Trends in New Zealand daily temperature and ... - Wiley Online Library

INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. 21: 1437–1452 (2001) DOI: 10.1002/joc.694

TRENDS IN NEW ZEALAND DAILY TEMPERATURE AND RAINFALL EXTREMES M.J. SALINGER* and G.M. GRIFFITHS National Institute of Water and Atmospheric Research, Auckland, New Zealand Recei6ed 26 June 2000 Re6ised 23 May 2001 Accepted 25 May 2001

ABSTRACT Trends in daily temperature and rainfall indices are described for New Zealand. Two periods were examined: 1951–1998, to describe significant trends in temperature and rainfall parameters; and 1930 – 1998, to ascertain the effects of two main circulation changes that have occurred in the New Zealand region around 1950 and 1976. Indices examined included frequencies of daily maximum and minimum temperatures, above and below specified percentile levels and at those levels, as well as frequencies of these above and below fixed temperature thresholds. Extreme daily rainfall intensity and frequency above the 95th percentile and the length of consecutive dry day sequences were the rainfall indices selected. There were no significant trends in maximum temperature extremes (‘hot days’) but a significant increase in minimum temperatures was associated with decreases in the frequency of extreme ‘cold nights’ over the 48-year period. There was a non-significant tendency for an increase in the frequency of maximum temperature extremes in the north and northeast of New Zealand. A decline occurred in frequency of the minimum temperature 5th percentile (‘cold nights’) of 10–20 days a year in many locations. Trends in rainfall indices show a zonal pattern of response, with the frequency of 1-day 95th percentile extremes decreasing in the north and east, and increasing in the west over the 1951–1996 period. Changes in the frequency of threshold temperatures above 24.9°C (25°C days) and below 0°C (frost days) are strongly linked to atmospheric circulation changes, coupled with regional warming. From 1930 – 1950 more south to southwest anomalous flow occurred relative to later years. In this period, 25°C days were less frequent in all areas except the northeast, and there was markedly more frost days in all but inland areas of the South Island compared with the 1951–1975 period. There was more airflow from the east and northeast from 1951 to 1975, the frequency of 25°C days increased and frost days decreased in many areas of New Zealand. In the final period examined (1976–1998), more prevalent airflow from the west and southwest was accompanied by more anticyclonic conditions. Days with a temperature of 25°C increased in the northeast only. Frost day frequencies decreased between 5 and 15 days a year in many localities, with little change in the west of the South Island and at higher elevation locations. Copyright © 2001 Royal Meteorological Society. KEY WORDS: climate

trends; New Zealand; rainfall extremes; temperature extremes

1. INTRODUCTION Many critical impacts of climate are controlled by extreme events rather than mean values. Aspects of human activity and the environment are usually well adjusted to mean climatic conditions, and show little sensitivity to moderate variations around these means. There is a general band of tolerance within which ecological, agricultural and human systems are adapted, outside which lie bands of hazard (Salinger, 1994). Their ability to adjust and respond, however, without stress and damage, declines as conditions become progressively more extreme. Systems that are particularly vulnerable are agricultural and forest ecosystems, coasts and water resources. * Correspondence to: National Institute of Water and Atmospheric Research, PO Box 109 695, Newmarket, Auckland, New Zealand; e-mail: [email protected]

