Report

Translating ENSO Parameters into Local Weather Variables

This session deliberated issues relating to downscaling global ENSO forecasts to make them actionable by national-level policy-makers. The presentations included the methodologies to assess possible mechanisms through which ENSO influences climate controls and local weather variables.

Presentation Highlights

For Indonesia, Sri Diharto noted that the meteorological agency of Indonesia (BMG) operates a seasonal prediction scheme dividing the country into 102 rainfall districts called seasonal forecasting areas.

The forecast methodology uses statistical models that still depend on rainfall data, which is variable across place and time. The seasonal forecast techniques are:

• Statistical (regression) techniques based on relationships between rainfall and SOI.
• Probability methods based on the rainfall time-series for that district.
• Auto-regressive techniques based on the time-series.
• Utilization of seasonal forecasts and current information on the SOI issued in the seasonal outlooks of the BOM (Australia).
• General synoptic experience in monitoring the situation at the time of issuance of the forecast

ENSO parameter sensitivity differs from one seasonal forecasting area to another. To account for this difference, the seasonal prediction scheme relies on statistical analogues of past rainfall patterns. A method that uses rainfall data and ENSO parameters needs to be developed further to incorporate other meteorological parameters such as upper air data (wind and water vapor content).

One of the major constraints to providing statistical analogue forecasts was non-availability of past rainfall data in a usable format. Hence, archiving and processing past rainfall data is a priority for the BMG. The major endeavour will be processing the data and storing it digitally for retrieval and use.

To overcome the constraints, future actions will include:

• Replacing conventional meteorological monitoring sensors at collaborating stations with tele-metered sensors and using satellite communication.
• Exploring other meteorological parameters in the forecast methodology.
• Improving database management to facilitate research activities.
• Improving data quality through instrument calibration.
• Encouraging staff to do applied research.

Aida Jose reported that there is a lack of understanding in the region of ENSO influences. The reduction of global ENSO forecasts into local climate forecasts using numerical regional climate models is non-existent, in part because of insufficient scientific personnel who can understand and translate the ENSO forecasts. This is crucial given the need for local ENSO-based forecasts and their limitations to be easily understood by various end-users.

In the absence of an appropriate numerical regional or local climate model for the Philippines, translation of ENSO forecasts was carried out using the approach of forecasting potential impacts on local climate by analogy. The following information was used as diagnostic tools for downscaling global ENSO forecasts into local seasonal climate forecasts.

El Nino Extended dry season Early end of rainy season Weak monsoon activity Less number of tropical cyclones          Above normal sea level pressure Above normal air temperature Dry weather conditions

La Nina Short dry season Early onset of rainy season Strong monsoon activity More number of tropical cyclones Below normal sea level pressure   Wetter weather conditions

The Philippines meteorological agency (PAGASA) also derived indicators to assess potential impacts of predicted climate variables on various sectors such as agriculture and water resources. The methodology was based on the principle of potential impact assessment by analogy, involving translation of ENSO forecasts to impacts on local climate and then to potential impacts on various sectors.

Some of the constraints encountered in the translation of ENSO forecasts into local climate forecasts include but are not limited to:

• Climate parameters other than ENSO and modes of atmospheric variability that interplay other than the air-sea interaction in the Central and Eastern Equatorial Pacific, or the ENSO that impacts the climate variability in the Philippines. Such parameters include SST anomalies in the South China Sea and the Indian Ocean, the Quasi-Biennial Oscillation (QBO) and the Madden Julian Oscillation (MJO).
• Existence of some degree of uncertainty (inherent?) in the various General Circulation Models used in generating ENSO forecasts and disparity in their results.
• Selection and application of suitable local climate models for the Philippines has yet to be done.

Suggested measures to overcome constraints include:

• More studies on the dynamics of ENSO influences in the region.
• Formal training and/or fellowship programs in graduate programs, or immersion experience in global climate centers.
• An operational numerical regional climate model to predict ENSO influence on the local climate.
• More interfacing opportunities between local climate forecast producers and end-users.
• A guidance manual for an integrated ENSO impact assessment is needed to provide a easy, coherent flow of information, effective collaboration and feedback mechanism.

Constraints in Vietnam, according to Pham Duc Thi, relate to non-integration of global ENSO parameters into a long-range forecasting scheme in quantitative terms. However, ENSO forecasts have been recently used for making-long range forecasts qualitatively. A considerable research effort is required to develop human resources as well as equipment to use global ENSO forecasts for translation of global ENSO parameters into local weather variables in quantitative terms.

