Climatology and trends of Atmospheric Water Vapour Transport in New Zealand

Atmospheric moisture transport is crucial for understanding New Zealand's climate dynamics, particularly in influencing extreme precipitation events. While the majority of previous studies have focussed on Atmospheric Rivers (ARs), this study examines the entire spectrum of water vapour transport and its link to extreme precipitation using 40 years (1981-2020) of Integrated Water Vapour Transport (IVT) data over the region. Although ARs are important drivers of extreme precipitation, they are infrequent as they only account for less than 10% of total moisture transport at most coastal locations. Extreme water vapour transport (defined by the 90th percentile IVT threshold) corresponds more closely with precipitation extremes than ARs alone, even using an expanded AR detection range. Here, IVT is classified into strength categories from weak to strong. Over the study period, all but the weakest category of IVT has increased in frequency of occurrence over most of South Island, while decreasing in northern North Island. Similarly, monthly IVT anomaly trends show a positive trend in the South Island and negative trend in the northern North Island during warmer months. Separate analysis of windspeed and total column water (TCW) reveals the driving factors underpinning these IVT trends: TCW has increased universally across New Zealand, while windspeed has decreased in more northern regions. Correspondingly, these findings indicate the importance of analysis both dynamic and thermodynamic factors in seeking to understand hydrometeorological variation and when investigating the responses to climate change.