Copyright © 2001 Royal Meteorological Society

1438

M.J. SALINGER AND G.M. GRIFFITHS

High temperatures can exacerbate drought conditions and increase the likelihood of fires. Very high temperatures also damage crops and reduce yields. Low temperatures can be expressed through frosts and heavy snowfalls. The former usually curtails yields in frost sensitive crops. However, low temperatures are important to some temperate horticultural and cereal crops in promoting yield and development. Heavy rainfall events are the cause of river flooding. Ericksen (1986) estimated the total cost to New Zealand of the nine most serious floods from 1968 to 1984 as averaging $NZ230 million ($NZ 2000). Increases in flood magnitude may increase costs exponentially, as areas previously considered safe would need securing. Changes in the frequency of heavy rainfall events can increase the occurrence of landslips and landslides. Drought directly impacts on the agricultural sector by reducing the number of days available for plant growth, leading to decreased yields. Droughts also affect both urban and rural water supplies. Climate change is expected to affect extremes, since small changes in mean conditions can lead to large changes in the frequency of extremes (Katz and Brown, 1992). Nicholls et al. (1996) concluded that globally, where there are adequate data, significant changes in extremes have occurred during the 20th century. There has been a clear trend to fewer extremes of low temperatures in several widely separated areas in the late 20th century. However, widespread significant changes in high temperature events have not been observed. This Intergovernmental Panel on Climate Change (IPCC) review (Nicholls et al., 1996) noted that few studies of trends in extreme rainfall and flood frequency had been made. In some regions with available data there is evidence of increases in the intensity of extreme rainfall events, but no clear, large-scale pattern has emerged. Other than in a few areas with longer-term trends to lower rainfall, there is little evidence of changes in drought frequency or intensity. Karl et al. (1999) updated some of this information, finding a reduction in the number of extremely cold days including fewer frosts, and an increase in the number of extremely hot days. Extreme precipitation had increased most rapidly over significant areas of the USA, China, Australia, Canada, Norway, Mexico and the former Soviet Union (Groisman et al., 1999). Global climate models, incorporating changes in greenhouse gases and including atmospheric (sulphate) aerosols, project temperature increases of between 1.4 and 5.8°C by 2100 (IPCC, 2001). This IPCC review concludes that the general warming would tend to lead to an increase in extremely high temperature events and a decrease in winter days with extremely low temperatures. With increasing greenhouse gas concentration, many climate models suggest an increase in the probability of intense precipitation. A number of simulations also show in some areas an increase in the probability of dry days and the length of dry spells (consecutive days without precipitation) (IPCC, 2001). New Zealand climate was cooler during the 19th century (Salinger et al., 1996). During the 20th century surface temperatures have increased and climate shifts have occurred in the New Zealand region, which might be expected to cause trends in climate extremes. Folland and Salinger (1995) note a warming in three surface mean temperature series (surface air temperature, sea-surface temperature, night marine air temperature) of 0.7°C between 1900 and the 1990s. Salinger and Mullan (1999) have noted two main circulation changes that have occurred in the New Zealand area during the 1930–1994 period: in the late 1940s and mid-1970s. From 1930 to 1950, more south to southwest anomalous flow occurred relative to later years. Airflow from the east and northeast increased during the period 1951 –1975. This was accompanied by increases in mean temperature in all regions and the national average temperature increased by 0.58°C. Wetter conditions occurred in the north of the North Island, and drier conditions in the southeast of the South Island. In the final period analysed (1976–1994), more prevalent west to southwest airflows occurred accompanying the higher incidence of El Nin˜ o events. Annual rainfall decreased in the north of the North Island and increased in the north, west, south and southeast of the South Island. However, in spite of more southwesterly flow, regional temperatures maintained their level of increase from the first period. In a study of daily maximum and minimum temperature, Salinger et al. (1996) noted for the period 1951 – 1990 that trends differ regionally. Minimum temperatures increased over this period by 0.18°C in the North Island, 0.19°C in Central Otago (in the central southern South Island) and by 0.22°C in Southland. Other South Island areas show less warming in minima. Maximum temperatures warmed by 0.14°C in the east of the North Island and by 0.11°C in the west of the South Island. Less warming in maximum temperatures occurred in other areas, amounting to around 0.05°C. Copyright © 2001 Royal Meteorological Society

Int. J. Climatol. 21: 1437 – 1452 (2001)

TRENDS IN NEW ZEALAND TEMPERATURE AND RAINFALL

1439

Only three previous studies have examined aspects of trends in New Zealand extreme climatic events. Salinger et al. (1990) identified sensitivities of extremes to changes in parameters of mean temperatures. Days above 30°C were found to increase in the order of 4–8 days/year per 1°C increase in mean maximum temperature in eastern areas of New Zealand and Central Otago. For the number of days below 0°C, there was a 3 – 20 days/year reduction per 1°C increase in mean minimum temperature throughout New Zealand. The frost-free period (days between the last air frost in spring and first air frost in autumn) increased by at least 20 – 30 days/year per 1°C increase in mean annual minimum temperature. Plummer et al. (1999) noted that the warming from 1940 to 1990 has resulted in about 10 fewer days per year with temperatures less than 0°C, and around 2 more days per year greater than 30°C in warmer locations. Manton et al. (2001) identified a decline in frequency of cool days and cold nights at the five New Zealand stations examined in a study of trends over Southeast Asia and the South Pacific. Although observed trends in extreme rainfalls were not examined, using scenarios of climate change, Salinger et al. (1990) projected that the 100-year 24-h rainfall amount would increase in most areas. The occurrence of moderate and severe drought, as measured by time series of the number of days of soil moisture deficit summations, decreased in the period 1951–1980 compared with the previous 30 years at a majority of 28 sites examined throughout New Zealand. Plummer et al. (1999) concluded that the observed changes in temperature extremes and drought frequency have been in response to atmospheric circulation changes in the region. This study has a twofold purpose: to examine the role of global warming on indices of daily temperature and rainfall extremes if any, and also to ascertain the role of climate shifts on temperature extremes above and below specific temperature thresholds. This paper examines trends and variations in annual frequencies of high and low temperature extremes over New Zealand over the period 1951–1998. Threshold temperatures of 0 and 25°C for the period 1930– 1998 were chosen for the examination of changes in mean frequencies with changes in regional circulation identified in 1950 and 1976 (Salinger and Mullan, 1999). Changes in indices of daily rainfall extremes are described for the period 1951 –1996. Results from New Zealand studies are important in the global context. They provide confirmation in the sparsely monitored Southern Hemisphere mid-latitudes, of trends in observed extreme indices described from other locations. Finally, the southwest –northeast-aligned axial ranges provide a significant topographic barrier to the prevailing west – southwest circulation. In this setting the responses of extremes to variations in circulation may strongly differ between districts in New Zealand.

