SEARCH

SEARCH BY CITATION

Keywords:

  • rainfall variability;
  • agricultural impacts;
  • Australia

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Rainfall and wet spell investigation
  6. 4. Climate diagnostics in extreme wet seasons
  7. 5. Summary
  8. Acknowledgements
  9. References

The sugarcane harvest season in northern New South Wales (NSW), Australia operates from June to November. Rainfall during this period can delay harvest practices and push the harvest window into the wetter summer period which can have severe economic consequences for industry. Farmer groups acknowledge that, whilst information about seasonal rainfall totals can assist forward planning activities impacted by harvest rainfall, knowledge about the number of wet spells during the harvest season would also be helpful. This diagnostic study investigated the interaction between harvest rainfall totals and wet spell frequency for Harwood, a major sugarcane growing region in Northern NSW. The El Nino-Southern Oscillation (ENSO) effect on harvest rainfall totals and wet spell frequency was also considered and climate diagnostics for extreme wet seasons were investigated. A moderate linear relationship between total rainfall and wet spell frequency exists for both winter (r = 0.68) and spring (r = 0.66). The ENSO had little influence in winter but La Niña events in spring favoured higher rainfall totals and wet spell frequency. Extreme wet seasons were largely influenced by positive pressure anomalies over southeastern Australia and the Tasman Sea. Copyright © 2009 Royal Meteorological Society


1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Rainfall and wet spell investigation
  6. 4. Climate diagnostics in extreme wet seasons
  7. 5. Summary
  8. Acknowledgements
  9. References

Farming enterprises across the world are vulnerable to variability in climate conditions. The El Nino-Southern Oscillation (ENSO) can explain part of this variability, but there exist other modes of variability, some known and some unknown that will also influence seasonal rainfall. Thus, it is necessary to understand both the strengths and limitations of ENSO, and in ENSO limiting circumstances we need to know what other climate patterns may have the potential to influence industry decision-making processes. Whilst improved understanding about the impacts of climate variability on agriculture systems is interesting and assumed to be helpful by climate practitioners, it does not guarantee industry adoption of climate forecasting technologies. Climate practitioners must at the very least be mindful of obstacles that impede the adoption process. At a time when the world is challenged to find sustainable solutions that will feed and fuel our expanding population, understanding how these issues can deliver benefits to agricultural industries is of paramount importance.

The connection between ENSO and agricultural enterprises such as sugarcane systems is well established. The ENSO has influenced sugarcane production in the West Indies, Florida, South Africa and Australia (Kuhnel, 1994; Singels and Bezuidenhout, 1998; Singels and Bezuidenhout, 1999; Hansen et al., 2001; Pulwarty and Eischeid, 2001; Everingham et al., 2003). Advance knowledge about crop size is important because it can help industry forward plan the sale of the crop. Whilst planning the sale of the crop is an extremely important task, there are many other planning activities that can be guided by probabilitistic knowledge about future ENSO conditions. Everingham et al. (2002, 2008a, b) described how ENSO can influence decisions about harvesting and milling practices and presented a method for linking knowledge about ENSO conditions with irrigation scheduling systems. Across the farming, harvesting, milling and marketing sectors of the industry value chain there is a wide array of actions that can be influenced by probabilistic knowledge of seasonal climate outlooks.

Whilst the connection between ENSO and sugarcane systems is clear, one should not assume that climate forecasting technologies which predict ENSO events will routinely be sought and integrated as part of regular planning practices within an industry context. The literature outlines many reasons for the impediments to the adoption of seasonal climate forecasting technologies (Stern and Easterling, 1999; Ziervogel and Downing, 2004; Hansen, 2005; Garbrecht and Schneider, 2007; Hayman et al., 2007; Jakku et al., 2007; Everingham, 2008a). Low skill or perceived low skill associated with the forecasting system is one reason that has impeded the adoption process. Difficulties associated with interpreting and understanding seasonal rainfall probabilities is another cause. A major factor, however, that has impeded the adoption pathway is the mismatch between farmers needs and scientific outputs. In many cases, forecasting systems are simply not forecasting what end users require. The importance of scientists working closely with industry practitioners to better align scientific expertise with industry needs is now well accepted. This type of participatory research (Carberry et al., 2002; Jakku et al., 2007) has lead to the development of ‘targeted forecasts’ (Meinke and Stone, 2005) to match the needs of industry stakeholders.