Introduction
The transport of atmospheric moisture is a key component of the global water cycle as it represents the atmospheric branch that links evaporation from the oceans and precipitation over the continents (Gimeno et al. 2016).In addition to factors leading to atmospheric instability and strong vertical motion, the potential for intense and sustained moisture transport is vital for driving extreme precipitation events (Trenberth et al. 2003;De Vries 2021;Gimeno-Sotelo and Gimeno 2023).This atmospheric transport of moisture can be quantified in a eulerian framework as Vertically Integrated Water Vapour Transport (IVT) and is one of the key parameters for defining atmospheric rivers (ARs) (De Vries 2021; Gimeno-Sotelo and Gimeno 2023).ARs are long and transient plumes of water vapour in the lower atmosphere that transport moisture from the tropics towards the poles and are "organised structures with high IVT" (Gimeno-Sotelo et al. 2023).Over 90% of meridional moisture transport is often attributed to ARs (Gimeno et al. 2014;Liang et al. 2023), but extreme precipitation events are not solely restricted to the occurrence of ARs (Gimeno-Sotelo and Gimeno 2023).IVT structures that are not considered as ARs (due to their geometric structure, magnitude, or duration) can also play an important and less studied role on hydrometeorological impacts (Lorente-Plazas et al. 2020;Reid et al. 2021).
In New Zealand, as with many regions globally, high levels of IVT are associated with major flood events.For example, the top winter maximum floods on the Waitaki River  were all associated with ARs (Kingston et al. 2016), and nationwide more than 48% of the extreme precipitation events and 50% of mean annual runoff have been attributed to ARs (Paltan et al. 2017).A comprehensive analysis of over 650 rain gauges in New Zealand, for example, revealed that 40-86% of rainfall totals and 50-98% of extreme precipitation event were associated with an AR, with the effect being more predominant in the austral summer season (Shu et al. 2021).It is noteworthy, however, that in some instances moisture transport structures attain magnitudes comparable to that of ARs but without conforming to the typical AR spatial patterns (Kingston et al. 2022).This suggests that high IVT values also play an important role in extreme hydro-meteorological events, particularly when ARs are not the dominant mode of moisture transport It should also be noted that different AR detection methods, such as those examined by the Atmospheric River Tracking Method Intercomparison Project (ARTMIP), utilize different criteria which can result in some differences in the detection of ARs.Even though AR landfall agreement between methods increases for stronger AR categories, there still exists a large spread in track density attributed to differences in how AR object size and shape, especially marginal features, are delineated (Zhou et al 2021;Lora et al. 2020).
There are several advantages to using IVT as a metric for extreme event hydrometeorological events, in particular corresponding to the dependence of IVT on both humidity (representing atmospheric thermodynamic variation) and wind (representing dynamic variation) (McClenny et al. 2020).In the extratropics for example, for extreme precipitation to occur there should be significant and sustained water vapour contributions from outside, as local atmospheric humidity alone is insufficient (Trenberth et al. 2003;Gimeno-Sotelo et al. 2023).
IVT is favoured over other variables such as Integrated water vapour (IWV) in AR detection algorithms and related studies because IVT better reflects the moisture flux and can be more related to extreme precipitation over complex terrain (Rutz et al. 2014;Swales et al. 2016 andGerushnov et al. 2017).Furthermore, studies in Europe and western United States show that there is a higher predictability of IVT compared to precipitation.The higher predictability and its strong connection with precipitation make it a potentially useful variable in forecasting extreme hydrometeorological events (Lavers et al. 2014, Lavers et al. 2016).IVT is not the sole requirement for precipitation to occur, but nonetheless the combined attributes of IVT, especially its application over complex terrain, make IVT particularly suitable for understanding extreme precipitation events.
In response to global warming, the amount of moisture in the atmosphere is expected to rise, following the thermodynamic constraints of water holding capacity set by the Clausius-Clapeyron equation (e.g.Trenberth et al. 2005).In tandem with these thermodynamic changes, dynamic factors may also be importantparticularly in regions where wind plays a significant role in influencing extreme precipitation (Gimeno-Sotelo 2023).Focusing on New Zealand, significant shifts in both thermodynamic and dynamic aspects have been observed.
Between 1909 and 2022, the annual average temperature in New Zealand increased by 1.26°C (± 0.27) (Ministry for the Environment & Stats NZ 2023).Since 1960, there has been an increasing trend in average annual rainfall and annual maximum one-day rainfall is seen in the southern South Island, with parts of the northern North Island seeing a decrease (MfE, 2023).Conversely, extreme winds are decreasing in most sites in New Zealand including at exposed coastal locations such as Wellington (in the North Island) and Invercargill (in the South Island) which are associated with the highest mean wind speeds across the country (Macara, 2018).Further changes are expected under climate change scenarios: a likely continued shift of storm tracks towards the poles in the Southern Hemisphere will result in an increase in ARs in those regions (Espinoza et al. 2018;Douville et al. 2021).
The aim of this paper is to construct a detailed climatology of IVT across New Zealand, going beyond the existing focus on ARs to get a broader understanding of moisture transport.
The relationship between IVT and precipitation will also be investigated, particularly in the context of extreme weather events.Finally, historical trends in IVT and its thermodynamic and dynamic constituents will be explored to investigate whether the linkage between moisture transport and extreme precipitation may have changed over time.This is pivotal in understanding how climatic and atmospheric changes are reflected in New Zealand's moisture transport, thereby providing critical insights for possible future climates and corresponding adaptations that may be required.

Study Location
The weather of New Zealand is governed by its maritime position, the prevailing westerly circulation pattern and modification of embedded midlatitude cyclonic systems by orographic lifting (Sinclair 1994).According to the Köppen climate classification system the majority of New Zealand has a temperate oceanic climate zone (Cfb), characterized by mild temperatures with no distinct dry season and a warm summer (Beck et al. 2018).Tropical cyclones that undergo extratropical transition and local convective systems also cause heavy rainfall and flooding to New Zealand (Kerr 1976;Sinclair 2004;Lorrey et al. 2014).Seasonal variation in the jet streams such as split jet structure during Austral summer also affect the storm track, and correspondingly the moisture transport in the region as well (Nakamura and Shimpo 2004;Ashok et al. 2007).In the warm season, the polar jet dominates the upper-level flow over New Zealand and is centred around 50°S (Prince et al. 2021).The bifurcation of the jetstream into subtropical jet (equatorward along 25°-30°S) and the polar front jet (at about 60°S) in the austral winter is a distinctive and persistent feature in the Southern Hemisphere.(Bals-Elsholz et al. 2001).While the equatorward position of the dominant jetstream leads to peak meridional transport over New Zealand in the warm season (Prince et al. 2021), the subdued cyclonic activity over the South Island during the austral winter as a result of poleward shift of jet leads to lowered IVT and greater negative anomalies relative to the North Island.Throughout this paper, the same simplified seasonal classification as Prince et al. ( 2021) is followed: a warm season (October to March) and a cool season (April to September).The study domain encompasses the region between 150°E to 180°E and between 20°S to 50°S, covering the entirety of New Zealand and including the surrounding oceanic regions of Tasman Sea.

