Extreme events (droughts and floods) have significant impacts on agricultural production and natural resource management. In the Pacific Rim including Papua New Guinea (PNG) and eastern Australia, the tele-connections (relationship over a long distance) of climate-related anomalies with El Nio and La Nia events are strong and are reliable enough for use in decision making.
In PNG, sweet potato is the dominant staple food and is therefore the most important crop in the country. Over 60% of the rural population depend on it as their main food source. About 75% of annual sweet potato production (which totalled 2.8M tonnes in 2000) is grown in the highlands. Climatic extremes, particularly high soil moisture, droughts and frosts are among the main constraints to production. Coffee is grown mainly for export, representing~40% of all agricultural exports. About 85% of coffee produced is by small land holders and represents a valuable source of income.
Historically PNG has experienced severe rainfall deficiencies or droughts in 1885, 1896, 1902, 1910, 1914, 1940-1941, 1955-56, 1960, 1965, 1972, 1982, 1987, 1991-1995 and 1997. Extremely wet years were experienced in 1908, 1916-1917, 1921, 1938-1939, 1943-1944, 1952-1953, 1961-1963, 1964, 1971, 1985-1986, 1988-1989. In most cases, droughts were associated with El Nio's and wet events were associated with La Nia's.
Lessons learned from the 1997-98 drought in PNG demonstrated the vulnerability of agricultural production to climate impacts both in terms of food security and farm income. A review of the current hazard monitoring capabilities and procedures after the 1997 drought recommended development of improved systems that provide early warning of developing threats and regularly updated information on their characteristics and progress. It is therefore a priority in PNG to develop an effective climate forecasting and warning system focussing on drought response strategies, information on quantitative measures of drought and improved crop management practices. This project retrieves long-term rainfall data for PNG, examines its relationship with El Nio Southern Oscillation (ENSO) and investigates the utility of drought warning tools to help maintain food security (sweet potato) and farm income (coffee).
Data preparation
The rainfall datasets were sourced from PNG National Weather Service (NWS) and consisted of data from the colonial PNG archive, NWS data collected post-1975 and one dataset from the National Agricultural Research Institute (NARI) research station at Aiyura. After some data pre-processing, good quality monthly rainfall data was created for 10 stations (Aiyura, Daru, Kavieng, Lae/Nadzab, Madang, Misima, Momote, Port Moresby, Rabaul/Tokua and Wewak) with a length of record between 52-106 years. These data were used in SCOPIC (Seasonal Climate Outlook for Pacific Island Countries) for analysis of drought and to determine the skill of seasonal climate forecasts (SCF) based on key ENSO indices.
Seasonal climate forecasting skill
In general ENSO-based predictive systems present significant skill, for at least some parts of the year, for most of the southern sea-level stations, and for the season lengths tested (3 and 5 months). This suggests that ENSO-based SCF of rainfall would be useful tools for agricultural and natural resource management in many parts of lowland Papua New Guinea during certain periods of the year. These periods vary depending on the geographic location of the rainfall stations, and the seasonal length of the forecast. The stations exhibiting the highest levels of skill are located on the coastal areas of the main island, and also in the southern outer islands. ENSO had little impact on the eastern islands closer to the equator and at high elevations (1200-1800 m), although Aiyura (1570 m) was the only station with data suitable for analysis in the highlands. It appears that using ENSO-based forecasts may be of limited value for locations in the highland areas, where much of the sweet potato and coffee are grown, although lack of adequate rainfall datasets make this assessment difficult.
Drought analysis
The drought analyses undertaken were based on a broad definition of "meteorological drought" defined as total rainfall over 4, 6, and 12 month periods using the decile and Standardised Precipitation Index (SPI) methods of analysis to trigger warnings for drought (decile 4, SPI -0.2) and identify periods of drought (decile 1, SPI -1).