Motivated by the potential benefits associated with targeted forecasts and the importance of understanding the strengths and limitations of ENSO, this diagnostic study presents a case study centred on Harwood, a major sugarcane growing region in New South Wales (NSW) and one of the southern most located sugarcane growing regions in Australia. One of the key periods for industry is June–November when the sugarcane is harvested. Ideal harvest conditions would have few interruptions to ensure the sugarcane is harvested before the onset of the summer wet season, approximately December. Wet weather interruptions can have disastrous economic consequences across the farming, harvesting, milling and marketing sectors of the industry value chain. Antony et al. (2002) showed that for one Australian sugarcane growing region located in North Queensland (Ingham 18°39′S, 146°10′E), wet conditions during 1998 cost the industry in excess of $ 19M AUD dollars. Costs because of wet weather harvesting can also flow through to the marketing level where wet conditions can delay the supply of sugar to the consumer. At the mill level, rainy conditions bring mud and other residue to the mill processing plant which causes machine breakdowns. Harvestors can also be bogged during wet conditions. When the ground dries, bogged areas become compacted and less productive for successive crops. In some years when the crop is not completely harvested, farmers will have to wait till the following year to harvest the crop, and experience delays to their cash flow situation. Improved understanding about influences that favour wet harvest seasons is important to guiding the development and integration of seasonal climate forecasts.

Seasonal rainfall forecasts for the harvest season can be accessed from a wide array of organisations across the world. Extension officers in NSW also disseminate climate forecasts to cane farmers via local newsletters. Most of these forecasts detail the chance that rainfall over consecutive 3-monthly periods will exceed the median or fall into a particular tercile. Whilst surveys have revealed that industry considers seasonal climate forecasts to be helpful, industry regularly comments that receiving a small amount of rainfall more frequently during the harvest season is worse than receiving a lot of rainfall in one downpour. In response to industry feedback and limited research conducted to date on both total rainfall and frequency of rain events, the purpose of this study is threefold. Firstly, we consider the relationship between harvest rainfall totals and the number of wet spells, i.e. the number of pentads that exceed a threshold amount of rain. This analysis is done for winter and spring separately. These seasons represent the first and second half of the harvest season. Secondly, we investigate the effect that ENSO has on the relationship between rainfall and wet spell frequency. Thirdly, we investigate atmospheric circulation and sea surface temperature (SST) features for extreme years that are both high in rainfall totals and the frequency of wet spells as these years can have devastating economic consequences for industry.

2. Data and methodology

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Rainfall and wet spell investigation
  6. 4. Climate diagnostics in extreme wet seasons
  7. 5. Summary
  8. Acknowledgements
  9. References

Daily rainfall data from 1950 to 2005 were sourced from the Australian Bureau of Meteorology for the Harwood Sugar Mill (station number 58027) which is located at 29°26′S and 153°15′E. The rainfall records were used to compute seasonal and pentad (5 day) rainfall totals during the harvest season. Three monthly rainfall totals for first half (June–August) and second half of the harvest season (September–November) were calculated. To compute pentad rainfall totals during the harvest season, the calendar year was first divided into 73 pentads. Pentads 31–49 represented the first half of the harvest season (winter) and pentads 50–67 represented the second half (spring). For convenience, and with little impact to this investigation, the 29th February was omitted from the study. A wet spell was defined to be a pentad that had more than 11.2 mm of rainfall which represented the upper quartile of rainfall across pentads 31–67. Scatterplots and the Pearson correlation were used to investigate the association between total winter rainfall and the frequency of winter wet spells. These calculations were repeated for spring rainfall and spring wet spell frequency.

The influence of ENSO on winter and spring rainfall and wet spell frequency was also assessed. The winter and spring ENSO classifications were derived from the three monthly averaged Nino 3.4 anomalies computed from the second version of the extended reconstruction of global SSTs (Smith and Reynolds, 2004). A winter was defined as an El Niño (La Niña) state if the monthly averaged Niño 3.4 anomaly from June to August was greater (less) than 0.5 °C (−0.5 °C). Winter was defined to be in a neutral pattern if the averaged monthly Nino 3.4 anomaly from June to August was between − 0.5 °C and 0.5 °C, inclusively. Spring ENSO classifications were as for the winter ENSO classifications but used monthly Niño 3.4 anomalies for September to November. Table I lists the yearly ENSO classifications for winter and spring. In most years, the winter and spring had the same ENSO classification (e.g. 1972, 1998), but in some years like 2000, the ENSO classification differed between winter and spring. A Kruskal-Wallis statistical test (Conover, 1980) was used to test for significant shifts in the rainfall and wet spell frequency distributions between El Niño, La Niña and neutral winters and springs.