IVT and primary dataset
Atmospheric data were sourced from ERA-5, which is the 5th generation of global reanalysis dataset produced by the European Centre for Medium-Range Weather Forecasts (ECMWF) (Hersbach et al. 2023).Spatial resolution is 0.25° × 0.25°, and a 40-year (1981-2020) study period was chosen, with a 6 hourly temporal resolution.IVT is the principal variable of interest as it characterizes moisture availability and transport and is defined in terms of specific humidity (q) as well as zonal(u) and meridional(v) components of wind, integrated for an entire vertical column of the troposphere (Gimeno-Sotelo and Gimeno 2023): where g is the acceleration due to gravity (9.8 m s -2 ) and dp is the pressure difference between two adjacent pressure levels.Vertical integral of northward water vapour flux (IVT v ) and vertical integral of eastward water vapour flux (IVT u ) (both units in kg m -1 s -1 ) are extracted from ERA5.Even though the derived IVT is a vector quantity, this paper focuses on the magnitude and not the direction of IVT, similar to the approach followed by Lavers and Villarini (2015).In addition to IVT, gridded data of Total Precipitation (TP) (hourly), Total Column Water (TCW, units in kg m -2 ) and wind speed are also obtained from ERA-5.

IVT classification and Atmospheric River Detection Technique (ARDT)
This paper classified IVT based on magnitude to help discern the variability of IVT patterns in different parts of New Zealand.Thresholds of IVT are classified similar to the AR categorisation developed by Ralph et al. (2019), such that weak IVT corresponds to ≥ 250 - Consistent with the aim of this study to provide a holistic assessment of moisture transport, the duration requirements outlined in the AR categorisation scheme are not applied.
Subsequently, decadal anomalies of three categories of IVT thresholds -weak, moderate, strong are counted for each grid cell and spatially mapped.
Even though the main aim is to characterise the climatology of moisture transport (IVT); this study also briefly deals with AR climatology.Detecting ARs enables comparison between IVT specifically associated with AR structures versus the full range of moisture transport.
Unlike previous AR detection methods applied over New Zealand such as Guan and Waliser (2019) and Reid et al. (2020) which are condition type algorithms (that looks for the conditions of an AR to be satisfied at a particular space and time at each grid points), the Instead of counting AR objects, here the focus was on the number of times (as days) each grid cell came under the influence of an AR as detected using the TEMPEST algorithm.In turn, this enables calculation of the relative contribution of ARs to total IVT (by dividing IVT on AR days by total IVT).
To gauge the connection between moisture transport on extreme precipitation, 15 locations with at least 40 years of daily rain data distributed across New Zealand were selected based on their local climates (Fig. 1).In each of the locations, the presence of ARs and extreme IVT on the top 100 rainy days in the previous 24 hours (from the recorded time of daily rain) was investigated.Overall, a two-pronged approach was used to characterize extreme water vapour transport.First, a location-specific approach was adopted with extreme IVT defined as exceeding the 90th percentile of IVT value in the corresponding grid cell.This allows characterization of extreme IVT relative to the local moisture transport climatology.
Secondly, a universal IVT categorization (as defined previously) on peak IVT recorded in the 24 hours before the event was applied, facilitating direct comparison across locations.
While previous studies such as Reid et al. (2021) have focused on the top 10 precipitation events for stations in New Zealand, here we expand the analysis to top 100 precipitation days so that large (i.e.impactful) precipitation events beyond just the most extreme are included.
Moreover, while detecting ARs a modified approach was followed.Not only was the presence of ARs on the location was checked, but also within a radius of 0.25° (roughly corresponding to 27.75 km).This also helps in capturing the effect of adjoining moisture structures, otherwise not categorized as AR by the TEMPEST algorithm (or any other algorithms) on extreme precipitation days.Rainfall data were accessed from the CliFlo national climate database (NIWA Taihoro Nukuraingi, 2023).

