Weather and Climate (2000) 20: 29-46 29

THE 1997-99 AND HISTORIC DROUGHTS IN NORTH

Zhijia (Harry) He Otago Regional Council, Dunedin

ABSTRACT INTRODUCTION The 1997-99 drought was considered a Droughts have received increasing major event in North Otago, with record attention in New Zealand recently due to the breaking high temperatures, low rainfall and impact of the 1997-99 drought on the economy low river flows. The driest area was centred of the country. Historic droughts in North in the coastal downlands, including , Otago have been analysed previously. Borlase the middle and lower catchment of the (1976) applied Herbst's drought evaluation and Waianakarua River, and the technique to evaluate rainfall data in the lower catchment of the Shag River. North Otago Downlands based on the Oamaru Herbst's drought evaluation technique rainfall record for the period of 1867-1974 (Herbst et al. 1966) was applied to 132 years (Figure 1). In total, 58 droughts of varying of monthly rainfall record at Oamaru to re- severity were identified. evaluate historic drought events with 75 Several techniques were used by the Otago droughts identified. The 1997-99 drought was Catchment Board and Regional Water Board ranked by the Herbst's technique as the 3rd (1983) to evaluate the Oamaru rainfall record most severe event in North Otago in the past for the period of 1867 — 1982. These techniques 132 years and the 2"d most severe drought include the Herbst's method, the method of during the 1980's and 1990's. The drought has deciles, the technique of cumulative departure been particularly lengthy but not particularly from the mean and a mathematical filter intense. analysis. A total of 65 separate drought events Low flow was significantly more persistent during this period were identified using in the Kakanui River during the 1997-99 Herbst's method. The most severe drought drought compared with other events in the occurred between 1963 and 1968 with a 1980's and 1990's. However, the extreme low duration of 53 months. Significant 10-11 and flows were not as low as those of extended 6-7 year rainfall cycles were demonstrated by droughts. Low flows were generally more the mathematical filter analysis, which were sustained in the Kakanui River catchment similar to cyclic patterns occurring in rainfall than in the Shag River catchment. records throughout New Zealand. However, There was generally no strong relationship different methods usually give different between monthly rainfall in North Otago and results even though they are used on the same the El Nino Southern Oscillation. However, record. Bee (1984), using outputs based on there appeared to be a general negative Herbst's method of the Otago Catchment relationship between spring SOI and summer Board and Regional Water Board, compared rainfall, with low rainfall relatively more likely the rainfall method of Herbst and the water during La Nina events. During El Nino events, balance method of Coulter (1973) and the possibilities of receiving either high or low concluded that Herbst's method generally rainfall were similar, but El Ni -no springs were overestimated the frequency of drought in slightly more likely to bring high rainfall to North Otago and identified longer droughts North Otago than springs of other episodes. than the water balance method. Otago 30 Historic Droughts in North Otago

Regional Council (1991) used a three year 1997-99 drought with historic events. The moving average technique to demonstrate the method was chosen over other commonly used variability of rainfall within the Shag River techniques in order to be able to compare the catchment but concluded that no cyclic rainfall results with previous studies. The spatial and pattern existed to allow predictions to be temporal distributions of rainfall during the made. 1997-99 event were also briefly examined. The relationship between El Nino / La Nina Streamflow was assessed to identify the and precipitation and streamflow values is impact of droughts on rivers. Low flow well documented in New Zealand. For regimes of the Kakanui River and Shag River example, research has shown that El Nino were compared (Figure 1). Relationship generally brings a phase of more westerly and between ENSO phenomenon and rainfall southwesterlies, and makes it drier in the variations at Oamaru and Palmerston were north and east, while La Nina promotes investigated, particularly between El Nifio easterly and northeasterly winds, with Southern Oscillation phases in spring and conversely wetter conditions in the north and rainfall variation during the following east, and drier western and southern regions summer. (Basher, 1998). A good correlation has been identified between below average rainfall in THE STUDY AREA the lower North Island and the La Nina phase (Gordon, 1986). A general positive relationship The study area extends from the Waitaki between summer rainfall in Christchurch and River in the north to the Pleasant River in the SOI was also found (Basher, 1998). south and includes the catchments of the Connection between spring (September- Kakanui, Waianakarua and Shag Rivers. November) SOI and summer (December- Figure 1 shows the study area, while Table 1 February) Clutha lake inflows (location of lists rainfall and current surface water Clutha lakes: S Lat 44°45', E Long 168°50') has monitoring stations within the study area. been identified by NIWA (Moss eta]., 1994 and The Kakanui River has a total catchment McKerchar, 1994). The studies indicated that area of 894 km2, which consists mostly of river low values of summer mean inflows are valley and rolling hills or downlands of less relatively unlikely when a moderate spring El than 600 m elevation. About a quarter of the Nino is experienced, but are relatively area is mountainous, reaching heights of some common when a moderate La Nina spring 1,640 m. Flow at Clifton Falls Bridge is proceeds. deemed as natural flow, since there is minimal Gordon (1985), in his study of effects of El abstraction above the station. Flow in the Nino Southern Oscillation on New Zealand lower catchment is, however, subject to the weather, established a relationship between influences of irrigation abstraction and seasonal rainfall and contemporary SOT. The groundwater-surface water interaction. Mean study indicated that there is a significant annual flow in the Kakanui River (Clifton Falls seasonal variation of SOI and rainfall Bridge) is 3.34 m's-'. correlation, with weak negative correlation The Shag River has a catchment area of 544 observed in Otago during spring and summer, km' and an elevation range of 1528 m. stronger negative correlation during autumn Comparatively low elevation hill country and positive correlation during winter. The forms the south and western boundary of the study also indicated that there is lag catchment. Flow at The Grange is influenced correlation between the SOT and subsequent by irrigation abstraction, particularly during weather elements in New Zealand. September to April. Flow in the lower For North Otago, the relationship between catchment is also affected by irrigation SOT and droughts was examined by Bee abstraction and groundwater-surface water (1984). This study showed that drought in interaction and the river sometimes dries up North Otago does not compare well with the naturally. Mean annual flow in the Shag River Southern Oscillation Index although most (The Grange) is 1.85 m3s-1. drought events occur at least partially during Most of the Kakanui river catchment a negative index phase. comprises greywackes, subschists and schists, This paper applies Herbst's method to while the downland topography is the most re-evaluate drought events and compares the distinctive of the region and is associated Historic Droughts in North Otago 31

