*4.1.5. Explaining the basis of seasonal forecasting*

498 Risk Management – Current Issues and Challenges

year of the project.

*4.1.2. Building trust* 

*4.1.3. Training* 

skillful seasonal forecasts at district level, presentation of probabilistic seasonal forecasts using easy-to-understand terminology, training farmers with the new information for them and translate this into decisions. This study concludes with lesson learned during the initial

The approach used in the project was to help the various stakeholders identify strategies for successful CRM using seasonal forecasting. The project sought to build the trust of farmers, while working with all relevant local organizations. In West African's culture, trust is the most important requirement in order to be effective in a viable partnership. A common saying states, "the manner to give is better than what is given". In order to build trust with farmers and foster a sustained relationship, a workshop that included 30 farmers from 6 villages around Kaffrine was planned. Since ANACIM does not have the legal mandate to implement any agricultural strategy, partnership with the local agriculture department representative (SDDR) was ensured. The SDDR has the mandate to monitor activities related to the farming system, and also has arbitration authority in case of conflicts over issues such as farm allocation, fertilizer subsidizing, buying harvest products. SDDR was a natural contact point with farmers, since they had already developed a long time partnership and has their trust. It was very important that the project team not appear as a stranger in the system, but work through a known entity. Association was developed with other local technical services including agricultural advisers from the national agency for agricultural and rural advice (ANCAR), which has a presence nationwide at district level and a mandate to advise farmers in term of agricultural strategies. Volunteers from World Vision, a Christian charity that assists children and invests a lot in agriculture in the district of Kaffrine, also participated. Participants included individual farmers, and members of farmer organizations such as

JAPPANDO. Women represented about 30 per cent of those participating.

Consistent with the strong oral tradition, time was reserved during the workshop to allow farmers to interact with the experts team. On the first day of the workshop, the floor was given to farmers to describe their experiences with forecasting the weather and climate. The technical experts started with differentiating the concepts of weather (imminent) and

A challenge for the CRM approach was to enable farmers to trust and use scientific information. There was a clear need for a common ground, where farmers would use the new scientific seasonal forecasting approach proposed to them without feeling that their indigenous approaches to seasonal forecasting were being rejected. In this culture, the scientists would lose credibility if the farmers were to think the scientists were saying that

climate (longer term), as these concepts are interchangeable in the local language.

*4.1.4. Connecting with farmers' indigenous knowledge* 

Another challenge was how to explain to farmers, in easy-to-understand terms, how climate forecasts work. Many farmers knew about weather forecasts, communicated through the weather bulletin on national TV. But the real challenges were to convince them that the rain could be forcast one to two months ahead, and to help them to understand the probabilistic nature of forecasts at this lead-time. The basis of seasonal forecasting was explained to the farmers by calling upon their intuition. When asked, "When it is hot, why do people go to the beach?" they responded that the sea breeze brings fresher air. They were then asked,

"Then why? Isn't it the same sun that heats both land and ocean? Why then does the ocean get cooler in summer?" It was explained that ocean has better memory of the past compared to the continent. Ocean remembers the heat of the past days and weeks. That's why, on a very hot day people go to the beach to benefit from ocean memory of the past weeks. Similarly, when it is cold, the ocean still remembers recent warm days. The ocean's heat memory is the basis for seasonal forecasting. As rain comes from clouds, clouds come from water vapor, and most water vapor from the ocean, they could see how ocean temperature could control rain. The farmers were also informed that satellites are used to monitor ocean temperature throughout the world, and computers quantify the likelihood of rain in Senegal. This very simplistic explanation helped them to make sense of scientific seasonal forecasts, and was sufficient to convince them to trust the forecast during this first contact.