Analysing relationship between moisture transport and precipitation
Kendall's τ correlation coefficient was used to assess the association between moisture transport and precipitation using daily mean IVT and daily precipitation totals for each grid cell.Kendall's τ is a non-parametric measure of correlation that assesses the strength and direction of a monotonic relationship between two variables and is particularly suitable when data may not meet the assumptions of linearity.Daily precipitation totals were calculated from the hourly TP data.

Trend analysis
Trend analysis of monthly anomalies of IVT, along with its dynamic (windspeed) and thermodynamic (TCW) components, was also undertaken.TCW data was accessed directly from ERA-5, whereas the horizontal 10 m wind speed was calculated using the u and v component of 10 m wind available from ERA5.The Mann-Kendall (MK) test which is a nonparametric statistical test was used to detect trends in the variables for each of the grid cells.

General climatology of moisture transport
An analysis of the multiyear seasonal mean of IVT over the 40 years (Fig. 2) showed that very low mean IVT was present in both the winter and summer half-year over New Zealand.
The highest IVT is typically observed during the warm season (October -March), while the lowest occur during the cool season (April -September).Notably, during the warm season there was a strong latitudinal gradient in the IVT, especially over the oceans surrounding New Zealand.Overall, the North Island has a better inland penetration of higher mean IVT in comparison to the South Island.In both the interior and high elevation regions of the North Island (Central/Waimarino Plateau) and the South Island (Southern Alps/Ka Tiritiri o te Moana), low average IVT was observed across both seasons.In relation to mean ARIVT/IVT percentages (Fig. 4) regions in the far south (below 45ºS), and northern regions, such as Northland and Auckland (34ºS), experienced a higher amount of total IVT contributed by ARs.Overall, however, most of the coastal regions have less than 10% of entire moisture transport attributed to ARs.Christchurch, Alexandra, Lake Pukaki and Dunedin (Table 1).These sites also showed a notably weaker correlation between IVT and gridded precipitation totals.In contrast, locations such as Milford Sound/Piopiotahi, Hokitika, New Plymouth, Kaitaia, and Auckland, which also exhibited higher IVT-precipitation correlations, recorded more than half of their peak IVT readings prior to extreme rain events as at least Moderate in intensity.Across all locations in New Zealand, there were a higher number of extreme precipitation days (top 100 events) associated with the exceedance of the 90th percentile local IVT threshold compared to days with ARs detected either at the location or within a 0.25° radius.Notably, all the top 100 days in Milford Sound/Piopiotahi had 90th percentile of IVT exceeded on days of extreme precipitation.The central and eastern South Island locations exhibited a substantially higher extreme precipitation day count associated with the 90th percentile local IVT compared to AR landfallsalthough these counts still remain below 50.