Table I. Seasonal ENSO classifications
 WinterSpring
La Niña1950 1954 1955 1956 1964 1970 1971 1973 1974 1975 1988 1998 19991950 1954 1955 1956 1961 1962 1964 1970 1971 1973 1974 1975 1983 1984 1988 1995 1998 1999 2000
Neutral1951 1952 1953 1958 1959 1960 1961 1962 1966 1967 1968 1969 1976 1977 1978 1979 1980 1981 1983 1984 1985 1986 1989 1990 1992 1993 1995 1996 2000 2001 2003 20051952 1953 1958 1959 1960 1966 19671978 1968 1979 1980 1981 1985 1989 1990 1992 1993 1996 2001 2005
El Niño1957 1963 1965 1972 1982 1987 1991 1994 1997 2002 20041951 1957 1963 1965 1969 1972 1976 1977 1982 1986 1987 1991 1994 1997 2002 2003 2004

Years when the rainfall and wet spell frequency were considerably high are of particular interest to industry. These extreme years were more closely investigated by considering a range of climate diagnostics to understand what made these years atypical.

3. Rainfall and wet spell investigation

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Rainfall and wet spell investigation
  6. 4. Climate diagnostics in extreme wet seasons
  7. 5. Summary
  8. Acknowledgements
  9. References

The average rainfall total for June–November for Harwood is approximately 450 mm, much lower than the mean rainfall from December to May which is 900 mm. Whilst aggregate figures are interesting, industry has asked researchers to consider the distribution of rainfall throughout the harvest season. This was done separately for winter and spring. As an example, Figure 1 shows the winter pentad rainfall profile for 1954 and 1999. The total amount of winter rainfall for these 2 years is similar. During the 1954 and 1999 winter harvests, 432 mm and 474 mm of rainfall was recorded, respectively. Although the total winter rainfall recorded for both years were similar, most of the rain was accumulated in one event during 1954, whereas winter rainfall during 1999 was more regularly distributed. Almost one decade on from the disastrous 1999 harvest season, farmers can still vividly recall the challenges that arose from this persistently wet season (personal communication, P. McGuire, NSW Extension Officer).

thumbnail image

Figure 1. Total amount of rainfall for each pentad during the winter harvest for (a) 1954 and (b) 1999

Download figure to PowerPoint

To investigate the relationship between total rainfall, and the number of wet events, we define a wet spell to be a 5-day period that had at least 11.2 mm of rainfall. This cut-off was chosen by first computing the 75th percentile of rainfall for each pentad across the years from 1950 to 2005; 11.2 mm was the median of the upper quartile for rainfall pentads (Figure 2).

thumbnail image

Figure 2. Upper quartile (y-axis) for each pentad (x-axis) computed across the years 1950–2005. The horizontal line shows the median height of all bars which represents the cut-off for a wet pentad. This cut-off occurs at 11.2 mm. Pentads 31–49 occur during the first half of the harvest season (June–August) and pentads 50–67 occur during the second half of the harvest season (September–November)

Download figure to PowerPoint

Figure 3 shows the relationship between total rainfall and wet spell frequency during the winter and spring harvest periods. There is negligible difference between the moderately positive Pearson correlation between rainfall and wet spell frequency 0.68 (winter) and 0.66 (spring). Although there are examples like 1954 and 1999 that motivate industry to question the relationship between rainfall total and wet spell frequency, Figure 3 reinforces the view that often where there is a high (low) amount of rainfall, there will also be a high (low) number of wet spells. The impact of ENSO on this relationship, however, is less consistent. From the scatterplot, ENSO has little impact on total rainfall and wet spell frequency during winter. During spring, La Niña classifications tend to favour higher rainfall totals and more wet spells. Boxplots for total rainfall and wet spell frequency during winter and spring by ENSO category support these findings (Figure 4). The Kruskal Wallis test was significant at the 0.10 level during spring for both rainfall (p = 0.016) and wet spell frequency (p = 0.075). Conversely, the test showed that ENSO classification did not impact the rainfall (p = 0.872) or wet spells (p = 0.524) during winter.