SCOPIC was used to study the impact of ENSO on droughts (Objectives 1.2, 1.3 and 1.4). Regular reoccurrences of El-Nio-based droughts occurred at locations exhibiting high seasonal climate forecasting skill (e.g. Daru, Madang, Port Moresby, Misima and Wewak). Locations whose droughts predominantly occurred during La Nia events generally exhibited limited or no SCF skill (e.g. Rabaul and Kavieng). The success rate of drought warnings (rainfall totals reaching decile 4 (SPI-0.2) and continuing to decile 1 (SPI-1) was useful (~60-80%) at Aiyura, Lae, Madang, Misima, Port Moresby, Rabaul and Wewak.
This analysis has provided an indication of which locations may be used to predict probable drought outcomes using ENSO, however this process will be further enhanced through proposed modifications to SCOPIC. These modifications will involve linking summary SCF skill scores (for predictand lengths equivalent to the drought monitoring period) to the SCOPIC drought analyses.
An additional study conducted was to assess the nature of the 'warning success rate' for varying definitions of drought, based on adjustment of the drought warning level (addressing Objective 1.2). Preliminary results demonstrate that an optimum warning-level exists (be it broadly defined) for maximising the success-rate of the drought monitoring with adequate lead-time. Initial results suggest that the utility of this process increases at locations where there is a high correlation of ENSO with drought, with noticeable increases in lead-time. A new analysis based on this process is now being developed for SCOPIC. Upon completion of this feature, SCOPIC will be able to automatically provide recommendations addressing Objective 1.2.
Upgrades to SCOPIC
The SCOPIC software was significantly enhanced during the reporting period with the latest Version 2.3.19 now being used by NWS and NARI staff. Enhancements included the addition of new data summary tools and reports (required to automate the assessment of PNG rainfall data), as well as significant modifications to the in-built drought analyses tool. In particular, the methodology for calculating drought using the decile-method was enhanced to better accommodate seasonality. Also, a new analysis for optimising the definition of drought-monitoring indices for particular locations has been developed based on the findings of the preliminary studies (60% complete). Finally, new automated drought monitoring reports (using English phrasing) were developed which are suitable for media release statements on drought monitoring.

The predominance of subsistence agriculture in Papua New Guinea (PNG) highlights the importance of food security. In general terms the villagers have learnt to manage (wantoks, purchase of food from the sale of cash crops e.g. coffee, potatoes) the localised shortages of food that occur regularly. It is the large scale shortages of food that occur irregularly that threaten human health and survival (e.g. 1997).
During these extreme events (droughts, floods and frosts) that cause widespread food shortages the PNG government has relied upon food aid (national and international) and more recently on villagers' self-reliance to purchase imported food. It is the more remote and isolated communities that are most vulnerable because of their poor access to food distribution points and markets to sell produce from cash crops.
This project retrieved long-term rainfall data for PNG, examined its relationship with El Nio Southern Oscillation (ENSO) and investigated the utility of drought warning tools to help maintain food security (sweet potato) and farm income (coffee). In parts of PNG the association between El Nino Southern Oscillation (ENSO) and rainfall are strong and reliable enough to improve decision making in crop management.
Through the training activities and tools developed in this project research and extension officers in the PNG highlands now have an understanding of ENSO, its association with rainfall in PNG and the climate tools (e.g. SCOPIC) that can provide an early warning of droughts and floods. However they require further training and continued support in order to usefully advise villagers of impending extreme events.
Sweet potato is a staple food for villagers and although it is relatively drought tolerant, excessive wet periods during tuber initiation followed by drought during tuber development significantly reduce tuber production, and the low yields may not be discovered until harvest. Repeated frost also significantly reduces tuber yield. Forecasting these events some months in advance can initiate alternative management such as the planting of varieties with a short growing season, planting in mounds to conserve moisture and planting under vegetation to protect against frost.
Smallholders produce about 85% of the coffee grown in PNG making it a valuable cash crop for many villages. The production of coffee is triggered by natural cycles of dry and wet conditions however extreme wet periods in poorly drained soils have as large a negative impact on production as drought, particularly during cherry ripening and development when large quantities of nutrients and water are required for high yields.