IVT categories
The occurrences of different categories of IVT magnitude (at 6 hourly intervals) were counted for the first and the late 10 years of the dataset and compared (Fig. 7).Compared to the first decade (1981)(1982)(1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990), there was an increase in the occurrence of Moderate and Strong IVT in the South Island.Most regions of North Island witnessed a decrease in incidence of both of the higher IVT categories.The patterns of occurrence of weak IVT were mixed for New Zealand, but still with the strongest increases over the southern South Island.Beyond the New Zealand landmass, two distinct patterns emerge for all three categories: firstly, a marked increase in positive anomalies across the southern Tasman Sea, and secondly, a decrease in occurrences towards the north and east of the study domain.

General climatology of moisture transport
New Zealand typically experiences low magnitude average IVT across all seasons, but with higher values for the North vs.South Island.While the latitudinal temperature gradient strongly drives the mean IVT state, seasonal shifts in storm tracks, which are tied to the position of jet streams, also lead to IVT variability between the North and South islands (Fig. 2 ).The Southern Alps/Ka Tiritiri o te Moana acts as a topographic barrier to the prevailing Southern Hemisphere westerly winds and distorts the synoptic features of the South Island (Sinclair et al. 1997;Henderson et al. 1999;Porhemmat et al. 2021).These orographic distortions lead to lower IVT in the lee of the Southern Alps/Ka Tiritiri o te Moana in the South Island, particularly in the regions of Canterbury and Otago.Higher mean IVT in the south of the South Island can also be attributed to the persistence of cyclonic systems south of 50°S across all seasons (Simmonds and Keay 2000;Keable et al. 2002), emphasizing the role of cyclones as essential mechanisms of moisture transport.
IVT percentiles averaged across the wider New Zealand region have previously been analysed (Prince et al. 2021), revealing seasonal patterns for both high and low percentiles.
The analysis of spatial patterns in both universal IVT categories grid-cell specific high and extreme (i.e. 90 th and 97 th percentile) IVT in the current study provide further detail on how water vapour thresholds vary across New Zealand.In particular, coastal regions of North Island and the extreme south of the South Island are much more exposed to higher levels of moisture transport.This is notably evident in major urban centres as Auckland and Wellington which has some of the highest water vapour transport values among the locations assessed.This heightened susceptibility to moisture transport indicates a greater vulnerability to the meteorological conditions that can contribute to major flood events.On the other hand, the leeward regions of the Southern Alps/Ka Tiritiri o te Moana and Central/Waimarino Plateau of the North Island's interior recorded significantly lower IVT, indicating a generally reduced exposure to atmospheric moisture transport and associated extremes.The relatively low 90 th and 97 th percentile values for the South Island west coast are somewhat unexpected in this context, given its status as the wettest region in the country.This may be related in part to the representation of topography within the ERA5 data, and/or an upwind blocking effect of the Southern Alps/Ka Tiritiri o te Moana on the prevailing westerly circulation.
The coincidence of IVTmax and AR IVTmax over the study domain indicates the complete overlap between the most extreme water vapour transport events and detected ARsconsistent with previous studies findings around the prominent roles that ARs aplay for the most extreme precipitation events in New Zealand (e.g.Prince et al. 2021;Reid et al. 2021).
This match of AR IVTmax with IVTmax is also significant in that tropical cyclones were excluded from the AR identification process.Although this may suggest that tropical cyclones play no role in the highest instances of IVT, the reality is likely more complex.For instance, Prince et al. (2021) found that the majority of tropical cyclones in the New Zealand domain between 1979 and 1998 had associated AR structures according to the Guan and Waliser (2015) ARDT.Further research is required to more fully understand the connection between these two types of extreme events and their representation within different ARDTs.
An important finding of this study is that despite their clear role in driving extreme water vapour transport, ARs account for less than 10% of total water vapour transport (as expressed in ARIVT/IVT %) in most coastal areas of New Zealand.The northeast of North Island (around 34°S) and some locations in the south west of South Island (poleward beyond 46°S) experience higher contributions, equivalent to 13% and 17%, respectively.This means that even though ARs are impactful for extremes, they are only a minor contributor to total IVT and its variation across the New Zealand domain.

