Log into your member account to listen to this article. Not a member? Join the herd.

Eastern Africa has been grappling with multiple humanitarian crises exacerbated by climate-induced drought emergencies, disease outbreaks, floods and social instability due to civil conflict and the prolonged effect of 2019 locust plagues and the COVID-19 pandemic. Between 2017 and 2023, the population needing humanitarian assistance in parts of Eastern Africa rose from 22.5 million to 68 million and, as reported in the financial tracking systems of the United Nation Office for the Coordination of Humanitarian Affair- UN-OCHA, the cost of humanitarian assistance doubled from US$4.1 billion to US$9.4 billion.

Of the crises besetting the region, severe drought is the most significant humanitarian emergency, especially for rural communities, as livelihoods primarily depend on animal husbandry and farming. Over the past 40 years, the region has experienced severe droughts: in 1976-1978, 1985-1988, 2010-2011, 2016-2017 and 2020-2022. Due to these crises, there has been significant interest in early warning systems and anticipatory planning in development and humanitarian contexts.

In particular, following the 1985 famine that resulted from severe drought and production failure, huge investments in early warning, preparedness and response were made. For example, the US Agency for International Development (USAID) created FEWSNET—Famine Early Warning System Network—an agency for evidence-based early warning information. FEWSNET’s primary focus is to provide scientific information on acute food insecurity, agro-climatic conditions and drought early warning to governments, international relief agencies, scientists, and NGOs, among others, for actionable response in preventing drought and famine emergencies. Equally, Eastern African states established ICPAC, initially known as the IGAD Drought Monitoring Centre in Nairobi (IDMC-N). ICPAC is the World Meteorological Organisation’s (WMO) regional climate centre of excellence. These agencies, together with others, work closely with government meteorological departments at the regional, national and local level to provide timely early warning information for preparedness, contingency planning and early action.

Over the last 20 years, the accuracy of  early warning information has improved, at least for short-term predictions, but the main challenge has been reaching local communities—what some call the “last mile”. The result is that early warning information often does not reach where it is most needed. Despite all the talk of early warning, disaster risk reduction, shock-responsive systems, contingency planning and anticipatory action, the end results are mixed to say the least. We need to ask, how can these big investments in early warning be linked to local approaches to prediction and response?

Predicting droughts and communicating the predictions through risk reports and early warning bulletins is now standard practice. In Kenya, the impressive National Drought Management Authority (NDMA), a government outfit based in 23 Arid and Semi-Arid Land (ASAL) counties, was established in 2011 with significant donor support. It produces monthly bulletins stacked with information derived from earth observations by satellites as well as surveys of key vulnerability indicators (household food consumption, market prices for livestock, food, water, livestock body condition, vegetation, status etc.) collected across each region. These bulletins are shared with the county government, the array of NGOs working in each area, and local communities.

Despite the deluge of high-quality information, the gap between early warning (which is increasingly accurate, at least for the short-term) and action on the ground is enormous. This has been a perennial problem. There are issues of trust (why should I believe the government?), inertia (surely if I wait a bit, then things will get better) and communication styles (a dozen pages in English rather than vernacular and visual versions, although this is apparently going to change). Moreover, those working on the ground know that there’s a drought right now (livestock is dying in numbers, there is no grass and water), so they don’t need information that the situation is dire. As one frustrated NDMA officer observed, “With early warnings you are telling them what they already see. We are ambassadors for what they already know!”

Deliberating on uncertainties: the need for local debate

The big problem with such information systems is that they are usually one-way: we have the information, you should listen and act. There is no space for dialogue, deliberation and debate. There are always uncertainties: Does this really apply here? Why wasn’t the drought predicted correctly last time? Is this information relevant to me right now? The assumption of specialised expertise filling a “deficit” in local knowledge and understanding has long prevailed in debates about science-policy interactions; it applies as much to early warning and drought alert information in pastoral drylands.

Despite the deluge of high-quality information, the gap between early warning and action on the ground is enormous.

This gap was recognised by a number of agencies that came together to design the Community-Managed Disaster Risk Reduction (CMDRR) approach, based on a participatory diagnosis of problems and joint construction of solutions. While the CMDRR committees are aimed at producing development and contingency plans that can then articulate with funding programmes from the government and NGOs, the most essential part of these committees is the process.

Meeting monthly and composed of a group of locally selected “experts”, they draw on local experience and knowledge and discuss impending or unfolding crises. This may be drought, but also conflict, animal disease or other challenges facing them, right there in their own context. This deliberation is crucial as diverse views are shared, dispute and contestation are possible, and in this way, uncertainties (for they are always there) are addressed.