thumbnail image

Figure 3. Relationship between total rainfall (y-axis) and wet spell frequency (x-axis) during (a) winter and (b) spring harvest. Years marked in blue/red/grey correspond to years when the Niño 3.4 region was defined to be in a La Niña/El Niño/Neutral state for the season of interest. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

Download figure to PowerPoint

thumbnail image

Figure 4. Distributional shifts in (a) winter and (b) spring harvest rainfall and (c) winter and (d) spring wet spell frequency between ENSO phases. The lower and upper edge of each box represents the 25th and 75th percentiles, respectively, and the horizontal line internal to the box is the median. Lines from the upper (lower) edge of the box extend to the largest (smallest) rainfall index that is not deemed to be extreme and the floating horizontal lines identify extreme cases that are more than 1.5 times the inter-quartile range above (below) the upper (lower) edge of the box. The shaded area defines the 95% confidence interval for the median

Download figure to PowerPoint

4. Climate diagnostics in extreme wet seasons

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Rainfall and wet spell investigation
  6. 4. Climate diagnostics in extreme wet seasons
  7. 5. Summary
  8. Acknowledgements
  9. References

The most negative economic consequences occur in years with a high amount of rainfall and a large number of rain events. Although both 1950 and 1999 had a high amount of rainfall and wet spells during winter as well as being La Niña, they are examined separately as they appear to be outliers (Figure 3). The spring harvest of 1972 had a high amount of rainfall and wet spell frequency. This season is interesting for two reasons—(1) the ENSO classification with spring rainfall and wet spell frequency is significant and (2) it has an El Niño classification which generally tends to favour drier conditions. We examine climate diagnostics associated with these extreme seasons to understand what influences may have contributed to the moist harvest conditions.

To obtain some insight into why winter 1950 was unusually wet, Figure 5 shows anomalies in SST, geopotential height and zonal wind at 850 hPa, and mid-tropospheric (500 hPa) omega. A cyclonic anomaly is apparent over coastal southern Queensland/northern NSW together with enhanced ridging further south leading to stronger easterly winds over the northern Tasman Sea/southern Coral Sea. Further upstream, warm SST anomalies are evident over the eastern Coral and Tasman Seas suggesting that the easterly flow towards Queensland and northern NSW would have contained moister marine air than average, favourable for rainfall. Globally, a La Niña pattern was in evidence. In addition to the favourable regional SST and horizontal circulation patterns, the omega plot shows a large negative anomaly (i.e. strong relative ascending air at 500 hPa) extending from inland southern Queensland/NSW right across the Tasman Sea/southern Coral Sea to the dateline. Thus, conditions here were favourable for uplift and convection and the advection of an unstable marine airmass towards the Harwood region, consistent with the wet conditions observed.

thumbnail image

Figure 5. June–August 1950 anomaly for (a) SST, (b) 850-mb geopotential height, (c) 850-mb zonal wind and (d) 500-mb Omega. The climatology was computed from winter seasons between 1950 and 2005, inclusively

Download figure to PowerPoint

The situation for winter 1999 (Figure 6) was somewhat similar to 1950 in terms of strong anticyclonic anomalies over southeastern Australia and the southern Tasman Sea and hence stronger easterly winds towards the Harwood region as well as an area of negative omega anomaly (relative uplift) along the northern NSW and southern Queensland coasts. However, the anomalies are less pronounced for 1999 than for 1950 and the regional SST anomaly shows a smaller area of warming upstream of eastern Australia with, in fact, a weak cool anomaly right near the coast. Consistent with these somewhat less favourable conditions for rainfall in this season, the positive rainfall anomaly for winter 1999 is substantially less than that for winter 1950 (Figure 3). Note that because there is little evidence of an ENSO signal during the winter harvest season, we have simply presented the anomalies for 1950 and 1999 as differences from the long term June–August mean.

thumbnail image

Figure 6. June–August 1999 anomaly for (a) SST, (b) 850-mb geopotential height, (c) 850-mb zonal wind and (d) 500-mb Omega. The climatology was computed from winter seasons between 1950 and 2005, inclusively

Download figure to PowerPoint

Another very anomalous case was the spring 1972 season whose very large rainfall amount and frequent wet spells were very unfavourable for the sugar industry. This season was quite different from the other El Niño spring seasons. Because there was a strong El Niño signal in spring, we show the difference in the various fields for this season relative to the composite of the remaining El Niño springs listed in Table I.