Water vapour transport and Precipitation
The spatial analysis of correlation between IVT and ARIVT with daily precipitation totals revealed moderate to strong correlation (Kendall's τ 0.4 to 0.6) along the north and western coasts of both the North Island and the South Island While previous studies, such as Reid et al. (2021), Prince et al. (2021) and Shu et al. (2021), have documented the effects of ARs on extreme precipitation events in New Zealand, several factors compel investigation of the full range of water vapour transport.Firstly, there can be variation in the identification of ARs depending on the ARDT employed (Rutz et al. 2019).
Additionally, non-AR structures significantly contribute to the background moisture flux (Fig. 4).The results show a consistent higher extreme precipitation event association with local 90 th percentile IVT thresholds compared to AR presence in all study locations.
Furthermore, weak or below-threshold IVT values are the primary precursor to extreme precipitation events in locations leeward of the Southern Alps/Ka Tiritiri o te Moana, e.g.
Christchurch (96%), Alexandra (92%) and Dunedin (89%).Results for these three locations correspond to well-known orographic blocking effects (e.g.Prince et al. 2021).However, the presence of low magnitude IVT precursors still warrants further investigation.Firstly, in terms of the specific (potentially non-AR) meteorological mechanisms driving these precipitation extremes, and secondly, to determine whether shorter duration IVT peaks within the 24-hour timeframe created a temporal mismatch in the approach of this study.Irrespective of the cause, these results highlight the additional information gained by analysing water vapour transport independently of formal AR definitions, which complements existing AR related studies in New Zealand.Thus, a multi-faceted approach considering both ARs and broader IVT patterns is necessary to fully understand hydroclimatic extremes in the region.