2. DATA AND METHODS

2.1. Data For examination of temperature extremes, daily maximum and minimum temperature data were extracted from the New Zealand National Climate database maintained by the National Institute of Water and Atmospheric Research (NIWA). The indices were derived from data from 22 stations (Figure 1) and were analysed over the 1951 – 1998 period. To ascertain changes in frequencies of key threshold temperatures with circulation, the period 1930–1998 was used. Because of climate station closures over the last decade, the latter temperature analysis required different sets to be used for the earlier and later periods of record. For the 1930 – 1975 period, 37 stations were used and 51 stations for the 1951 –1998 period. Daily rainfall data from 25 stations throughout New Zealand, and the three offshore islands, continuous for the period 1951 – 1996, were used for the analysis of extreme rainfall indices. The examination of extreme rainfall indices used data only up to 1996 because of recent station closures at these long-term rainfall sites.

2.2. Quality control Inhomogeneities or discontinuities caused by changes in climate station site location, exposure, instrumentation and observation practices will influence climatic extremes. Salinger et al. (1992a,b) have Copyright © 2001 Royal Meteorological Society

Int. J. Climatol. 21: 1437 – 1452 (2001)

1440


Figure 1. Map of New Zealand showing all stations used (dots) and place names referred to in the text. Outlying islands include Raoul (29°15%S, 177°55%W), Chatham (43°57%S, 176°34%W) and Campbell (52°33%S, 169°9%E)

made considerable efforts to homogenize maximum and minimum temperatures and rainfall at 25 climate stations, including those used here. The records were carefully refined by a number of methods. They were screened for inhomogeneities by examining meta-data and comparisons were made with neighbouring stations to identify unrecorded site changes, or other environmental changes near the climate station site. Procedures used to homogenize the data (Rhoades and Salinger, 1993) included cumulative sum plots and neighbour station comparisons. For the 22-station temperature and 25-station rainfall set used in the analysis, artificial trends were minimized to give equivalent ‘single-site’ series of daily rainfall and maximum and minimum temperature. The threshold analyses used more temperature stations to provide counts of days below 0°C and above 24.9°C. For this set of stations, adjustments for inhomogenieites determined by Salinger (1981) were used for the additional stations using the Rhoades and Salinger (1993) method. Although adjustment by these methods of daily data is sub-optimal (Manton et al., 2001), these approaches minimize the influence of inhomogeneities on the results of the analysis.

2.3. Analysis methods Percentiles were used to calculate annual extreme indices of maximum and minimum daily temperatures. The percentiles were computed using all the non-missing days. The maximum temperature 95th percentile and the minimum temperature 5th percentile were chosen, corresponding to the 18th Copyright © 2001 Royal Meteorological Society

Int. J. Climatol. 21: 1437 – 1452 (2001)


1441

Figure 2. Trends in the frequency of days with maximum temperature above the 1961– 1990 mean 95th percentile ( frequency of hot days) over the period 1951–1998. The sign of the linear trend is indicated by +/ − symbols; bold indicates statistically significant trends (95%), with symbol indicating trends close to zero

highest daily maximum and 18th lowest daily minimum temperature in a 365-day year. The following four annual indices of temperature extremes were calculated for each year over the period 1951– 1998:

The maximum temperature 95th percentile (hot days). The minimum temperature 5th percentile (cold nights). Frequency of days with maximum temperature above long-term 95th percentile value* ( frequency of hot days). Frequency of days with minimum temperature below long-term* 5th percentile ( frequency of cold nights).