Improving Climate Risk Management at Local Level –

Techniques, Case Studies, Good Practices and Guidelines for World Meteorological Organization Members 501

It was decided to express the forecast as a probability of exceedence of rain instead of the probability of occurrence of the three tercile categories that meteorological services in West Africa officially issue in their seasonal forecasts. Farmers understand the notions of uncertainty and probability, but understanding and acting on formal probability formats is challenging. To help them understand the new probability of exceedance format, an exercise of classifying the last 5 years of rainfall that they recollected from memory was conducted (Fig. 4). A chart of 30 years of Kaffrine rainfall data was provided. The farmers could see that it is very likely that they would get at least as much rainfall as the driest year, and very unlikely that they would get more than the wettest year. The middle years represent "normal" conditions. How to identify the 25th, 50th and 75th percentiles of rainfall from the graph was discussed. The idea that a dry forecast would shift the distribution toward the left and wet forecast to the right was introduced. Hypothetical wet and dry forecasts were discussed until the farmers appeared to understand what they meant. As the probability of exceedence is a cornerstone of the training, the farmers were divided into four groups. Two groups were given probability of exceedence forecasts for hypothetical dry years and the other two groups were given hypothetical wet years. The farmers were asked to discuss what they would do differently if this were the actual forecast for the upcoming season. Each group reported back on their forecast and strategies. The whole group was encouraged

After the training, the next step was to build a communication strategy to ensure that the information will reach the farmers effectively. A discussion of the best way to communicate the seasonal forecast revealed a number of options. Among the means identified, cell phones appeared first as a cheap technology. This option is accessible, since most of the farmers have a cellular phone, and is consistent with the traditional use of oral communication. Local radio was the other promising means of communicating forecast information. All of

*4.1.7. Probability of exceedance graphs* 

to comment on these strategies.

**Figure 4.** Farmers sorting seasonal rainfall.

*4.1.8. Communicating the forecast* 

#### *4.1.6. Getting past the technical language barrier*

The next challenge was to explain the probabilistic nature of the forecast, which is less intuitive than a deterministic rainfall amount. To start with, the farmers were asked to recollect from their memory the last 5 rainy seasons and rank them from the wettest to the driest. With a pluviograph (Fig. 3), it was explained how rainfall is recorded in millimeters, and what 1mm of rain means. One mm of rain was poured into the soil, then the farmers were asked to compare it with the quantity of rain that they consider sufficient to plant their crops. They indicated that they plant when the soil wetting front is greater than the span of an average man's hand. To help them interpret what a seasonal total means, a discussion was held on how the temporal distribution of rain relates to the seasonal total. It was clear for the farmers that a seasonal forecast gives an idea of the total, but no information about its distribution in time. One farmer explained the difference between a good rainy season and a good cropping season, which was a clear indication that they understood the seasonal forecast output. The farmers were informed also what could potentially be forecast and what could not.

**Figure 3.** Training farmers to read a rain gauge.

### *4.1.7. Probability of exceedance graphs*

500 Risk Management – Current Issues and Challenges

*4.1.6. Getting past the technical language barrier* 

**Figure 3.** Training farmers to read a rain gauge.

what could not.

"Then why? Isn't it the same sun that heats both land and ocean? Why then does the ocean get cooler in summer?" It was explained that ocean has better memory of the past compared to the continent. Ocean remembers the heat of the past days and weeks. That's why, on a very hot day people go to the beach to benefit from ocean memory of the past weeks. Similarly, when it is cold, the ocean still remembers recent warm days. The ocean's heat memory is the basis for seasonal forecasting. As rain comes from clouds, clouds come from water vapor, and most water vapor from the ocean, they could see how ocean temperature could control rain. The farmers were also informed that satellites are used to monitor ocean temperature throughout the world, and computers quantify the likelihood of rain in Senegal. This very simplistic explanation helped them to make sense of scientific seasonal forecasts, and was sufficient to convince them to trust the forecast during this first contact.