• River level/flow station

A Flow gauging station

• Rainfall station

0 1 2 3 4 5

Kilometres

WAIKOURA U A

1 U AIR

11 I AT CLIFTON FALLS BR LDE HORPE DO WINDSOR PARK

NEWHAVEN

AMARU ,11/ FUSHIA CREEK • Oamaru KAURU HILL DVIEW AT MILL DAM

efr-THE DASHER

NS Walanakarua

DEEPDELLOK-Ai GOLDEN POINT • • HAG AT THE GRANGE GLENDALE 1iITrERS CREEK

HAG AT CRAIG • STONEMAN ENTHORN OODWOOD PUMP PALMERSTON Palmersto SHEY PARK CENTREWOOD

Figure 1. Study area, with location of rainfall and current surface water monitoring stations. 32 Historic Droughts in North Otago

Station Station Name Station Type Record Catchment Area No Start End (km') 510001 Hilderthorpe Rainfall 1983 1998 140981 Waikoura Rainfall 1964 141901 Oamaru Airport Rainfall 1941 1985 141902 Oamaru AWS Rainfall 1982 15005A Dome Hills Rainfall 1975 150062 Fuchsia Ck Rainfall 1953 150072 Newhaven Rainfall 1952 150082 Elderslea Rainfall 1893 1908 150083 Windsor Park Rainfall 1892 1908 150085 Enfield Rainfall 1977 150091 Oamaru Rainfall 1867 1965 150161 The Dasher Rainfall 1936 150172 Kauru Hill Rainfall 1890 1930 150241 Islay Downs Rainfall 1969 150341 Glendale Rainfall 1961 150382 Trotters Ck Rainfall 1908 150431 Nenthorn Rainfall 1969 150471 Palmerston Rainfall 1969 150472 Bushey Park Rainfall 1908 1979 150571 Centrewood Rainfall 1912 1997 71703 Kakanui R. at Clifton Falls Br. River level/flow 1981 286 71713 Kakanui R. at Mill Dam River level/flow 1989 539 72603 Shag R. at The Grange River level/flow 1989 319 72604 Shag R. at Goodwood Pump Flow gauging 72627 Deepdell Ck at Golden Point Weir River level/flow 1985 72690 Shag R. at Craig Rd River level/flow 1993 428 Table 1. Rainfall and current surface water monitoring stations mainly with tertiary limestone deposit. In the significance of the frequency of drought Shag River, much of the catchment is occurrence in the North Otago region has often underlain by schist, while in the valley floor been referred to in earlier reports (Bee, 1984, quartz conglomerates, sandstones, mudstones Borlase, 1976 and Waugh et al., 1997). These and limestone occur and are interspersed with droughts have resulted in irrigation bans as volcanic outcrops. well as livestock and financial losses to The Kakanui, Waianakarua, Shag, Pleasant farmers. and other catchments in the area are naturally subject to low flows, particularly between late THE 1997-99 DROUGHT spring (November) and mid-autumn (April), primarily due to climatic factors. North Otago The 1997-99 drought was a significant event is not as dry as some inland areas but still with widespread impact reported in New experiences a relatively low rainfall Rainfall Zealand, including extreme moisture stress in - ranges from less than 600 mm per annum near Marlborough, North Canterbury, parts of the Oamaru to in excess of 1000 mm in the Wairarapa and Hawke's Bay (Basher, 1998). Kakanui Mountains. The majority of the Record breaking high temperatures, low downlands have rainfall in the order of 600 to rainfall and river flow were also experienced 700 mm per annum. in Central Otago and the east coast from The area is sheltered from the predominant Oamaru to Dunedin. westerly rainstorms by mountain ranges but In North Otago, below average rainfall is exposed to storms from the south and started in May 1997. The drought intensified easterly quarters. The mean annual rainfall during January 1998, with all stations is low and highly variable and so it is recording record breaking low rainfall (Table occasionally subject to long dry periods. The 2). Other months with particularly low rainfall Historic Droughts in North Otago 33