* Calculated by averaging the 30 annual 95th percentiles in the years 1961 –1990. An annual index value was deemed missing if the amount of missing data in the temperature series for any 1 year meant that the probability of lying in the top (or lower) 18 values exceeded 0.5. In the records chosen, there were very few missing values. Changes in threshold temperatures for the periods (a) 1930–1950, (b) 1951– 1975 and (c) 1976–1998 were examined for the following thresholds:

Frequency of days below 0°C ( frost days). Frequency of days above 24.9°C (25 °C days).


Int. J. Climatol. 21: 1437 – 1452 (2001)

1442


Figure 3. Trends in the frequency of days with minimum temperature below the 1961– 1990 mean 5th percentile ( frequency of cold nights) over the period 1951–1998. The sign of the linear trend is indicated by +/ − symbols; bold indicates statistically significant trends (95%), with symbol indicating trends close to zero

For rainfall extremes, the following annual indices were calculated at stations over 1951 –1996:

The 95th percentile of daily rainfall (mm) (extreme intensity). Frequency of daily rainfall exceeding the long term* mean 95th percentile (extreme frequency). Maximum number of consecutive dry days (dry spells).

A dry day is defined as a day where the rainfall total is less than 1 mm, and a rain day as a day with a rainfall total of 1 mm or more. Only rain days were used to calculate the extreme intensity and extreme frequency indices, because the distribution is skewed by many low and zero rainfall values. Trends in the annual series of each of the various indices except warm days and frost days were examined for statistically significant trends using the Mann Kendall ranked tau statistic (Tarhule and Woo, 1998). For both temperature and rainfall indices, a 95% level of confidence was chosen for significance. To identify the occurrence of change points in a series, the Pettitt test was used. This test signals a change of distribution in a sequence of observations when the initial distribution is unknown (Pettitt, 1979). As the annual seasonal cycle in temperatures is not removed, the hot day index reflects trends in summer, and the cold day index reflects trends in winter. The frost day index is an important temperature threshold for many biological activities. The warm days index was chosen to capture regional trends towards hotter summers, and 25°C is a suitable threshold for a typical New Zealand station. The rainfall indices were selected to describe whether extreme 1-day rainfalls are increasing or decreasing in either Copyright © 2001 Royal Meteorological Society

Int. J. Climatol. 21: 1437 – 1452 (2001)


1443

Table I. Year of occurrence of statistically significant change points in (a) hot days and (b) cold nights for the sites analysed, at the 0.05 probability levela Site

Hot days

Cold nights

Waipoua Forest Auckland Tauranga Taupo Ruakura New Plymouth Gisborne Napier Palmerston North Wellington Hokitika Milford Sound Appleby Ashburton Lincoln Waimate Dunedin Queenstown Ophir Invercargill

1979 1945, 1971 1970

1953

1982

1953 1953 1955 1950, 1970

1953 1947

1969 1970 1972 1970 1951, 1978

1951

1968

1968 1980

a

The analysis was performed on the entire series available from commencement of records to 1998. The symbol indicates that there is no significant change point detected in the index.

Figure 4. Nationally average frequency of days with minimum temperature below the 1961– 1990 mean 5th percentile ( frequency of cold nights) over the period 1951– 1998 for 21 stations including outlying islands Copyright © 2001 Royal Meteorological Society

Int. J. Climatol. 21: 1437 – 1452 (2001)

1444


Figure 5. Changes in the average annual frequency of days above 24.9°C (25 °C days) for the periods (a) 1951– 1975 relative to 1930–1950, and (b) 1976– 1998 relative to 1951– 1975

magnitude or frequency, and the dry spells index is a measure of whether changes in the longest dry periods are increasing or reducing each year.

3. RESULTS

3.1. Temperature The number of hot days, calculated from linear trends over the 48-year period, has increased (not significantly) in the north and northeast of the North Island and at Campbell Island (Figure 2). There are significant increases in the frequency of hot days at both Gisborne and Raoul Island by 10 and 31 days, respectively, per annum over the period. In most other parts of New Zealand there are small decreases, which are also not significant, except at Taupo. The trends in the 95th percentile value, again calculated from linear trends over the 1951 – 1998 period, were similar: generally non-significant increases between 0.5 and 1°C occurred in the north and northeast of the North Island. However, a significant warming of about 1°C occurred in the hot day index at Gisborne and Raoul Island. In contrast, values decreased non-significantly by 0.3 – 0.6°C in most other areas, except for a significant decrease at Taupo of about 1°C. Abrupt changes in the hot days index (Table I) are indicated in the early 1950s for the north and west of the North Island and in the west of the South Island. The mean value of the hot days index significantly increased from the early 1950s for the stations in these areas over the entire 48 years analysed. The frequency of cold nights shows significant trends over much of the country (Figure 3). These have declined by between 10 and 20 days per annum over the period. Only at one higher altitude location (Taupo), in the far southwest (Milford Sound) and at Campbell Island have the number of cold nights increased, but not significantly. The 5th percentile index shows significant increases of between 0.6 and Copyright © 2001 Royal Meteorological Society

