when there are strong direct anthropogenic impacts on surface water resources such as in the Ōmakō River.
This study demonstrated how a natural and a human-influenced site in Ōmakō experienced hydrological droughts in different ways and compared the characteristics of both meteorological and hydrological standardised indices in classifying drought periods. It is hoped from this research that an attempt at quantifying the human impact on drought is made through the site comparison. The wider use of this study is in seeing if meteorological indices can be used to highlight periods of hydrological drought.