Jun-97 • Rainfall between 9 and 14 mm, 70% below average for June. Sep-97 • Average flow in the Kakanui River at Clifton Falls was 1.31 m3/s, lowest for September since 1981. Jan-98 • Rainfall between 3 and 24 mm, 70 — 95% below average for January, lowest for January on record. • Average flow in the Kakanui River at Clifton Falls was 0.80 m3/s, 68% below average for January, second lowest for January. Feb-98 • Average flow in the Kakanui River at Clifton Falls was 0.48 m3/s, 89% below average for February, second lowest for February. Mar-98 • Average flow in the Kakanui River at Clifton Falls was 0.52 m3/s, 76% below average for March, lowest for March. Jan-99 • Average flow in the Kakanui River at Clifton Falls was 0.54 m3/s, 79% below average for January, lowest for January. • Average flow in the Shag River at The Grange was 0.06 m3/s, 96% below 9-year average for January, lowest for January. Feb-99 • Rainfall between 8 and 13 mm, 73 — 85% below average for February. • Average flow in the Kakanui River at Clifton Falls was 0.39 m3/s, 85% below average for February, lowest for February. • Average flow in the Shag River at The Grange was 0.04 m3/s, 95% below 9-year average for February, lowest for February. • Lower reach of the Shag River basically dried up. May-99 • Rainfall between 3 and 19 mm at most stations, 80 — 97% below average for May. Table 2. Low rainfall and low river flows during the 1997-99 drought were February, May and August, with 75 — 95% Figure 3 shows the total rainfall distribution below average recorded for the months. for May 1997 through May 1999 over North October 1998 and July 1999 were the only Otago, presented as the departure from months with the whole area receiving well average, based on stations with records of 20 above average rainfall. The drought years or longer. The driest area was centred terminated in January 2000 with 90 — 130 mm in the coastal downlands, including Oamaru, rain received in the area, followed by 50 — 110 the middle and lower catchment of the mm in March and 65 — 95 mm in April. Kakanui and Waianakarua Rivers. The lower During the dry seasons of 1998 and 1999, catchment of the Shag River was also dry. river flows in the Kakanui River at Clifton Falls and the Shag River at The Grange were HISTORIC DROUGHT ANALYSIS significantly below the mean 7-day annual low flow (Q7m), with flows closely oscillating about Methods the 10 Year 7-day annual low flow (Q, ,o) for a Long-term monthly rainfall data for considerable period of time. In the lower Oamaru (1867-1999) and Palmerston (1908- Kakanui and Shag Rivers, river flow dropped 1999) were available for this study. The 132 below consent limits set by the Otago Regional year rainfall data for Oamaru was obtained Council, despite voluntary irrigation by combining records from Oamaru (150091, restriction and irrigation bans. The lower 1867-1942), Oamaru Airport (141901, 1943- Shag River dried up in February 1999 (Table 1982) and Oamaru Airport Automatic Weather 3, Figure 2). Station (141902, 1983-1999). The distance 34 Historic Droughts in North Otago

4000 between Oamaru and OamamAirport is about

3000 15 km The period with overlapping duration for Oamaru and Oamaru Airport is about 25 2000 years. Monthly rainfall patterns between the two stations were very similar, with mean 1000 annual rainfall of 573 mm and 528 mm A ! respectively. Records from the two stations 01-APR-97 OCT-97 APR-98 'OCT-98 'APR-99 'OCT-99 were combined to demonstrate the long-term Kakanui at Clifton Falls Br rainfall pattern. Gaps in the record before A: C)7.m (562 Us) B: Q7,10 (325 Us) 1997 were filled by averaging data from three stations, Kauru Hill (150172), Windsor Park (150083) and Elderslea (150082) (Otago Catchment Board and Regional Water Board 1983), while gaps in the record after 1997 were filled using data from the Hilderthorpe station (510001) (Figure 1, Table 1). The 91 year rainfall record for Palmerston was the combination of records from Bushey 01- R-97 OCT-97 APR- 8 OCT-98 APR-99 OCT-99 Park (150472, 1908-1975) and Palmerston Kakanui at Mill Dam (150471, 1976-1999). Bushey Park and A: Consent limit (250 Us) Palmerston are about 5 km apart, with mean annual rainfalls of 638 mm and 648 mm 500 respectively. Rainfall records for the two stations overlap by about seven years. 1000 Monthly totals for the two stations during the overlap period are generally the same. Gaps 0 500 in the record were filled using data from

, A Centrewood (150571) or Trotters Creek 0 • , B (150382). 01-AP R-97 OCT-97 APR-98 OCT-98 APR-99 OCT-99 Monthly cumulative departure of recorded Shag at The Garage rainfall from average was plotted for Oamaru A: QT. (198 Us) B: 07,10 (52 Us) and Palmerston to examine trends in the long-

1500 term rainfall record. Herbst's drought evaluation technique was applied to the

1000 rainfall data from Oamaru for drought severity, intensity and duration. These drought parameters for the 1997-99 event were then compared with historic events.

A The Herbst's technique is based on the -97 OCT-97 APR-98 OCT-98 APR OCT-99 following premises: Shag at Craig Road • Farming in any region is adapted to the prevailing climate pattern. Only rainfall A: Consent limit (110 Us) deficits that exceed the average deficit for any particular month should prove harmful to farming operations. Benefit to vegetation of above-average rainfall persists for some time after the rain actually falls due to storage of moisture in the soil. Conversely, vegetation does not recover immediately

0- -97 OCT-97 APR 9 OCT- APR-99 OCT-99 after a drought is broken. Depending on the amount of rainfall received in a Shag at Goodwood Pump particular month, a carry-over effect is A: Consent limit (28 Us) passed on to the next month. Figure 2. River flow in the Kakanui and Shag Rivers. Historic Droughts in North Otago 35