Int. J. Climatol. 21: 1437 – 1452 (2001)


1445

Figure 6. Changes in the average annual frequency of days below 0°C ( frost days) for the periods (a) 1951– 1975 relative to 1930–1950, and (b) 1976– 1998 relative to 1951– 1975

1.8°C over the 1951 – 1998 period. A significant date of change in this cold nights index for northern and central areas of the North Island is around 1970 (Table I), with a significant increase in the mean value of the index after this time. Nationally averaged frequency of cold nights for 21 stations (Figure 4) show a clear trend. This index decreases by 10 days per annum over the 48-year period. The longer period analysis of changes in threshold 25°C and frost days for the three periods show distinct regional definition (Figure 5). Compared with the 1930–1950 period, in the years from 1951 to 1975 (when east and northeast airflows became more prevalent), the average number of 25°C days increased by over 10 in central Canterbury, and by between 5 and 10 in the remainder of the east and south of the South Island. Only in the northeast of the North Island did the number of 25°C days decrease. The last period (1976 – 1998), which had more west to southwest airflow, shows a different pattern of change. The frequency of 25°C days decreases by 5 days or more per annum in the central North Island. Many other areas show a decrease in frequency, except the east of the North Island, and in the northeast of the South Island. The number of threshold 25°C days increases most in Gisborne. The spatial pattern of change in frequency of frost days from the first (1930–1950) to second (1951– 1975) period is regionally distinct (Figure 6). In most areas there is a decrease, by over 10 days per annum in the central North Island, and inland northern areas of the South Island. In many other areas the decline is between 5 and 10 days per year. However, the number of frost days increases in the inland southeast of the South Island by up to 5 days a year. Between the 1951–1975 and 1976–1998 time periods most areas show a decrease in frost days. These are largest in the southeast of the South Island, with a decrease in frequency by 10 days. Only in the west of the South Island and the odd higher elevation site is there little change in frequency of frost days. Copyright © 2001 Royal Meteorological Society

Int. J. Climatol. 21: 1437 – 1452 (2001)

1446


Figure 7. Trend in the average intensity of daily rainfall events equal to the 95th percentile (extreme intensity) calculated on only rain days with rainfall totals of 1 mm or more, over the period 1951– 1996. The sign of the linear trend is indicated by +/ − symbols; bold indicates statistically significant trends (95%), with symbol indicating trends close to zero

3.2. Rainfall The extreme intensity index (Figure 7) shows significant decreases at Taihape and Lincoln. The decrease in the extreme intensity amounts to between 15 and 25% over the 1951–1996 period at these locations. Three western sites at New Plymouth, Milford Sound and Queenstown show a non-significant increase. At many other locations no trend is observed in this index. Changes in the extreme frequency index shows a similar pattern of change (Figure 8). The number of days above the 1961 – 1990 average 95th percentile shows an increasing frequency of exceedences at Ruakura and New Plymouth, and in the west and south of the South Island. These increasing trends are non-significant. A decrease in the extreme frequency occurs in the north and east of the North Island, Lincoln, Timaru, Raoul and Chatham Islands, ranging between 0.5 and 4 days less by 1996. In the 48-year record an abrupt change is indicated around the mid-1950s for the north of the North Island (Table II), when the extreme frequency increased significantly. A shift around 1977 in the southwest of the South Island also occurred, with a significant increase in the extreme frequency. These features are clearly seen in the nationally averaged extreme frequency index for 24 stations (Figure 9). The index overall shows a decline of about 2 days from 1951 to 1998. There is an abrupt change indicated in 1978 to lower values, which is significant by the Pettitt test at the 95% confidence level. Copyright © 2001 Royal Meteorological Society

Int. J. Climatol. 21: 1437 – 1452 (2001)


1447

Figure 8. Trend in the frequency of daily rainfall exceeding the 1961– 1990 mean 95th percentile (extreme frequency) for rain days with rainfall totals of 1 mm or more, over the period 1951– 1996. The sign of the linear trend is indicated by +/ − symbols; bold indicates statistically significant trends (95%), with symbol indicating trends close to zero

Finally the dry spells index shows a pattern of decrease; sometimes significantly, in many areas over the period considered (Figure 10). The maximum number of consecutive dry days in many areas has decreased by between 2.5 and 4.5 days since 1951. Only in Gisborne, Wellington, Blenheim, Timaru and Dunedin have these increased. The change point analysis yielded no favoured date of change (Table II).