The next challenge was to explain the probabilistic nature of the forecast, which is less intuitive than a deterministic rainfall amount. To start with, the farmers were asked to recollect from their memory the last 5 rainy seasons and rank them from the wettest to the driest. With a pluviograph (Fig. 3), it was explained how rainfall is recorded in millimeters, and what 1mm of rain means. One mm of rain was poured into the soil, then the farmers were asked to compare it with the quantity of rain that they consider sufficient to plant their crops. They indicated that they plant when the soil wetting front is greater than the span of an average man's hand. To help them interpret what a seasonal total means, a discussion was held on how the temporal distribution of rain relates to the seasonal total. It was clear for the farmers that a seasonal forecast gives an idea of the total, but no information about its distribution in time. One farmer explained the difference between a good rainy season and a good cropping season, which was a clear indication that they understood the seasonal forecast output. The farmers were informed also what could potentially be forecast and It was decided to express the forecast as a probability of exceedence of rain instead of the probability of occurrence of the three tercile categories that meteorological services in West Africa officially issue in their seasonal forecasts. Farmers understand the notions of uncertainty and probability, but understanding and acting on formal probability formats is challenging. To help them understand the new probability of exceedance format, an exercise of classifying the last 5 years of rainfall that they recollected from memory was conducted (Fig. 4). A chart of 30 years of Kaffrine rainfall data was provided. The farmers could see that it is very likely that they would get at least as much rainfall as the driest year, and very unlikely that they would get more than the wettest year. The middle years represent "normal" conditions. How to identify the 25th, 50th and 75th percentiles of rainfall from the graph was discussed. The idea that a dry forecast would shift the distribution toward the left and wet forecast to the right was introduced. Hypothetical wet and dry forecasts were discussed until the farmers appeared to understand what they meant. As the probability of exceedence is a cornerstone of the training, the farmers were divided into four groups. Two groups were given probability of exceedence forecasts for hypothetical dry years and the other two groups were given hypothetical wet years. The farmers were asked to discuss what they would do differently if this were the actual forecast for the upcoming season. Each group reported back on their forecast and strategies. The whole group was encouraged to comment on these strategies.

**Figure 4.** Farmers sorting seasonal rainfall.

#### *4.1.8. Communicating the forecast*

After the training, the next step was to build a communication strategy to ensure that the information will reach the farmers effectively. A discussion of the best way to communicate the seasonal forecast revealed a number of options. Among the means identified, cell phones appeared first as a cheap technology. This option is accessible, since most of the farmers have a cellular phone, and is consistent with the traditional use of oral communication. Local radio was the other promising means of communicating forecast information. All of

the participating farmers listen to the radio, but the listening quality of the radio is very poor when they are on the farm. Some NGOs and farmers association leaders recommended e-mail as a possibility. The administrative authority who was present mentioned the government's network of heads of village. In case of an extreme event, this can be used to reach each village within an hour. The local authority showed his support and promised to help with access to this facility. To avoid conflicts between farmers' organizations, other farmers recommended sending the information through the SDDR, who knows how to contact them.

Improving Climate Risk Management at Local Level –

Techniques, Case Studies, Good Practices and Guidelines for World Meteorological Organization Members 503

the word, how the rain was, etc. It was not a forecast but rather a monitoring exercise. It was good to touch base. During the first field trip in selected villages (October 12-13 2011), some farmers made major decisions, such as borrowing money from the bank to invest more in their farm, or hiring workers. Another field visit was made around the end of the cropping

In January, three months after the rainy season, when farmers have sold their crops and finished their farming work, an evaluation workshop was organized in Kaffrine to assess the use and usefulness of the seasonal forecast strategy. Local extension services were present, as well as farmers' organizations. Fifteen of the farmers who attended the training workshop in June were invited back, along with 13 other farmers who hadn't received information about seasonal forecasting. During the January workshop, participants assessed both 2011 seasonal rainfall and the performance of various crops grown in the district. The participants took the opportunity to discuss in three groups, and interpret the information presented. One group included 12 farmers that had received the forecast and adjusted some decisions in response to the forecast (group I). The next groups included 3 participants who did receive the forecast but didn't make any adjustment to their farming practices (group II), and the last group consisted of 13 farmers who had never received any climate forecast information (group III). They were asked to document actions taken, problems encountered, and recommendations. Group I understood from the workshop that a short cycle crop was suitable because the season was to be less than 2010, but rainfall would be enough. The main problems they listed were: the high spatial variability of the rainfall, the late occurrence of the first rainfall which made it difficult to judge when to start planting, a long dry spell, and early termination of the season. They wanted to know or get: the starting date, finer forecast information in space, a weather bulletin each two weeks, and more training to better understand the forecast. Group II did not use the seasonal forecast because they had already bought their seeds at that time which made it difficult to change any of their farming strategy. Group III, who had never received any climate information, indicated that they had thought 2011 would be like 2010. They missed the opportunity of a long season in 2010, and were prepared to catch up the next year by choosing a long cycle variety, buying fertilizers and hiring wage laborers. The group members concluded that their problem was that they didn't know anything about the course of the rainy season and needed to be part

The Workshop participants were given the chance to evaluate the whole process – from farmer selection, to organization of the workshop, to training agenda – in order to identify what is needed for improvement. There is a need to improve the communicating system by using already existing channels. World Vision recommended that training more trainers

season (October 18-22, 2011) to conduct surveys on expected yield.