Kakanui River at Clifton Falls Bridge Shag River at The Grange • Q1(1998)•,_, 375 Lis (rp = 3 years) • Qty(1998)* • 45 Lis (rp = 7 years) '•11(1999)• 263 Lis (rp 5 years) '%(1999): 25 Lis (rp = 20 years) • Qd(1998):. 377 Lis (rp = 3 years) Qd(1998)•. 49 Us (rp = 7 years) Qd(1999)* 291 L/d (rp = 7 years) Qd(1999 • Q7d(1998): 393 Lis (rp = 5 years) • 0-:7(1(1998). 55 Us (rp =-- 8 years) 38 Lis (rp = 17 years) Q7c1(1999)• 315 Lis (rp = 12 years) Q7d(1999)* • Q30(1(1998): 422 Lis (rp -,- 7 years) •64Q30(1(1998): Lis (rp -, 9 years) • Qi(1998):Q30d(1999)• 338 Lis (rp = 16 years) Q305(1999): 50 Lis (rp = 14 years) • No. of days below Q7,m: 180 • No. of days below Q7,m: 210 • Qd(1998)••No. of days below Q7,10: 8 • No. of days below Q7,10: 30 Kakanui River at Mill Dam Shag River at Craig Road 132 Lis • Qi(1998)• 29 Lis

Qi(1999)*. 153 Lis Qi(1999): 18 Lis 146 Lis Qd(1998)• 50 Lis Qd(1999)• 178 Lis Qd(1999)• 19 Lid Q7d(1998)• 266 Lis Q7d(1998): 54 Us Q7c1(1999•)• 213 Lis Q7d(1999): 25 Lis Q30(1(1998):• 310 Lis Q3041(1998. 67 Lis Q30c1(1999)• 312 Lis Q30c1(1999)• 28 Lis No. of days below consent limit 250 Lis: 30 No. of days below consent limit 110 Lis: 113 (4: minimum instantaneous flow Qd: minimum daily average flow minimum 7-day average flow

Q30d minimum 30-day average flow Q, m: mean 7-day annual low flow. 562 L/s for the Kakanui River at Clifton Falls, 198 L/s for the Shag River at The Grange Q • 10 year 7-day annual low flow. 325 L/s for the Kakanui River at Clifton Falls, 52 L/s for the Shag River at The Grange rp: return period

Table 3. Flow analysis during the 1997-99 drought

A single month can constitute a drought RESULTS only if no rain falls during the month of highest mean rainfall. Figure 4 illustrates cumulative monthly Drought is terminated either by a rainfall departure from average for Oamaru continuous period of slightly above and Palmerston. The five-year moving means average rainfall or a few successive are also included. While the cumulative months of abnormally high rainfall. rainfall departures from average for the month The Herbst's technique can be divided into do not indicate any apparent cyclic pattern, the following steps (Herbst et al., 1966, Bee, Figure 4 illustrates a period of increased 1984): rainfall between 1942 and 1947 at Oamaru. A • Calculation of effective rainfall and period of lower than average rainfall is monthly surplus or deficit indicated during the period 1948 to 1992, • Tests of onset and termination of drought followed by a period of generally average • Calculation of drought index rainfall until the present day. 36 Historic Droughts in North Otago

-..-•••••" -35% Total rainfall departure contour

0 35 Station with rainfall departure in %

Figure 3. Total rainfall departure from average for the 1997-99 drought (May 1997 — May 1999). Historic Droughts in North Otago 37

5000

4000 .t.•

WAw , t 2 3000

2000 0. 70 A 1 000 '1ikt. L_AeL , A TI-a 2,

• -low q '

0 -2000 7-- COo CO0) CO COco C7)t, c‘.10 LCC)0 1- CO CI (5) CO 0 00) CO CO 7--0 0 LO CO csJ c‘i cv a) co 0LO 0 CO 7-- 7-- C.73 CO 0) 0) co CO CO CO CO CO CO CO Cr) a) c7) co a) a) o) o) o) o) a) a) 0) 0) 0) 0) 0) 0) CM Year

- - Oamaru -Oamaru 5-yr moving mean Palmerston -Palmerston 5-yr moving mean

Figure 4. Cumulative rainfall departures from average at Oamaru and Palmerston, with 5-year moving means.

Figure 5. Drought-severity indices at Oamaru during 1867 — 1999. 38 Historic Droughts in North Otago

