IS CLIMATE CHANGE EVIDENT IN WELLINGTON’S RAINFALL RECORDS? IMPLICATIONS FOR STORMWATER CATCHMENT MODELLING

D. Maxwell (SLR Consulting Ltd), A. Osborne (Wellington Water Ltd)

ABSTRACT

As a result of climate change, temperatures are expected to increase. Warmer temperatures mean that more water vapor will enter the atmosphere, while also increasing the air’s ability to hold moisture. This may result in changes in rainfall that are then amplified in runoff.

Accounting for future changes in runoff due to climate change is a key element of stormwater hydraulic modelling and essential in the management of stormwater runoff and flood risk. As a critical input to these models, Wellington Water wanted to better understand future potential rainfall due to climate change and identify if these effects are already apparent in rainfall records across the region. This paper presents the results of analyses undertaken to investigate this and compares these findings with NIWA’s climate change projections for the region.

Records from nine rain gauges, well distributed across Wellington stormwater catchments, were used to identify any trend in annual and seasonal rainfall maxima. The sites were selected based on record length (including consideration of gaps) and the quality of the data. Two sites have over 100 years of record – Kaiwharawhara Stream at Karori Reservoir and Wainuiomata River at Wainuiomata Reservoir. A long-term record for Kelburn was formed by combining data from two gauges at the same location. In most cases sub-hourly observations commenced in the late 1990s or early 2000s providing a much shorter period for analysis.

Changes in rainfall intensity were also investigated using rainfall maxima from a range of different event durations (i.e. 10 minutes to 24 hours). Regression analysis was used to determine whether there was any trend in observed rainfall records to date.

To determine if the frequency of large events has changed over time, the number of events exceeding a threshold was also investigated for each duration. The selection of the threshold value for each duration is subjective but the aim is to reflect any possible temporal change in the frequency distribution that could be from climate change. As such, the thresholds chosen aim to avoid sampling biases while being able to account for seasonal and long-term variations. The choice of threshold provides on average 3-5 exceedances per year of record. To avoid dependence (i.e. consecutive daily rainfalls resulting from a single weather event), there needed to be at least a 12-hour gap between rainfall events.

The results show that, to date, although there are some significant increases or decreases in rainfall, these are often not consistent over the gauge network nor over various durations. For the most part, trends in rainfall are relatively weak, which is consistent with the relatively small changes (within ±2%) in rainfall that are projected by 2040 (i.e. 2031-2050) from the 1986-2005 long term mean (NIWA, 2022).

However, in general, annual rainfall has decreased at the Wainuiomata gauges and other areas of the (upper) Hutt Valley. In Wellington City (Kaiwharawhara site) appears to be increasing. These findings are consistent with projected changes presented by NIWA (2022).

Rainfall maxima over several durations (rainfall intensity), rainfall appears to be increasing Wellington city (at Kaiwharawhara and Kelburn sites) for durations of 1-hour or greater, with decreasing trends for the Orongorongo area and Whakatikei sites, although these were not significant.

Seasonally, the most significant trends have been in rainfall maxima during autumn, with increases at Kaiwharawhara and Whakatikei in the south-west and north-west, respectively, for durations greater than 1-hour. A decrease in autumn and winter rainfall maxima at Orongorongo has also been observed for the same durations. There are very few obvious or consistent trends for rainfall maxima in other seasons or for other durations, possibly due to the localised nature of very short intense rainfall bursts. These intense rainfall events may not always occur over a gauge, and may therefore be missed in the observational records.

In general, the frequency of large events has been increasing, particularly at Kaiwharawhara, Whakatikei and Mangaroa across most event durations. A decrease in the intensity of these events has been observed at the Orongorongo site.

In conclusion, although there is some evidence of climate change in rainfall records across Wellington, the relationships identified are generally weak and inconsistent. The amount of change projected for the 2031-2050 period is relatively small (±2%) for most parts of Wellington and therefore could account for the few clear trends identified to date.

Despite this, there is some evidence that Wellington city and western areas of the region have observed an increase in rainfall over time while eastern areas are largely experiencing a decrease in rainfall. This is consistent with NIWA’s (2022) most conservative climate change projections (RCP 8.5), noting that the there is considerable variability depending on the timeframe (i.e. annual or seasonal; 2031-2050 or 2081-2100) and Representative Concentration Pathway modelled.