Anomalies in occurrence of IVT Thresholds and trend analysis of IVT and its associated components
The Importantly, studies such as Tabari (2020) have shown that an increase in water availability can lead to increased extreme rain events with a larger increase predicted for regions and seasons with high moisture availability.
In contrast to TCW, more substantial regional variation occurred for windspeed.The north/south contrast in trends has strongly influenced overall regional variation in IVT trends, indicating the importance of dynamical control on vapour transport (in addition to the underlying forcing from a warming climate).These findings are consistent with the intensification and poleward movement of the Southern Hemisphere westerlies under anthropogenic forcing (Toggweiler 2009;Deng et al. 2022).Further, this change is also partly related to changes in the Southern Annular Mode (SAM), the dominant source of internal variability in the extratropics of the Southern Hemisphere.SAM has trended towards a positive phase in recent years, especially in the austral summer due to stratospheric ozone hole depletion and increase in greenhouse gases (Thompson and Solomon 2002;Thompson et al. 2011;Swart et al. 2015).Accordingly, the poleward displacement and intensification of storm tracks associated with a positive SAM phase could explain the reduced wind speed trends seen in the northern North Island, and thus the overall trends in IVT.Studies elsewhere have similarly shown the important of variation in the wind field for regional trends in vapour transport (Nygård et al. 2020).The connection from trends in IVT to trends in precipitation is While there was an increase in the number of Strong and Moderate IVT occurrences in the South Island, negative anomalies were observed in the North Island, compelling investigation of overall trends of anomalies of IVT and its components.An upward for IVT was present in southern New Zealand and a downwards trend in the North, which was more pronounced during warm months.This was accompanied by overall increase of TCW in the study region and varying trends of wind speed anomalies with a pronounced negative trend in the north.
As global warming progresses, characterizing historical and possible future fluctuations of IVT and its constituent components allows improved understanding and attribution of the shifting connections between water vapour transport processes and extreme event response.
This interplay between dynamic and thermodynamic factors is particularly important to understand given that the expected intensification of winds associated with mid-latitude cyclones under high emission scenarios (Douville et al. 2021).Thus, future research should further investigate how the interplay between shifting circulation patterns and the thermodynamic influence of warming will modulate the frequency and characteristics of extreme precipitation events.For New Zealand, this is further complicated by the projected near-term weakening of the forced SAM change in austral summer due to the ozone recovery (Thompson et al. 2011), future studies should also consider the relative contributions of all anthropogenic drivers alongside natural variability when analysing water vapour transport changes and precipitation extremes in New Zealand.
employed here is the Tempest Extremes algorithm version 2.1 (TEMPEST; Tracking type Algorithm which works on a Lagrangian framework by tracking the propagation of ARs as whole objects over space and time)(Shields et al. 2018;Ullrich et al. 2021).TEMPEST tracks coherent moisture plumes as ARs based on laplacian edge detection of ridges in the IVT field instead of IVT itself.Filters used in this ARDT include considering only the grid points poleward of 15° N/S (to filter features too near the equator) and a minimum area of the blob ≥ 4 × 10 5 km 2 (to filter features that are too small)(Ullrich et al. 2021).Even though restricting grid points poleward of 15°N/S filters out a substantial number of tropical cyclones (TC) and TC-like objects, additional filtering was done to completely exclude TCs.
While the MK test evaluates whether the variables tend to increase or decrease over time, Theil-Sen slope estimates the magnitude of the trend.The significance of the trend for each grid cell was evaluated at the 90% confidence interval.Following the approach ofNygård et al. (2020), the trends were calculated for three time periods: whole timesteps(1981-2020, 40   years); 1981-2000 (first 20 years) and 2001-2020 (latter 20 years).The trend results over 40 years reveal the long-term changes in the magnitude of the variables, whereas results over the 20-year time periods reveal whether long-term trends have changed or strengthened over time.
. The highest correlations (τ > 0.5) were observed along the West Coast of the South Island highlighting the role of the Southern Alps/Kā Tiritiri o te Moana in enhancing precipitation on the windward regions, matching similar findings by Prince et al. (2021) and Shu et al. (2021).In contrast, the marked weakening of correlation in leeward regions clearly demonstrates the rain shadow effect created by the Southern Alps/Kā Tiritiri o te Moana, as these high relief features block the intrusion of moisture transport, resulting in weaker IVT-precipitation relationships in areas east of the mountain range.
increasing frequency of Strong and Moderate IVT over the South Island in the last two decades of the study period contrasts with a decrease in occurrence of these categories over most of North Island.Similarly, monthly anomalies of IVT have seen a positive trend over the South Island and a slight negative trend over the northern North Island, more especially during the warm season.Monthly TCW anomalies were found to have increased across all of New Zealand and it is interesting to note that the recent increasing trends (2001-2020) in TCW anomalies were predominantly observed during the warmer months and over the South Island.This matches with the stronger increases in temperature for the New Zealand region observed for summer vs. winter across the 1980-2015 period(Gutiérrez et al. in press), and the according increases in saturation vapour pressure under the Clausius-Clapeyron equation.

a
matter for further research, but recent analysis of trends in measured annual rainfall across New Zealand have indicated an increase at many sites in South Island and decreases in the northern North Island (Ministry for the Environment & Stats NZ 2023).5 ConclusionsUsing IVT as the principal variable, this paper has characterized the spatial and temporal patterns of atmospheric water vapour transport based on ERA5 reanalysis data from 1981-2020 over New Zealand.The study has extended beyond ARs, encompassing IVT in general including the constituent variables of Total Column Water (TCW) and wind speed.Despite the importance of ARs for extreme precipitation, their relative contribution to total water vapour transport was low.Additionally, the 90th percentile local IVT threshold for each grid location was more strongly associated with extreme precipitation days compared to the presence of ARs detected at or within 0.25° of that location .A strong correlation was also shown between IVT and daily precipitation totals, especially in areas influenced by orographic enhancement of precipitation.

Table 1 -
Station based categorization of peak IVT and occurrence count of ARs and extreme IVT during 24-hour lead-up to the top 100 highest precipitation days.