The figure also shows general north of Dunedin, low rainfall is correspondence in peaks and troughs between a well-known characteristic. Annual and Oamaru and Palmerston. However, rainfall seasonal rainfall is highly variable so this area patterns of the two stations are different from is occasionally subject to long dry periods the late 1960's to the early 1990's when a trend (Otago Catchment Board and Regional Water of below average rainfall was generally Board, 1983, Otago Regional Council, 1991). observed at Oamaru, while above average Drought analysis indicates that droughts were rainfall was recorded at Palmerston from the common phenomena in North Otago, late 1960's to the early 1980's and average particularly droughts with short duration, rainfall since the early 1980's. This pattern is with 22 out of 75 droughts (29%) having more obvious in the five-year moving mean duration not longer than six months and 24 plots. droughts (32%) from seven to 12 months. A total of 75 droughts were identified during Droughts longer than 24 months were rare, 132 years by the Herbst's technique, with with only four of the 75 events identified in individual events ranging from two to 37 this category (Figure 5). months (Figure 5). Drought events identified Intense droughts were generally of short in this study for the period 1867-1982 were duration, with the eight most intense events generally the same as the events identified by lasting less than eight months. The most the Otago Catchment Board and Regional intense drought, drought 1983, lasted only two Water Board (1983) using the 116-year record. months. On the other hand, six of the ten least The most severe drought identified by the intense events lasted more than 15 months. Otago Catchment Board and Regional Water Severe droughts were generally of long Board was the drought starting in October duration, but with varied intensities. Of the 1963 and terminating in February 1968, with 20 most severe droughts, 17 of them were a duration of 53 months. However, with the found to have duration longer than 15 months longer rainfall record, this event was identified (Table 4). On the other hand, of the 20 least as a series of three droughts: October 1963 to severe droughts, the longest duration for each February 1965, June 1965 to September 1966 individual event is seven months. and February 1966 to February 1968. The 20 most severe droughts fall at an Previous studies indicated that in North average interval of seven years (1871-72, 1877- Otago, as over much of the east coast of the 79, 1880-81, 1889-92, 1896-98, 1906-07, 1909- Drought Duration Duration Intensity Intensity Severity Severity Total Normal Total rain as rank rank rank rain total rain % of average month mm mm mm mm 1889-92 37 1 19 42 710 1 1188 1677 71 1988-90 29 3 21 29 602 2 883 1307 68 1997-99 32 2 18 50 590 3 1056 1437 73 1896-98 23 8 25 16 564 4 631 1041 61 1984-86 24 6 23 21 551 5 659 1084 61 1914-16 22 9 24 18 522 6 595 990 60 1906-07 18 13 26 9 473 7 417 810 51 1958-59 18 15 24 17 439 8 480 791 61 1981-83 20 11 20 35 398 9 608 911 67 1923-24 16 20 25 15 395 10 377 708 53 1909-11 24 5 16 63 395 11 839 1084 77 1968-69 15 25 26 10 393 12 365 671 54 1972-74 17 19 23 20 393 13 535 790 68 1947-49 25 4 16 67 391 14 928 1133 82 1877-79 23 7 16 65 371 15 882 1039 85 1880-82 17 17 20 33 343 16 505 766 66 1963-65 17 18 19 41 327 17 559 784 71 1936-37 20 10 15 69 309 18 653 885 74 1871-72 15 23 19 40 291 19 529 673 79 1966-68 15 24 18 54 272 20 526 693 76 Table 4. Twenty most severe droughts during the last 132 years Historic Droughts in North Otago 39

Drought 1980-81 1981-83 1984-85 1986-87 1988-90 1992-93 1994-95 1997-99 Onset Jul-80 Aug-81 Jan-84 Dec-86 Mar-88 Dec-92 Aug-94 May-97 Termination Jun-81 Apr-83 Jan-86 Dec-87 Aug-90 Nov-93 Sep-95 Jan-00 Duration (month) 11 20 24 12 29 11 13 32 Duration_rank 36 11 6 32 2 37 29 2 Intensity (mm) 23 20 23 17 21 17 13 18 Intensity_rank 24 36 21 61 29 60 73 50 Severity (mm) 248 398 551 201 602 184 173 590 Severity_rank 25 9 4 35 2 37 41 3 Total Rain (mm) 293 608 659 435 883 378 500 1056 Normal total rain (mm) 497 911 1084 542 1307 494 582 1437 Total rain as % of normal 59% 67% 61% 80% 68% 76% 86% 73% Table 5. Major droughts during the 1980's and 1990's. 11, 1914-16, 1923-24, 1936-37, 1947-49, 1958- Flow duration curves of the Kakanui and 59, 1963-65, 1966-68, 1968-69, 1972-74, 1981- Shag Rivers during the 1997-99 drought were 83, 1984-85, 1988-90 and 1997-99), with a 2— compared with those during droughts in the 13 year range between the starting dates. 1980's and 1990's. Minimum flows during each The 1997-99 drought lasted 32 months, with event were calculated. Percentages of time an intensity of 18 mm and severity of 590 mm. flows fell below certain levels (mean 7-day The Herbst's drought evaluation technique annual low flow,,,) and 10 year 7-day ranks the 1997-99 drought as the 3rd most annual low flow (Q710) were also estimated in severe event at Oamaru in the past 132 years order to compare the impacts of droughts on and the 2nd most severe drought during the the two catchments. Return periods were 1980's and 1990's (Figure 5, Table 4, Table 5). estimated for minimum instantaneous flow The drought was a particularly lengthy but (Qi), minimum daily flow (Qd), minimum 7-day not particularly intense event, with its average flow (Q7 d) and minimum 30-day duration and intensity ranked 2nd and 50th average flow (Q304) at upstream stations with among the 75 historic droughts, and the 1st minimum or less irrigation abstraction effects. and 8th among droughts during the 1980's and The Log-normal method was used for this 1990's respectively. analysis. Specific discharge duration curves of the IMPACT OF DROUGHT ON two catchments during drought events in the RIVER FLOWS 1990's were plotted and recession constants of the two catchments were estimated using Methods low flow data during the three most recent Low flow records were available for the drought events to compare low flow regimes Kakanui River at Clifton Falls (Figure 1) for of the two catchments. Low flow data were most of the drought events during the 1980's selected after eliminating suspect data that and 1990's. Records were not complete during could be attributed to rainfall, irrigation the 1980-81 drought. Significant gaps (more influence and poor site conditions. The than 250 days) occurred in the flow records selected data were checked for consistency by during the 1986-87 drought. As such, data plotting semi-logarithmically with straight during this period could not be used. Flow lines fitted to the data, from which a recession records were continuous during the droughts constant was calculated for each set of data. of 1984-85, 1988-90 and 1997-99, although The recession constant for a catchment is the small gaps are present in the data set. average of those from all the data sets. In the Shag River at The Grange, low flow records were available during the droughts of RESULTS 1992-93, 1994-95 and 1997-99. Substantial gaps (17 days) appeared in records during Table 6 and Figure 6 indicate that the November 1994 and some small gaps also impact of the 1997-99 drought on streamflow appeared in other months during the 1994-95 was much more significant than droughts with event. Gaps of about 4.5 days appeared during shorter duration (the 1992-93 and 1994-95 January 1998. droughts), with all minimum flows in the 40 Historic Droughts in North Otago

