What the Climate Does to the Hills

Preparation begins with an honest forecast. This chapter sets out what climate science actually projects for Rwanda and the Great Lakes region (no denial, no theatre), and then does what the previous chapter said must be done: reads the projections politically, because Rwanda's climate exposure runs through its hillsides and its neighbours before it runs through any global average.

One fact frames everything. Rwanda's entire historical contribution to the problem rounds to nothing: per-capita emissions around a tenth of a tonne of CO₂ a year against roughly fifteen for the United States, and none of that innocence buys exemption from a single consequence.¹ Chapter 31 called that the boundary of obligation working as designed. Here it is simply the starting condition.

What the models agree on

For East Africa the model ensembles assessed by the IPCC agree on more than the popular impression suggests.²

Begin with what has already been measured. Rwanda's national communications under the climate convention report a warming trend of roughly 1.4 °C since the early 1970s, ahead of the global average for the same period.³ The gauge network is thin (a chronic East African problem), which is one reason the farmers of the epigraph deserve to be read as instruments: the long rains (itumba) arriving late and quitting early, the short rains (umuhindo) breaking into false starts. The projections below extend a measured trend; they do not invent one.

Heat, unambiguously. The region has already warmed by roughly a degree; every scenario adds more, with the 2040s (the rupture window) carrying regional warming in the range of 1.5 to 2.5 °C over pre-industrial depending on emissions. For a temperate-highland country whose settlement pattern, crops, and disease burden were all calibrated to altitude-cooled climate, warming does not mean discomfort. It means the entire altitudinal zoning of Rwandan life shifts uphill.

Harder rain, unambiguously. Warmer air holds roughly seven per cent more moisture per degree, and the ensembles agree that when rain falls over the Lake Victoria basin and the Albertine highlands, more of it will fall in intense bursts. Rwanda does not need a model to believe this; it has the death tolls. The rains of early May 2023 killed at least 130 people in the Western and Northern Provinces in roughly two days, destroyed thousands of homes, and cut national roads, one storm system.⁴ On terraced volcanic slopes where a third of the country farms gradients above sixteen degrees, intense rain is not a weather event. It is a geological one: landslide, gully, and the loss of the soil that took the terrace decades to hold.

The mechanism explains why annual totals mislead. A slope fails when water arrives faster than the soil can pass it on: weeks of ordinary rain load the deeply weathered soils of the Congo–Nile divide, pore pressure builds, and one violent hour can then cut the cohesion that holds a saturated hillside to its bedrock. On steep land the killing variable is intensity, not accumulation. May 2023 followed the script exactly: a saturated April underneath, one extreme system on top.

More variable seasons, with longer dry spells inside them. Rwandan agriculture runs on two rainy seasons: umuhindo, the short rains around September–December, and itumba, the long rains around February–May. The projections and the observations both point the same direction: later onsets, earlier or erratic cessations, and longer dry spells embedded within nominally wet seasons, while total annual rainfall stays flat or rises modestly. The average survives; the calendar dies. Every planting decision in the country is keyed to that calendar.

What they do not agree on

Honesty requires the other column. The famous embarrassment of East African climate science is the "East African climate paradox": most global models project the long rains getting wetter over the coming decades, while the observed record from the late 1990s to the mid-2010s showed them drying, with a run of catastrophic droughts in the Horn.⁵ The candidate explanations (Indian Ocean sea-surface patterns, aerosols, natural variability that the models underweight) remain unsettled. A chronicle that mocked certainty elsewhere will not manufacture it here: the direction of mean rainfall over the Great Lakes at mid-century is genuinely uncertain.

The paradox rewards a closer look, for what it teaches about the tools. The drying that sits behind the Horn's drought emergencies was real and sustained, yet the CMIP-generation ensembles, asked to reproduce those same decades, mostly generate the opposite sign, and carry their wetting forward into the projections. Both escapes are uncomfortable. Either the models misweight a real mechanism (the leading suspects involve Indian and western Pacific sea-surface warming patterns, which the models also struggle to reproduce), in which case their regional rainfall projections deserve limited trust; or the drying was a long swing of natural variability that will eventually release, in which case the wetter future may still arrive, on a date no model can name.

Beneath the paradox sits a humbler limit: the scales the models cannot see. Global ensembles resolve East Africa in grid cells tens of kilometres wide, while Rwandan rain is organised by ridges and lake breezes far below that scale; downscaling sharpens the map while inheriting the parent model's biases. Nor can any ensemble sequence the decades: whether the 2030s or the 2050s carry the worse droughts is not answerable. Anyone selling a district-level rainfall forecast for 2045 is selling something other than science.

The planning consequence is not paralysis; it is the one drawn throughout this movement. Rwanda must prepare for a widened distribution: wetter extremes and drier extremes, in the same decade, sometimes in the same year, rather than for a trend. Water storage, terracing, drainage, and crop diversity are precisely the investments that pay under every branch of the uncertainty. That is what robust means.