For example, in one village some way off the main road near Moyale, we met the chair of the local committee who explained its functioning. There are 23 members, 15 men and 8 women. The roles are voluntary although they have been supported—now over nine years—by a local NGO. The membership includes elders with long, historical experience of past crises and how these were addressed, and several people with specialist expertise.

The assumption of specialised expertise filling a “deficit” in local knowledge and understanding has long prevailed in debates about science-policy interactions.

Among these local experts is a man who is an expert in treating sick and injured animals (specialised in local techniques for bone-setting). His knowledge is sought by community members when animals become sick in “normal” times, but when a particular disease spreads dramatically, he is a crucial point of contact. With veterinary officers few and far between, he must link with those selling drugs, but also those with knowledge (as he has) of traditional herbs and treatments. The local “disease reporters” pass information upwards to their superiors, but their local knowledge is also crucial in understanding disease at a local level. Connecting these networks is crucial in responding to a crisis, as described for North Horr, also in Marsabit county. The CMDRR is thus a vital platform for integrating and sharing this knowledge.

Local early warning: the role of community-based prediction and response

In addition to those with expertise in particular facets of crisis response, there are others who act as the local early warning system; they claim that they never make use of the NDMA bulletins but have their own system. This is perhaps not surprising: there is no phone network in the village, and they are not provided with data bundles to download the documents with all their graphs and tables. Instead, they make use of locals who are experts in predicting droughts and other crises.

Two such experts are members of the committee. One woman recently inherited the role of Uchu from her mother, expert reader of animals’ intestines. Her mother was renowned throughout the area as someone who could accurately predict what will happen by inspecting the intestines of a recently slaughtered goat, cow or bull. They must be animals that have been born and raised in the area and ideally are young calves or kids. Usually, the intestines of animals slaughtered for weddings, funerals or naming ceremonies are used by such experts. If the signs are unclear, the process is repeated with a newly slaughtered animal of the right type. Those who read the signs are offered a fried portion of the liver. Once eaten the predictions are made, and people discuss. Sometimes there are conflicting versions from different people, and further deliberations have to be made. Even in the indigenous science of making predictions using animal intestines, there are uncertainties.

Although intestine readers can divine the future across a range of hazards, others may be referred to. Some throw shoes to see what the future might bring, while others gaze at the stars. These indigenous astronomers are especially well regarded. In the same village where we conducted our interviews, an interpreter of the patterns of the stars was also present. People view the local astronomer as especially good at predicting future climate events, usually over a more extended period than those who read from the intestines of slaughtered animals.

Even in the indigenous science of making predictions using animal intestines, there are uncertainties.

Of course, predictions only happen at a certain point in time, and in relation to a certain set of questions that community members pose. But droughts, conflicts, disease outbreaks and so on unfold over time in uncertain ways. This is why predictions must be repeated, and adaptations and responses to these must be continuous, part of a process. Combining multiple knowledge is essential, along with discussions around uncertainties, if a humanitarian crisis is to be contained based on early warning information.

Closing the early warning gap 

The problem with the centralised early warning systems, and the whole paraphernalia of reporting that follows, is that they too often do not reach the “last mile”—the affected communities. This is where the early warning’s “missing  link” has long been identified. Often distrusted and perceived as alien to local contextual knowledge, recommendations are frequently rejected.

This is why the NDMA in Moyale has, with the encouragement of a local NGO, started to work with local early warning specialists in workshops where external, “scientific” information is shared at the local level and debated alongside the local interpretations and predictions. In Moyale sub-county the NDMA has invited traditional forecasters from across the region, including different ethnic groups. At a workshop, they slaughter a goat, and each individual inspects the intestines. After completing their inspections, they share the results and compare them with the ICPAC and Meteorological Department forecasts.

Often distrusted and perceived as alien to local contextual knowledge, recommendations are frequently rejected.

As the local NDMA officer explained, despite debate about the specifics, there was remarkable convergence between the different views. Building trust with local communities through using local knowledge in tandem with external, “scientific” sources is seen as an important route to communication, with community radio programmes planned where the results can be discussed.

And yet, the huge investments in early warning systems using the very best satellite technologies and highly sophisticated interpretation techniques often assume a linear transformation of information, from those who know and those who don’t. But this ignores the fact that local pastoralists are well practised in predicting and responding to drought. In the end, the fancy technological solutions are no match for the local deliberations on the ground about uncertain futures using multiple sources of knowledge.

No-one expects these predictions to be correct all of the time—whether local or external—but it’s the deliberation around uncertainties that ensues following a prediction that is important in shaping local responses. Effective responses always have to be embedded in local contexts, drawing on local knowledge and social relations, and this is why too often external interventions around “resilience” fail and why alternatives are needed.

This is article is adapted from the second of a series of three blogs written as part of a scoping study and supported by ACIAR (Australian Centre for International Agricultural Research).