Kakanui River at Clifton Falls Drought Minimum flows (L'S) Time flow below Q7m Time flow below Q7 10

Q, Qd Q7d Q30d Days % Days 1980-81 1981-83 399 428 454 568 6 36 0 0 1984-85 245 258 280 367 11 80 1 7 1986-87 - 1988-90 117 230 345 484 9 78 1 9 1992-93 489 568 604 659 1 3 0 0 1994-95 429 438 471 585 7 27 0 0 1997-99 263 291 315 338 19 182 1 10

Shag River at The Grange Drought Minimum flows (I/s) Time flow below Q7„,,, Time flow below Q7 10

% Days % Days Q, Qd Q7c1 Q30d 1980-81 1981-83 1984-85 1986-87 1988-90 1992-93 268 269 289 375 1994-95 153 155 165 215 10 39 0 0 1997-99 25 28 38 50 22 211 3 29 minimum instantaneous flow Qd: minimum daily average flow Q 7d: minimum 7-day average flow Qsod. minimum 30-day average flow Q7 m: mean 7-day annual low flow. 562 Us for the Kakanui River at Clifton Falls, 198 Us for the Shag River at The Grange Q zio: 10 year 7-day annual low flow. 325 Lis for the Kakanui River at Clifton Falls, 52 Us for the Shag River at The Grange

Table 6. Low flows during droughts in the 1980's and 1990's.

Kakanui and Shag Rivers during the 1997-99 highest for all the droughts in the last 20 years drought significantly lower, and percentages (Table 6, Figure 6). of time flow below Q7m and Q7 io significantly The impact of the 1997-99 drought on. higher, than those dnring the 1992-93 and streamflow was more severe than the 1981- 1994-95 droughts. 83 event, which had a duration of 20 months. Comparison of flows during the 1997-99 Minimum flows during the 1997-99 drought drought with those during extended droughts were much lower than those during the 1981- in the 1980's and 1990's (the 1981-83, 1984-85 83 period and the percentage of time below and 1988-90 droughts) showed that there was Q7,„ was much higher than that during the a lack of high flows in the Kakanui River 198-1-83 event. during the 1997-99 event compared with other Although general low flow was more droughts. Low flow was significantly more persistent during the 1997-99 event, the persistent, with the percentage of time flow extreme low flows were not as low as those below (47,m during the 1997-99 event the during the 1984-85 and 1988-90 events which Historic Droughts in North Otago 41 Sp 10000 Kakanui at Clifton Falls May 97 — Dec 99

8000 - Kakanui at Clifton Falls Dec 92 — Oct 93

Kakanui at Clifton Falls Aug 94 Aug 95 -(7,- 6000

u_-9- 4000 -

2000 -

0 0 10 20 30 40 50 60 70 80 100 Percent smaller

10000 Kakanui at Clifton Falls May 97 — Dec 99

8000 Kakanui at Clifton Falls Aug 81 — Mar 83

Kakanui at Clifton Falls Jan 84 — Dec 85 ::(-7f 6000 Kakanui at Clifton Falls Mar 88 — Jul 90

-a-.c' 4000

2000

0 10 20 '30 140 '50 '60 70 '80 100 Percent smaller

40 Kakanui at Clifton Fails May 97 — Dec 99

30 Kakanui at Clifton Falls Dec 92 — Oct 93 Kakanui at Clifton Falls Aug 94 Aug 95

FkIMAIM Shag at The Grange May 97 — Dec 99 20 Shag at The Grange Dec 92 — Oct 93

Shag at The Grange Aug 94 — Aug 95 10

10 20 30 40 50 60 70 80 100 Percent smaller

Figure 6. Flow duration curves in the Kakanui River and specific discharge (Ls lkm-2) flow duration curves in the Kakanui and Shag Rivers during droughts. 42 Historic Droughts in North Otago had a duration of 24 and 29 months in Shag River flow record during November respectively. Most of the minimum flows 1994, which was indicated as irregularities in during the 1997-99 event were higher than, the flow duration curve for the 1994-95 or close to, those during the 1984-85 and 1988- drought. The difference in low flow between 90 events, and the percentage of time flow was the two catchments is generally consistent. lower than the Q7,, was similar to those for Low flows were generally more sustained in the 1984-85 and 1988-90 droughts. the Kakanui River catchment than in the Shag Frequency analysis based on the Log- River catchment, with the 80 percentile normal method indicates that in the Kakanui specific discharge in the Kakanui River 1.47 River at Clifton Falls, the return period of the Is-likm2 greater than in the Shag River on minimum 30-day average flow during the 1999 average for the three droughts, and the 95 dry season was 16 years, the highest among percentile specific discharge 1.18 Is-Vkm2 all the minimum flows (Table 3). The greater than in the Shag River (Table 7, Figure minimum 30-day average flow during the 1998 6). dry season was 7 years, also the highest. In Low flow recession constants estimated the Shag River at The Grange, the return using data during the three most recent period of both the minimum instantaneous droughts also confirmed that the Kakanui flow and the minimum daily average flow River generally produced more sustained flows during the 1999 dry season were the highest than the Shag River, with the average (20 years). During the 1998 dry season recession constants for the Kakanui River however, the return period of the minimum 0.9400 per day and Shag River 0.9371 per day 30-day average flow was the highest, but only (Table 8). 9 years. This shows that during the 1997-99 drought, the critical duration in the Kakanui RELATIONSHIP BETWEEN RAINFALL River was long (30 days), while in the Shag AND EL NINO SOUTHERN River, the critical duration was short (1 day). OSCILLATION Comparison of specific discharge duration curves of the two catchments during the Methods drought events of the 1990's indicated the Monthly values of the Southern Oscillation difference in low flow regimes in the two Index (SOT) since 1933 were available from catchments. There are some substantial gaps the National Institute of Atmospheric and

