Abstract
Predictive understanding of precipitation H-2 and O-18 in New Zealand faces unique challenges, including high spatial variability in precipitation amounts, alternation between subtropical and sub-Antarctic precipitation sources, and a compressed latitudinal range of 34 to 47 degrees S. To map the precipitation isotope ratios across New Zealand, three years of integrated monthly precipitation samples were acquired from >50 stations. Conventional mean-annual precipitation H-2 and O-18 maps were produced by regressions using geographic and annual climate variables. Incomplete data and short-term variation in climate and precipitation sources limited the utility of this approach. We overcome these difficulties by calculating precipitation-weighted monthly climate parameters using national 5-km-gridded daily climate data. This data plus geographic variables were regressed to predict H-2, O-18, and d-excess at all sites. The procedure yields statistically-valid predictions of the isotope composition of precipitation (long-term average root mean square error (RMSE) for O-18 = 0.6 parts per thousand; H-2 = 5.5 parts per thousand); and monthly RMSE O-18 = 1.9 parts per thousand, H-2 = 16 parts per thousand. This approach has substantial benefits for studies that require the isotope composition of precipitation during specific time intervals, and may be further improved by comparison to daily and event-based precipitation samples as well as the use of back-trajectory calculations.