*4.1.11. Evaluation of the seasonal forecast* 

of the group that received seasonal forecast training.

*4.1.12. Lesson learned and way forward* 

## *4.1.9. From theory to practice*

A week after the training work, ANACIM sent a group of experts to call a meeting to communicate the actual July-September 2011 seasonal forecast with the farmers. Twentytwo attended. Some key points from the training were revisited: good rainy season versus good cropping season, probability of exceedence interpretation, plausible management response strategies, definition of 1mm. Rain gauges were distributed to some representatives of farmers' organizations who expressed need for this tool, and the meaning of a millimeter of rain was again demonstrated. The forecast was presented with an explanation on how to interpret it (Fig. 5). The forecast in this case was "normal to abovenormal." As the year before, 2010, had been exceptionally wet – the highest on record – it was indicated that rainfall this year (2011) would probably be less than 2010. Some explanations about what the seasonal forecast did not say were also offered. Recommendations on any particular management strategies were not made, but rather it was left open to each farmer to decide. Considering that this was a first contact with them, it was preferred to build trust first before offering recommendations.

**Figure 5.** Training on interpreting the probability of exceedance.

#### *4.1.10. Keeping in touch*

Through this project, funds were available to undertake two field visits during the season, and also to call selected farmers from time to time. When the first big rainy event occurred, some farmers were asked if they planned to use the seasonal forecast, whether they spread the word, how the rain was, etc. It was not a forecast but rather a monitoring exercise. It was good to touch base. During the first field trip in selected villages (October 12-13 2011), some farmers made major decisions, such as borrowing money from the bank to invest more in their farm, or hiring workers. Another field visit was made around the end of the cropping season (October 18-22, 2011) to conduct surveys on expected yield.

### *4.1.11. Evaluation of the seasonal forecast*

502 Risk Management – Current Issues and Challenges

contact them.

*4.1.9. From theory to practice* 

the participating farmers listen to the radio, but the listening quality of the radio is very poor when they are on the farm. Some NGOs and farmers association leaders recommended e-mail as a possibility. The administrative authority who was present mentioned the government's network of heads of village. In case of an extreme event, this can be used to reach each village within an hour. The local authority showed his support and promised to help with access to this facility. To avoid conflicts between farmers' organizations, other farmers recommended sending the information through the SDDR, who knows how to

A week after the training work, ANACIM sent a group of experts to call a meeting to communicate the actual July-September 2011 seasonal forecast with the farmers. Twentytwo attended. Some key points from the training were revisited: good rainy season versus good cropping season, probability of exceedence interpretation, plausible management response strategies, definition of 1mm. Rain gauges were distributed to some representatives of farmers' organizations who expressed need for this tool, and the meaning of a millimeter of rain was again demonstrated. The forecast was presented with an explanation on how to interpret it (Fig. 5). The forecast in this case was "normal to abovenormal." As the year before, 2010, had been exceptionally wet – the highest on record – it was indicated that rainfall this year (2011) would probably be less than 2010. Some explanations about what the seasonal forecast did not say were also offered. Recommendations on any particular management strategies were not made, but rather it was left open to each farmer to decide. Considering that this was a first contact with them, it

Through this project, funds were available to undertake two field visits during the season, and also to call selected farmers from time to time. When the first big rainy event occurred, some farmers were asked if they planned to use the seasonal forecast, whether they spread

was preferred to build trust first before offering recommendations.