Specific discharge (Lstkm-2) Drought 80 percentile 95 percentile Kakanui R. at Shag R. at Kakanui R. at Shag R. at Clifton Falls The Grange Clifton Falls The Grange 1992-93 2.99 1.54 2.19 1.17 1994-95 2.48 0.86 1.90 0.55 1997-99 1.80 0.47 1.34 0.18 Table 7. Comparison ofspecific discharge flow duration curves during droughts in the Kakanui and Shag River catchments.

Periods Recession constants Kakanui R. at Clifton Falls Shag R. at The Grange 7 - 17 Mar 93 0.9751 0.9616 5 - 24 Jun 93 0.9485 0.9303 16 - 25 Jul 93 0.9684 0.9742 11 - 19 Feb 95 0.9389 0.9227 21 Apr - 5 May 95 0.9694 0.9831 11 - 21 Jul 98 0.9016 0.9344 21 - 31 Dec 98 0.8784 0.8534 Average 0.9400 0.9371 Table 8. Low flow recession constants. Historic Droughts in North Otago 43

Water Research (NIWA). Rainfall data were RESULTS the Oamaru long-term records, obtained by combining records from Oamaru and Oamaru Figure 7 indicate that in North Otago, there Airport, and the Palmerston long-term is generally no obvious relationship between records, obtained by combining records from monthly rainfall variation and SOL Drought Bushey Park and Palmerston. events could occur when El Nino Southern Two-year moving average of the SOT and Oscillation was in either El Nino, La Nina or monthly rainfall departure from normal for neutral phases. For example, the 1981-83, the month were plotted for Oamaru for the 1988-90 and 1984-85 droughts were coincident period 1933-99 to examine any consistent with El Nino, La Nina and neutral phases relationship between SOT and monthly rainfall respectively. variation. Monthly rainfall during summer Examination of rainfall of all summer (December-February) for the period was also months during the period 1933-99, reviewed for variations. disregarding El Nino Southern Oscillation, Further analysis selected a group of El indicated that summer rainfall was highly Nino, La Nina and Neutral springs for the variable. In general, the distribution of period 1933-98. The selection criteria was a summer monthly rainfall was positively September-November average SOI of -5.0 or skewed, with long tail on the right side (Figure less ("El Nino"), +5.0 or more ("La Nina") and 8). As such, summer rainfall was relatively more than -5.0 but less than +5.0 ("Neutral"). more likely to be below average, with 62-64% Summer months following these springs were of the months recording below average also selected. The relationship between spring rainfall SOT and the number of summer months with The examination of all summer months high rainfall (rainfall departure from normal following particular El Nino I La Nina springs > 20%), low rainfall (rainfall departure from since 1933 indicates that the relationship normal < -20%), and normal or near normal between SOT in spring and rainfall variation rainfall (rainfall departure from normal E 20% during the following summer was generally and 3 -20%) was examined weak, particularly between El Nifio and

80

– 60

– 40

– 20 4-2 – 0

– -20 cL

– -40 71

– -60 cc

80 CY> co (0 CO 0 C71 CO CY) 1- CO co o o 1— CO CO CO CO srt. CO LC) LO 111 LC) CO CO N- N. co co co co co 0) 0) CY) C7) C3) cr) CT) 01 0> 0) 0) 0) 0) cr) 0) c() cr) a) 0) CC) C7>

— SOI —Rainfall departure from mean

Figure 7. Two-year moving average of SOL and monthly rainfall departure from average at Oamaru 44 Historic Droughts in North Otago

20

18

16

14

F 12 Z..' :21 10 .0co E' 8 cl.

6

4 I I

2

0 I II , 1111 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 Monthly rainfall during summer (rnm)

Figure 8. Distribution of summer monthly rainfall at Oamaru.

250

• ez 200

co E 150

=Li! 100 -

• • • Trs • • • 00 . • ----• • • 0 ... " • et II ro • • -W." • Pa io lb am 11!: •31,.. • ,.2 , • • til • 0 0 • • 0 •flank, • ID • •• • •• • • • -50 - . t • 9) e

-100 -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 Spring SOI

Figure 9. Relationship between spring SOI (average SOI for September-November) and summer rainfall departure from average (mean rainfall departure from average for December-February) at Oamaru. Historic Droughts in North Otago 45 neutral springs and the following summers, 75 droughts were identified by the as indicated by Figure 9. And the relationship Herbst's technique, with individual appears to be a negative correlation, with low events ranging from two to 37 months. rainfall relatively more likely during La Nina Drought events identified for the period events. During El Nino events, the possibilities 1867-1982 were generally the same as of high rainfall and low rainfall to be received those identified by the Otago Catchment are similar. However, El Nino springs are Board and Regional Water Board using relatively more likely to bring high rainfall to the 116-year record. With longer rainfall North Otago than springs of other episodes records, the most severe drought (Table 9). While it has been identified that identified previously was in fact three during El Nino events, low rainfall generally separate events. occurs in the east due to more frequent and Droughts are a common phenomenon in stronger winds from the west such as in North Otago, particularly droughts with Canterbury, this pattern does not suit North short duration. Intense droughts were Otago. In fact, the relationship between spring generally of short duration and severe SOT and summer rainfall in North Otago is droughts were generally of long duration, similar to that between SOT and summer but with varied intensities. Clutha lakes inflow identified by McKerchar The 1997-99 drought was considered to in 1994. be the 3rd most severe event in North CONCLUSIONS Otago during the past 132 years and the 2nd most severe drought during the 1980's The 1997-99 drought was considered a and 1990's. The drought has been a major event in North Otago, with record particularly lengthy but not particularly breaking high temperatures, low rainfall intense event. and river flows recorded. The driest area There was a lack of high flows in the was centred in the coastal downlands, Kakanui River during the 1997-99 including Oamaru, the middle and lower drought compared with other events in catchment of the Kakanui and the 1980's and 1990's. Low flow was Waianakarua River, and the lower significantly more persistent. catchment of the Shag River.