4. DISCUSSION AND CONCLUSIONS The nationally averaged mean, maximum and minimum temperature for New Zealand, calculated from 21 locations (Salinger et al., 1992a) over the 1951 – 1998 period show an increasing trend of 0.2, 0.1 and 0.4°C, respectively. The decreasing trend in the value and frequency of hot days for much of New Zealand is small, and is not consistent with the small warming observed in maximum temperatures. Only in the north and northeast of the North Island a non-significant increase in the frequency and value of hot days is observed. Copyright © 2001 Royal Meteorological Society

Int. J. Climatol. 21: 1437 – 1452 (2001)

1448


Table II. Year of occurrence of statistically significant change points in the (a) extreme intensity and (b) extreme frequency indices of daily rainfall, and (c) dry spells index for the sites analysed, at the 0.05 probability levela Site

Extreme intensity

Extreme frequency

Dry spells

Raoul Island Kaitaia Auckland Ruakura Rotorua New Plymouth Taihape Gisborne Napier Masterton Paraparaumu Wellington Nelson Blenheim Lincoln Timaru Dunedin Tara Hills Queenstown Ophir Hokitika Milford Sound Invercargill Chatham Island Campbell Island

1953

1950 1952 1957 1955 1973

1957 1947 1957 1976

1986

1973 1939 1948 1984

1954, 1978 1939

a

1973 1974 1984

1937 1967 1978 1977

1991 1966, 1977, 1992 1976

1948

1985 1943

1968 1980

1994

1975

1965

The years indicate that the change point is significant at the 0.05 probability level. The indicates that there is no significant change point detected in the index.

symbol

In contrast, the cold night index value (i.e. the 5th percentile of minimum temperature) has increased over most of the country, rising by 0.6 – 1.8°C. This, and the 10– 20 day decreases in the frequency of cold nights from the 1961 – 1990 average trend in the same direction to the increase recorded in minimum temperatures. These trends are consistent with those observed globally towards fewer extremes of low temperature. Between 1950 and 1993, on average, night time minimum air temperatures over land increased by 0.2°C per decade, about twice the rate of increase in day time maximum air temperatures of 0.1°C per decade (IPCC, 2001). The pattern of change of 25°C days and frost days for the 1930 – 1998 period link well with the changes in annual and seasonal circulation over the New Zealand region identified by Salinger and Mullan (1999). In the second period (1951– 1975) the annual frequency of 25°C days increased the most in the south and east of the South Island, and in the west of the North Island, compared with the first period (1930–1950). Decreases occurred in the northeast of the North Island. This contrasts with a different pattern of change in the frequency of frost days below 0°C. These show a decrease in all areas except inland south eastern areas of the South Island. The changes in these exceedances of temperature thresholds are consistent with the change in circulation to more airflow from the east and northeast between the first and second periods. The increases in 25°C days in the North Island occur in those areas most sheltered from strengthened onshore flow in summer, i.e. western and southwestern parts of the North Island. In contrast, areas exposed to the enhanced airflow show a decrease in frequency. For the South Island the general increase in 25°C days, more marked in the east and south reflects the change to more northerly quarter airflow and anticyclonic conditions. The frost day changes also match the circulation shifts. Associated with the airflow from warmer sources there is a widespread decrease in frost days. However, the inland South Island areas show an increase in frost frequency with the higher frequency of Copyright © 2001 Royal Meteorological Society

Int. J. Climatol. 21: 1437 – 1452 (2001)


1449

Figure 9. Trend in nationally averaged frequency of daily rainfall exceeding the 1961– 1990 mean 95th percentile (extreme frequency) for rain days with rainfall totals of 1 mm or more, over the period 1951– 1996, for 24 stations