Compared with other El Niño years, 1972 shows more positive SST anomalies in the central equatorial Pacific, but interestingly also shows stronger positive SST anomalies in the Coral and Tasman Seas east of Australia (Figure 7). A large anticyclonic anomaly extending from southwest of Western Australia right across the continent and well into the South West Pacific is reflected in the easterly wind anomalies over coastal Queensland/NSW. In addition, there is a negative omega anomaly relative to the other El Niño seasons (i.e. more uplift) over the Harwood region. Thus, these conditions suggest more advection of relatively unstable marine air towards Harwood and increased rainfall compared to the other seasons. Consistent with these plots, spring 1972 was not only much wetter than other El Niño seasons (Figure 3(b)) but also relative to the record as a whole and plots with similar anomaly patterns to Figure 7 are found if the anomaly of 1972 relative to the long term mean is calculated (not shown).

thumbnail image

Figure 7. September–November 1972 anomaly for (a) SST, (b) 850-mb geopotential height, (c) 850-mb zonal wind and (d) 500-mb Omega. The climatology was computed from El Niño springs between 1950 and 2005, inclusively

Download figure to PowerPoint

5. Summary

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Rainfall and wet spell investigation
  6. 4. Climate diagnostics in extreme wet seasons
  7. 5. Summary
  8. Acknowledgements
  9. References

In this study, we considered both total rainfall and the number of wet spells as per the request of industry. Although we considered both total seasonal rainfall and wet spell frequency, these two variables were correlated. Knowledge about this correlation is important. Not only does it allow researchers to concentrate efforts on exploring just one response instead of two, but this process is deemed acceptably by industry and thus avoids a potential barrier to the adoption of climate forecast technologies.

The ENSO had little impact in winter but had a stronger influence in September–November. Analysis of regional circulation patterns for harvest seasons with an anomalously high amount of rainfall and/or wet spells suggests that strong ridging over southeastern Australia/southern Tasman Sea together with cyclonic anomalies further north and above average SST in the eastern part of the Coral/Tasman Seas play a significant role. These patterns lead to stronger easterlies impacting the Harwood region which likely contain more moisture than average because of the warmer SST over their fetch. Local low pressure anomalies and relative uplift then provide favourable conditions for rainfall. This knowledge was found to be useful to industry to explain the anomalously wet harvest seasons, especially in 1972, which was an El Niño year. Figure 8 suggests that these findings need not be limited to extreme seasons because it shows strong positive correlations between wet spell frequency and sea level pressure over southeastern Australia and the Tasman Sea. These correlations are stronger and more spatially extensive in spring than in winter.

thumbnail image

Figure 8. Correlations between wet spell frequency and NCEP re-analysis mean sea level pressure for (a) June–August and (b) September–November. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

Download figure to PowerPoint

We emphasize that the impact of climate circulation features and ENSO were investigated from a diagnostic perspective. The harvest season was considered closely because wet conditions during this time can have negative economic consequences for the Harwood sugarcane growing regions. This knowledge coupled with further research can guide the development of prognostic studies to provide early warning about the likelihood of extreme wet harvests through identification of climate mechanisms that can lead to devastating conditions. This information can then allow industry to implement the appropriate preparation needed to minimise the impacts of a wet harvest season.

Although we have presented an application for Harwood, part of the NSW sugar industry, the framework can easily be extended to other sugarcane growing regions around the world. Similarly it can be extended to other cropping systems where knowledge about the interaction between total rainfall and/or wet spell frequency is needed to minimise the impact of interruptions to major industry operations such as harvesting. By so doing, sustainable efficiencies may be obtained through increased profitability and productivity. Agricultural industries have and always will be vulnerable to climate conditions, so it is important that we understand the impacts that different atmospheric and oceanic circulation patterns can have on industry practice in pursuit of gaining sustainable efficiencies through increased profitability and productivity across the supply chain.