There are significant sub-national variations and interactions with other non-ENSO processes. In using tropical cyclones as a local weather variable, there is a need to know not just frequency, but also cyclone intensity. An inability to produce quantitative forecasts and limited experience with dynamic modeling are other constraints. In Vietnam, there are plans to:

• Study ENSO interaction with other relevant processes.
• Incorporate local knowledge in making forecasts.
• Focus on parameters where skill is high.

Discussion Points

In considering the translation of ENSO parameters into local weather variables, participants questioned how dynamic modeling is done for multiple impacts. In Vietnam, multiple events can be a limitation on resources. For example, a serious drought may be followed by floods, which could then be followed by storms and landslides. The sequential occurrence of multiple extreme weather events like droughts followed by floods, which would then be followed by storms and landslides, needs to be factored into the modeling of long-range forecasts.

Climate variability is a continuous process and causes severe weather events even in neutral years. Hence, the sea surface temperature patterns in the Central Eastern Pacific constantly cause regional climate variations. Climate forecasts need to be evolved to capture all flavors of sea surface temperature patterns, regardless of the occurrence of El Nino and La Nina.

One of the biggest problems in translating ENSO parameters into local weather variables is that the current models come from "developed countries". The variables are fed into global models, but then the information from them must be translated to reflect impacts at local level.

The release of climate information to the general public and the government can be problematic; therefore, the role of the media is crucial to developing awareness and improving reactions to an ENSO forecast. The same definitions and vocabulary may not be shared by sector agencies and the national meteorological agencies. The lack of common vocabulary poses a serious difficulty for decision-makers and the media to appreciate uncertainties inherent in long-range forecasts. The development of a common, understood vocabulary may be the link to effective communication with the media, the sector agencies and the public. One example of this is in Indonesia, where the agriculture sector requests forecasts from the meteorological agency to make a forecast, but has a different interpretation of the meaning of the results, which are reported as "normal", "above normal" or "below normal" rainfall. When the agriculture agencies are not satisfied with the information, they try to do their own analysis and forecasts, which wastes money and resources. The key to this problem may be improved communication about climate information among the sectors.

End-users of climate forecast information have unrealistic expectations of the meteorological service, which shoulders a huge responsibility for the information that they release, given that different users will interpret the information to serve their own interests. To the extent that the public does not know how to interpret the degree of uncertainty in the forecasts, some countries have found that it helps to focus on one type of forecast that can be done well. For example, in Vietnam, forecasts are concentrated on typhoon predictions. As confidence builds in the meteorological services, it will be easier to convey other types of forecasts.

Recommendations

Following the presentations and discussions, the participants focused on several issues of importance. These included the need to downscale information and make models useful locally, as well as ways that meteorological agencies should work to present useful information. The participants suggested the following:

• Undertake a comprehensive assessment of long-range forecast methodologies in each country to identify gaps and measures to address them, to enable meteorological agencies to downscale global long-range forecasts into actionable formats for use by policy-makers and end-users.
• Integrate user' needs and suggestions to provide user-friendly long-range forecasts.
• Conduct more studies on the dynamics of ENSO influences in the region.
• Develop an operational numerical regional climate model to predict ENSO influence on the local climate.
• Develop a mechanism for intermediate interpretation of long-range forecasts that are based on ENSO parameters into locally usable information to remove the barriers to incorporating probabilistic forecast information into decision-making processes.
• Address the critical gap between short-range and long-range forecasts. Mechanisms exist in all three countries for transmission of short- and medium-range forecast information by meteorological agencies to forecast user organizations. The long-range climate forecasts are also routinely transmitted by meteorological organizations to user organizations. However, since long-range forecasts based on global ENSO indices are of a probabilistic nature, user organizations are unable to utilize them effectively for decision-making.
• Encourage capacity-building efforts to enable user departments and applied research organizations to process and refine long-range forecast information products into usable information for decision-making and policy-planning purposes. This may also effectively reduce the gap between short- and long-term forecasts.
• Develop an institutional arrangement to promote continuous dialogue among local climate forecast producers, intermediary research institutions, policy-makers and various end-users.
• Develop and promote formal training and fellowships in graduate schools, or immersion experiences for trainees in global climate forecast centers.
• Undertake concerted efforts to educate the public about the nuances of climate issues and climate forecasts.
• Undertake systematic efforts to develop common definitions and vocabulary.