anticyclones east of New Zealand (Trenberth, 1976), and lighter winds and less winter cloud cover leading to more radiation frosts (Salinger et al., 1996). Changes in the last period (1976– 1998) to the 25°C and frost days are also consistent with circulation changes. Increases in 25°C days are detected in the east of the North Island and in the northeast of the South Island, and decreases occur elsewhere, which relate well to more prevalent airflow from the west and southwest. The circulation changes accompanied the higher incidence of El Nin˜ o events. The pattern of frost day change also reflects the circulation switch: no change or increases are detected in the west of the South Island, and at higher elevations cooled by the increased westerlies. Other areas show frost day decreases associated with the regional warming in surface land and marine temperatures (Folland and Salinger, 1995). Non-significant increases in the frequency of hot days over the period 1951 –1998 only occur in the north and northeast of the North Island, areas that became more anticyclonic during the last circulation phase (1976– 1998). Other areas show a non-significant decrease in hot day frequency and values, accompanying the higher frequency of westerly winds in that period. The regional nature of these changes in hot days shows a likely link with the interaction between circulation and the country’s orography (Salinger, 1980a). The general decrease in the frequency of cold nights in most locations reflects the rise observed in regional minimum surface temperatures. The rainfall indices display no trend towards an increase in extreme intensity or frequency foreshadowed by global models for climate warming (IPCC, 2001). These show rather the opposite over much of New Zealand, with significant decreases in frequency, and at a few locations, intensity, over the 1951–1996 period. However, the rise in mean temperature over the period considered has been slight (0.2°C). Instead, a regional response in 1-day rainfall extremes and dry spell trends is evident. Although there are few places where the extreme intensity showed a significant change, the non-significant trends were for decreases in areas sheltered from the west. Increases occur in those locations that are exposed to the prevailing west to southwesterly circulation. The extreme frequency index shows the same zonal Copyright © 2001 Royal Meteorological Society

Int. J. Climatol. 21: 1437 – 1452 (2001)

1450


Figure 10. Trend in the annual maximum number of consecutive dry days (dry spells) where the rainfall total is less than 1 mm, over the period 1951– 1996. The sign of the linear trend is indicated by +/ − symbols; bold indicates statistically significant trends (95%), with symbol indicating trends close to zero

pattern of trends, but with more sites showing a significant change. The largest increasing trends in the extreme frequency index occur in the west and south of the South Island. A change point date identified in the extreme frequency index in these regions around 1978 is consistent with the period when a shift in circulation occurred to enhanced westerlies. These results are in accord with the strong sensitivity New Zealand rainfall changes show to the interaction between circulation and orography (Salinger, 1980b). Finally dry spells generally show a decrease in length except in the east of the South Island, Gisborne and Wellington. The trends in the three rainfall indices are regional in nature, and show a west/east difference in trends. Daily air masses from the westerly climate regime affect New Zealand; intensities and frequencies decrease in sheltered areas and increase in exposed areas. The decrease in dry spell length may indicate the more frequent rain producing synoptic disturbances that occur in an enhanced westerly circulation regime, where showery rainfall events are more predominant. This paper is the first national analysis of trends in indices of temperature and rainfall extremes and thresholds for New Zealand, over common periods of the observed record. The most important results are that there are regional responses evident in the extreme index trends, and that significant changes in some Copyright © 2001 Royal Meteorological Society

Int. J. Climatol. 21: 1437 – 1452 (2001)


1451

extreme parameters are consistent with changes in atmospheric circulation over the New Zealand region. The cold night changes have practical significance in that the frequencies of cold extremes over most of the locations in this study show a decline of between 10 and 20 days a year. However, the 46 –68-year period of analysis is rather short for measuring climate trends, particularly for those involving rainfall. Climate station closures during the 1990s are of concern as it is crucial that monitoring stations are continued into the future so as to detect future changes in extreme indices. Such closures have been highlighted by IPCC (2001) as priority areas of action. The changes in agreement with global climate model projections are the decrease observed in cold nights and decrease in frost day frequency (IPCC, 2001). The responses of New Zealand climate extremes to changes in atmospheric circulation demonstrate the marked sensitivity of local climates to small circulation changes. The coherent but distinct responses demonstrate the important role New Zealand’s orography has interacting with shifts in the regional circulation. These modify the expression of temperature and rainfall extremes. This is a bountiful area for future investigations involving further stations and over longer time periods. Results from New Zealand confirm the trends towards higher minimum temperature extremes and fewer frost days observed in other parts of the globe. In this remotely monitored region of the Southern Hemisphere mid latitudes, the response of the minimum temperature extremes is consistent with increases in marine and land surface mean temperatures observed in New Zealand and the surrounding oceans. ACKNOWLEDGEMENTS