The specific threats, ranked by what actually breaks

Three of the threats, in close-up

Three rows deserve the mechanism spelt out, because the mechanism tells a planner which lever moves.

Malaria climbs by arithmetic. Transmission has a thermal floor: below roughly eighteen degrees the Plasmodium falciparum parasite matures too slowly to complete its cycle inside a mosquito's short life, and the Anopheles vectors slow with the cold. Altitude has therefore been Rwanda's quiet epidemiological subsidy, and the highland fringe (the band of settlement just above the transmission ceiling, dense precisely because it was healthy) is where warming spends it, among populations with little acquired immunity. The upslope shift is documented, not merely modelled: Siraj and colleagues, in Science, tracked malaria cases in the Ethiopian and Colombian highlands moving measurably uphill in warmer years. The saving clause is also real: malaria is the most governable threat in the table (nets, spraying, diagnosis and surveillance all work), and chapter 28 counted the health system among the state's achievements. Climate moves the arena's boundary; the contest inside it remains winnable, if funded like one.

Coffee is a thermometer with a price attached. Coffea arabica evolved in Ethiopian highland forest and performs inside a narrow band of mean temperatures around the high teens; quality falls and pests such as the coffee berry borer advance as the thermometer rises. The suitability studies are cautious on numbers and united on direction: a substantial share of the land now growing arabica in East Africa turns marginal by mid-century, while new suitability opens higher up.⁸ On most continents "higher up" is an inconvenience; on Rwandan hills it is a collision with land already farmed by somebody or protected as the montane forest the country restored at real cost. The Ethiopian modelling of Moat and colleagues suggests a coffee sector can ride the shift with deliberate relocation and forest conservation, but that is a migration of slow-bearing trees managed over decades, not something a smallholder improvises. And since coffee and tea earn the foreign exchange that pays for chapter 33's imports, the squeeze is not an agronomic footnote; it is a balance-of-payments problem with roots.

Lake Kivu is the tail risk, and warming is not the trigger. The lake keeps its gas because it is permanently stratified: a density staircase of heat, salt and dissolved gas seals the deep water off from the surface. The precedent is Lake Nyos in Cameroon, which overturned in 1986 and asphyxiated some 1,700 people with roughly a thousandth of Kivu's gas load. The honest reading cuts both ways: gradual surface warming does not by itself unseal a lake stabilised at depth (if anything it strengthens the stratification), and the credible triggers are volcanic and tectonic, as the 2021 Nyiragongo eruption uncomfortably rehearsed. But the deep gas creeps toward saturation over the decades, which is why methane extraction (chapter 33) is the rare investment that is simultaneously baseload electricity and a degassing service: every megawatt burned is pressure the lake no longer stores.

The regional reading: Rwanda's climate is its neighbours

Now the political layer, which the impact tables always omit and which chapter 31 insisted on. Rwanda is small, landlocked, and embedded in a region of a hundred and sixty million people whose climate exposure is worse-governed than its own. Four transmission channels matter more than any isohyet:

Movement of people. The World Bank's regional modelling puts East African internal climate migrants in the tens of millions by 2050. The Horn's droughts already displace millions in bad years; the eastern Congo displaces millions without any help from the climate. Rwanda (one of the few states in the region that functions) will face arrival pressure in a period when, as chapter 27 showed, the rich world has stopped pretending it will share that burden. A country that hosted refugees at 1.5 per cent of its population through the 2010s, and that has been paid by Europe to hold people Europe refuses, should plan for arrival pressure an order of magnitude larger, arriving during its own bad harvests.

The recent regional record shows both ends of the tap. Between 2020 and 2023 the Horn endured five consecutive failed rainy seasons, its worst drought in at least four decades, displacing millions across Somalia, Ethiopia and northern Kenya; in the same years Lake Victoria rose to record levels and flooded its shorelines, and the swelling of Lake Tanganyika and the Rusizi plain drove tens of thousands from their homes in Burundi and the eastern Congo.⁹ Too little water and too much, in adjacent basins, inside four years: the widened distribution wearing faces, and the arrival-pressure scenario Rwanda should be costing now, in the calm.

The corridors. Every litre of fuel and most fertiliser arrives by truck over 1,400 kilometres from Mombasa or Dar es Salaam. Those corridors cross regions (the Kenyan drylands, central Tanzania) with their own worsening drought and flood exposure, and ports whose surrounding cities have their own coming emergencies. Climate does not have to touch Rwanda to cut its supply lines; chapter 33 draws the consequence.

The watersheds. The Congo–Nile divide runs through Rwanda's western mountains; the country's rivers feed both basins, and the Nile's headwater politics (eleven states, one existentially anxious downstream power armed with a longstanding legal claim to most of the water) sharpen with every dry year. Rwanda signed the Cooperative Framework Agreement that Egypt rejects. Water diplomacy in a drying decade is not a technical file; it is a security file.