Acknowledgements

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Rainfall and wet spell investigation
  6. 4. Climate diagnostics in extreme wet seasons
  7. 5. Summary
  8. Acknowledgements
  9. References

This research has been supported through funding provided by the Australian Government through the Sugar Research and Development Corporation and by James Cook University. Images have been provided by the NOAA-ESRL Physical Sciences Division, from their Web site at http://www.cdc.noaa.gov/” and from the Climate Explorer website at http://climexp.knmi.nl.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Rainfall and wet spell investigation
  6. 4. Climate diagnostics in extreme wet seasons
  7. 5. Summary
  8. Acknowledgements
  9. References
  • Antony G, Everingham YL, Smith DM. 2002. Financial Benefits From Using Climate Forecasting —A Case Study. Proceedings of the Australian Society for Sugar Cane Technologists 24: 153159.
  • Carberry P, Hochman Z, McCown R, Dalgliesh N, Foale M, Poulton P, Hargreaves J, Hargreaves D, Cawthray S, Hillcoat N, Robertson M. 2002. The FARMSCAPE approach to decision support: farmers', advisers', researchers' monitoring, simulation, communication and performance evaluation. Agricultural Systems 74: 141177.
  • Conover WJ. 1980. Practical Nonparametric Statistics, 3rd edn. John Wiley and Sons: New York.
  • Everingham YL, Baillie B, Inman-Bamber NG, Baillie J. 2008a. Forecasting water allocations for canefarmers. Climate Research 36: 231239.
  • Everingham YL, Clarke AJ, Van Gorder S. 2008b. Long lead rainfall forecasts for the Australian Sugar Industry. International Journal of Climatology 28: 111117.
  • Everingham YL, Muchow RC, Stone RC, Coomans DH. 2003. Enhancing sugarcane yield forecasting capability using SOI phases: a case study for north eastern Australia. International Journal of Climatology 23: 11951210.
  • Everingham YL, Muchow RC, Stone RC, Inman-Bamber NG, Singels A, Bezuidenhout CN. 2002. Enhanced risk management and decision-making capability across the sugar industry value chain based on seasonal climate forecasts. Agricultural Systems 74: 459477.
  • Garbrecht JD, Schneider JM. 2007. Climate forecast and prediction product dissemination for agriculture in the United States. Australian Journal of Agricultural Research 58: 966974.
  • Hansen JW. 2005. Integrating seasonal climate prediction and agricultural models for insights into agricultural practice. Philosophical Transactions of the Royal Society B 360: 20372047.
  • Hansen JW, Jones JW, Irmak A, Royce F. 2001. El Niño-Southern Oscillation impacts on crop production in the southeast United States. ASA Special Publication No. 63; 5576.
  • Hayman P, Crean J, Mullen J, Parton K. 2007. How do probabilistic seasonal climate forecasts compare with other innovations that Australian farmers are encouraged to adopt? Australian Journal of Agricultural Research 58: 975984.
  • Jakku E, Thorburn PJ, Everingham YL, Inman-Bamber G. 2007. Improving the participatory development of decision support systems for the sugar industry. In Proceedings of the 29th Australian Society of Sugar Cane Technologists Conference: Mackay, Queensland, Vol. 29; 4149.
  • Kuhnel I. 1994. Relationship between the Southern Oscillation index and Australian sugarcane yields. Australian Journal of Agricultural Research 45: 15571568.
  • Meinke H, Stone RC. 2005. Seasonal and inter-annual climate forecasting: the new tool for increaseing preparedness to climate variability and change in agricultural planning and operations. Climatic Change 70: 221253.
  • Pulwarty RS, Eischeid J. 2001. The impact of El Niño-Southern Oscillation events on rainfall and sugar production in Trinidad. In Proceedings of the 27th West Indies Sugar Technologists Conference: Trinidad, 11.
  • Singels A, Bezuidenhout CN. 1998. ENSO, the South African climate and sugarcane production. Proceedings of the South African Sugar Technologists Association 72: 1011.
  • Singels A, Bezuidenhout CN. 1999. The relationship between ENSO and rainfall and yield in the South African sugar industry. South African Journal of Plant and Soil 16: 96101.
  • Smith TM, Reynolds RW. 2004. Improved extended reconstruction of SST (1854–1997). Journal of Climate 17: 24662477.
  • Stern PC, Easterling WE. 1999. Making Climate Forecasts Matter. National Academy Press: Washington, DC.
  • Ziervogel G, Downing TE. 2004. Stakeholder networks: improving seasonal climate forecasts. Climatic Change 65: 73101.