This research was supported by the New Zealand Foundation for Research, Science and Technology, contract No. CO1628. Our thanks go to Isabelle Leleu of Meteo-France who developed the software for analysing indices of rainfall extremes. Our thanks go to Drs A.B. Mullan and J. Renwick who provided useful comments in the drafting of this manuscript. REFERENCES Ericksen NJ. 1986. Creating flood disasters? Water and Soil Miscellaneous Publication No. 77. Folland CK, Salinger MJ. 1995. Surface temperature trends in New Zealand and the surrounding ocean, 1871– 1993. International Journal of Climatology 15: 1195 –1218. Groisman P, Karl T, Easterling D, Knight R, Jamason P, Hennessy K, Suppiah R, Page C, Wibig J, Fortuniak K, Razuvaev V, Douglas A, Førland E, Zhai P. 1999. Changes in the probability of extreme precipitation: Important indicators of climate change. Climatic Change 42: 243–283. IPCC. 2001. In Climate Change 2001: The Scientific Basis, Houghton JT, Ding Y, Griggs D, Noguet M, van der Linden P, Dai X, Maskell K, Johson CA (eds). Cambridge University Press: Cambridge. Karl TR, Nicholls N, Ghazi A. 1999. CLIVAR/GCOS/WMO workshop on indices and indicators for climate extremes. Climatic Change 42: 3 – 7. Katz RW, Brown BG. 1992. Extreme events in a changing climate: Variability is more important than averages. Climatic Change 21: 289 – 302. Manton MJ, Della– Marta PM, Haylock MR, Hennessy KJ, Nicholls N, Chambers LE, Collins DA, Daw G, Finet A, Gunawan D, Inape K, Isobe H, Kestin TS, Lefale P, Leyu CH, Lwin T, Maitrepierre L, Ouprasitwong N, Page CM, Pahalad J, Plummer N, Salinger MJ, Suppiah R, Tran VL, Trewin B, Tibig I, Yee D. 2001. Trends in extreme daily rainfall and temperature in Southeast Asia and the south Pacific: 1961–1998. International Journal of Climatology 21: 269 – 284. Nicholls N, Gruza G, Jouzel J, Karl TR, Ogallo LA, Parker DE. 1996. Observed climate variability and change. In Climate Change 1995: The Science of Climate Change, Houghton JT, Meira Filho LG, Callander BA, Harris N, Kattenberg A, Maskell K (eds). Cambridge University Press: Cambridge; 133–192. Pettitt AN. 1979. A non-parametric approach to the change point problem. Applied Statistics 28: 126 – 135. Plummer N, Salinger MJ, Nicholls N, Suppiah R, Hennessy KJ, Leighton RM, Trewin B, Page CM, Lough JM. 1999. Changes in climate extremes over the Australian region and New Zealand during the twentieth century. Climatic Change 42(1): 183 – 202. Rhoades DA, Salinger MJ. 1993. Adjustments of temperature and rainfall records for site changes. International Journal of Climatology 13: 899 – 913. Salinger MJ. 1980a. New Zealand climate: 1. Precipitation patterns. Monthly Weather Re6iew 108: 1892 – 1904. Salinger MJ. 1980b. New Zealand climate: 2. Temperature patterns. Monthly Weather Re6iew 108: 1905 – 1912. Salinger MJ. 1981. New Zealand climate: The instrumental record. PhD Thesis, Victoria University of Wellington. Salinger MJ. 1994. Climate variability, agriculture and forestry. World Meteorological Organisation Technical Note No. 196, Geneva, Switzerland. Salinger MJ, Allan R, Bindoff N, Hannah J, Lavery B, Lin Z, Lindesay J, Nicholls N, Plummer N, Torok S. 1996. Observed variability and change in climate and sea level in Australia, New Zealand and the South Pacific. In Greenhouse: Coping with Climate Change, Bouma WJ, Pearman GI, Manning M (eds). CSIRO Publishing: Melbourne; 100– 126. Copyright © 2001 Royal Meteorological Society

Int. J. Climatol. 21: 1437 – 1452 (2001)

1452


Salinger MJ, McGann R, Coutts L, Collen B, Fouhy E. 1992a. Temperature trends in New Zealand and outlying islands. New Zealand Meteorological Service, June. Salinger MJ, McGann R, Coutts L, Collen B, Fouhy E. 1992b. Rainfall trends in New Zealand and outlying islands. New Zealand Meteorological Service, June. Salinger MJ, Mullan AB, Porteous AS, Reid SJ, Thompson CS, Coutts LA, Fouhy E. 1990. New Zealand climate extremes: Scenarios for 2050 AD. Report for the New Zealand Ministry for the Environment, May. Salinger MJ, Mullan AB. 1999. New Zealand climate: Temperature and precipitation variations and their links with atmospheric circulation 1930– 1994. International Journal of Climatology 19: 1049 – 1071. Tarhule A, Woo M. 1998. Changes in rainfall characteristics in northern Nigeria. International Journal of Climatology 18: 1261 – 1271. Trenberth KE. 1976. Fluctuations and trends in indices of the southern hemisphere circulation. Quarterly Journal of the Royal Meteorological Society 102: 639–653.


Int. J. Climatol. 21: 1437 – 1452 (2001)