**Figure 5.** Training on interpreting the probability of exceedance.

*4.1.10. Keeping in touch* 

In January, three months after the rainy season, when farmers have sold their crops and finished their farming work, an evaluation workshop was organized in Kaffrine to assess the use and usefulness of the seasonal forecast strategy. Local extension services were present, as well as farmers' organizations. Fifteen of the farmers who attended the training workshop in June were invited back, along with 13 other farmers who hadn't received information about seasonal forecasting. During the January workshop, participants assessed both 2011 seasonal rainfall and the performance of various crops grown in the district. The participants took the opportunity to discuss in three groups, and interpret the information presented. One group included 12 farmers that had received the forecast and adjusted some decisions in response to the forecast (group I). The next groups included 3 participants who did receive the forecast but didn't make any adjustment to their farming practices (group II), and the last group consisted of 13 farmers who had never received any climate forecast information (group III). They were asked to document actions taken, problems encountered, and recommendations. Group I understood from the workshop that a short cycle crop was suitable because the season was to be less than 2010, but rainfall would be enough. The main problems they listed were: the high spatial variability of the rainfall, the late occurrence of the first rainfall which made it difficult to judge when to start planting, a long dry spell, and early termination of the season. They wanted to know or get: the starting date, finer forecast information in space, a weather bulletin each two weeks, and more training to better understand the forecast. Group II did not use the seasonal forecast because they had already bought their seeds at that time which made it difficult to change any of their farming strategy. Group III, who had never received any climate information, indicated that they had thought 2011 would be like 2010. They missed the opportunity of a long season in 2010, and were prepared to catch up the next year by choosing a long cycle variety, buying fertilizers and hiring wage laborers. The group members concluded that their problem was that they didn't know anything about the course of the rainy season and needed to be part of the group that received seasonal forecast training.

#### *4.1.12. Lesson learned and way forward*

The Workshop participants were given the chance to evaluate the whole process – from farmer selection, to organization of the workshop, to training agenda – in order to identify what is needed for improvement. There is a need to improve the communicating system by using already existing channels. World Vision recommended that training more trainers

would be the best way. Overall, the farmers appreciated the experience of last year and welcomed more training.

Improving Climate Risk Management at Local Level –

Techniques, Case Studies, Good Practices and Guidelines for World Meteorological Organization Members 505

8°15`N and 12°50`N and longitudes 74°50`E and 77°30`E). The location map of Kerala is

**Figure 6.** Location map of Kerala. Source: www.mapsofindia.com

The annual rainfall across Kerala is highly variable, averaging about 3000mm, but varying

Seasonally, rainfall is bimodal, due to the influence of both the summer and winter monsoons, with maximum monthly rainfall (>600mm) during the summer monsoon in June and July, and winter monsoon rainfall (200-300 mm) during October. Heavy rainfall during the summer monsoon, followed by a prolonged dry spell is a characteristic feature of the humid tropics, which is particularly prominent in the case of the northern districts,

Annual average surface air temperature varies between 25 and 30°C, with a seasonal range between around 18°C in winter and 35°C in winter. The altitude across Kerala varies from

including Kasaragod, where the influence of winter monsoon is negligible (Fig. 7).

*4.2.1. Rainfall and thermal regimes of Kerala* 

between less than 1000mm to greater than 5500mm.

given in Fig. 6.

Seasonal climate forecasts could have considerable potential to improve agricultural management and livelihoods for smallholder farmers. But constraints related to legitimacy, salience, access, understanding, capacity to respond and data scarcity have so far limited the widespread use and benefit from seasonal predictions in the Sahel region. The existing constraints reflect inadequate information services, policies or institutional processes in the region. However there is great potential to overcome these constraints. An approach is suggested that packages: i) seasonal and onset forecasts, ii) opportunity for farmers to implement strategies, and iii) insurance tools in case of extreme variable or dry years. Even when the seasonal rainfall or onset matches the forecast, poor farmers wouldn't profit if they don't have access to funds or crop varieties to implement any forecast-based strategy. And it turns out that in Kaffrine, there is often false start of the rainy season, making it imperative to provide farmers with alternatives, for example through index insurance.

As work with farmers in Kaffrine on the forecast continues, research is being conducted and a working group on improving prediction of intra-seasonal variability has been set up. Crop producers and seed bank will be invited into the process, to allow farmers to access suitable varieties for forecast-based strategies. There is some work on index insurance in the region, and it is planned to reach out to involve such groups in this effort. Through this approach it is hoped to gain success, avoid frustration and build long-term partnerships.