Oamaru No. of months No. of months No. of months following following following El Nino spring neutral spring La Nina spring No. of months with high rainfall 24 19 12 No. of months with near normal rainfall 16 22 8 No. of months with low rainfall 26 37 31 Total 66 78 51

Palmerston No. of months No. of months No. of months following following following El Nino spring neutral spring La Nina spring No. of months with high rainfall 21 18 10 No. of months with near normal rainfall 18 29 11 No. of months with low rainfall 27 31 30 Total 66 78 51 High rainfall: rainfall departure from mean > 20%. Near normal rainfall: rainfall departure from mean < 20% and > -20%. Low rainfall: rainfall departure from mean < -20%.

Table 9. Relationship of spring SOI and rainfall during summer months 46 Historic Droughts in North Otago

The impact of the 1997-99 drought on REFERENCES streamflow was more significant than the droughts 1992-93 (11 months duration), Basher, R., 1998. The 1997/98 El Nino Event: Impacts, Responses and Outlook for New Zealand. Report by 1994-95 (13 months duration) and 1981- the National Institute of Water and Atmospheric 83 (20 months duration). However, the Research for the Ministry of Research, Science and extreme low flows during the 1997-99 Technology. drought were not as low as those during Bee, A.W., 1984. Long-term Agricultural Drought in North Otago, Thesis, University of Otago. the 1984-85 and 1988-90 events, which Borlase, 0.M., 1976. Determination of Drought had duration of 24 and 29 months Frequencies and the Impact of Droughts in North respectively. Otago, Otago Catchment Board publication, Dunedin. During the 1997-99 drought, the critical Coulter, J.D., 1973. A Water Balance Assessment of the New Zealand Rainfall. Journal of Hydrology (NZ), duration in the Kakanui River was long vol. 12, no. 2, pp83-91, 1973. (30 days) and was short in the Shag River Gordon, N.D., 1985. The Southern Oscillation: a New (I day). Zealand perspective. Journal of the Royal Society of Low flows were generally more sustained New Zealand, vol. 15, no. 2, pp137-155, 1985. Gordon, N.D., 1986. The Southern Oscillation and New in the Kakanui River catchment than in Zealand Weather. Monthly Weather Review, vol 114, the Shag River catchment. pp371-387. • There is no obvious relationship between Harrison, W., 1988. The Influence of the 1982-83 Drought monthly rainfall in North Otago and SOT. on River Flows in Hawke's Bay. Journal of Hydrology (NZ), vol 27, no 1, pp 1-25, 1988. • Summer rainfall was highly variable and Herbst, PH., Bredenkamp, D.B. and Barker, HAIG., relatively more likely to be below normal, 1966. A Technique for the Evaluation of Drought from disregarding the El Nino Southern Rainfall Data. Journal of Hydrology vol 4, pp 264- Oscillation phenomena. 272, 1966. McKerchar, A., 1994. Long Term Variability in River There appears to be a general negative Flows. Water & Atmosphere, 2(3), pp14-15, 1994. relationship between spring SOT and Miller, I.D., 1980. Meteorological Events Which Lead to summer rainfall, with low rainfall Prolong Dry Weather. In: McMahon TA (ed.) Low relatively more likely during La Nina Flow Workshop Keynote Address and Abstracts of Other Papers. Christchurch, 24-25 July 1980. events. During El Nino events, the Moss, M.E., Pearson, UP. and McKerchar, AI, 1994. The probabilities of high rainfall and low Southern Oscillation index as a predictor of the rainfall are similar but El Nino springs probability of low streamflows in New Zealand. Water are relatively more likely to bring high Resources Research, vol. 30, No. 10, pp2717-2723, 1994. rainfall to North Otago than springs of Otago Catchment Board and Regional Water Board, 1983. other episodes. The Kakanui River Catchment Water Resource Inventory Otago Catchment Board and Regional ACKNOWLEDGEMENTS Water Board Technical Report. Otago Regional Council, 1991. The Shag Catchment Grateful thanks to the Hydrology Team of Resource Description Issues and Options for Management. Otago Regional Council Report. the Otago Regional Council, which conducted Otago Regional Council, 2000. The effects of the 1999 the river flow measurement and part of the drought on Otago's rivers. Otago Regional Council rainfall measurement. I also thank the Report. National Institute of Water and Atmospheric Waugh, J., Freestone, H. and Lew, D., 1997. Historic floods and droughts in New Zealand. In: Mosley & Research Ltd. for providing rainfall and SOI Pearson (ed.) Floods and Droughts: the New Zealand records. Experience, New Zealand Hydrological Society Inc, Wellington, New Zealand.