The precedent next door. The eastern Congo is the standing demonstration of what the region looks like when resource wealth, weak statehood, and external appetite combine, the very forces chapter 31 projected forward. Any regional climate crisis will be fought over that terrain first. Rwanda's involvement there is documented and criticised in chapter 28; the point here is narrower: the region's climate future and its conflict future are the same file, and planners who separate them are planning for a different continent.

What the record shows Rwanda already does

The chronicle's rule is to argue from the record, and the record here is genuinely unusual. Rwanda was the first least-developed country to submit a climate strategy of the modern kind (the Green Growth and Climate Resilience Strategy of 2011), and its updated national plan prices the decade to 2030 at roughly eleven billion dollars, about forty per cent of it adaptation.¹⁰ It banned plastic bags in 2008, hosted the Kigali Amendment to the Montreal Protocol in 2016 (the single most effective piece of climate law yet enacted, done under a Rwandan presidency of the meeting), runs a national green fund (FONERWA) that has moved real money into terracing, wetland restoration, and off-grid power, and has restored forest cover past its thirty-per-cent target.

Two readings of this record are available, and the chronicle takes both seriously. The first: this is exactly the calibrated, nobody-is-coming statecraft that chapter 31 prescribed, begun a decade early. The second: about half of that eleven billion is, by the plan's own text, conditional on external finance, which makes the strategy, in part, a bet on the very pledges this movement assumes will fail. The synthesis is the agenda of the next two chapters: keep the plans, and close the dependency inside them.

The handoff

The forecast, compressed: hotter everywhere and sooner than comfortable; rain that arrives harder and less punctually on land that cannot afford either; disease and crops both climbing the hills; genuine uncertainty about the means and none about the extremes; and a regional environment (corridors, watersheds, neighbours) through which the global rupture, if it comes, will arrive years ahead of the weather.

Nothing in that list is answerable by protest or by summit. Every item is answerable by storage, soil, seeds, watts, medicine, and manufacturing. That is the work of the next two chapters.

Footnotes

1. Per-capita CO₂ emissions from fossil sources: Global Carbon Project / Our World in Data series: Rwanda in the range of 0.1 tCO₂/person/yr against a US figure near 15. ↩
2. IPCC, Sixth Assessment Report, Working Group I (2021), ch. 10–12 and the Africa regional atlas; Working Group II (2022), ch. 9 (Africa): the source for the heat, heavy-precipitation, and variability statements, and for their confidence levels. ↩
3. Republic of Rwanda, Third National Communication under the United Nations Framework Convention on Climate Change (2018): an observed warming trend of roughly 1.4 °C since the early 1970s, with continued warming projected under all scenarios. ↩
4. Government of Rwanda / MINEMA situation reports and international press, 2–4 May 2023: at least 130 dead across Western, Northern and Southern Provinces, with thousands of houses destroyed. On erosion, Rwanda Water Resources Board national erosion-control mapping, which classes a majority of the country's farmland as needing protective works. ↩
5. On the paradox: Bradfield Lyon and David DeWitt, "A recent and abrupt decline in the East African long rains," Geophysical Research Letters 39 (2012); Wenhao Dong et al. and subsequent literature reviewed in W. Yang et al., "The East African long rains in observations and models," Journal of Climate (2014–2015 debate); IPCC AR6 WGI Atlas discussion of East Africa. ↩
6. A. S. Siraj et al., "Altitudinal Changes in Malaria Incidence in Highlands of Ethiopia and Colombia," Science 343 (2014), pp. 1154–1158; Mercedes Pascual et al., "Malaria resurgence in the East African highlands," PNAS 103:15 (2006). ↩
7. On Kivu's gas inventory and hazard: Klaus Tietze's foundational surveys; Martin Schmid et al., "Weak mixing in Lake Kivu," Geochemistry, Geophysics, Geosystems 6 (2005); comparisons with the 1986 Lake Nyos disaster (~1,700 dead from a lake holding three orders of magnitude less gas). ↩
8. Christian Bunn et al., "A bitter cup: climate change profile of global production of Arabica and Robusta coffee," Climatic Change 129 (2015), pp. 89–101; Justin Moat et al., "Resilience potential of the Ethiopian coffee sector under climate change," Nature Plants 3, 17081 (2017). ↩
9. On the Horn drought: WMO and IGAD/ICPAC statements on five consecutive failed rainy seasons, 2020–2023. On the lakes: Lake Victoria's record 2020 levels (exceeding the 1964 maximum); IOM reporting on flood displacement around Lake Tanganyika and the Rusizi plain, 2020–2024. ↩
10. Republic of Rwanda, Green Growth and Climate Resilience Strategy (2011; revised 2023); Republic of Rwanda, Updated Nationally Determined Contribution (2020): US$11 billion to 2030, roughly 40% adaptation, roughly 60% of the total conditional on international support. ↩