Mozambique 2023

Tropical Cyclone Freddy in 2023

In 2023, Mozambique experienced extreme flooding driven by Tropical Cyclone Freddy, one of the longest-lasting cyclones on record at 35 days. Between February and mid-March, the country received more than a year’s worth of rainfall, causing widespread saturation and large-scale inundation. At the flood’s peak, an estimated 18,500 km² was underwater. Freddy made its first landfall on 24 February, delivering 250–300 mm of rain in just three days. After an unusual trajectory change, it made a rare second landfall, bringing intense rainfall exceeding 200 mm per day across Zambezia, Sofala, Manica, Tete, and Niassa—regions already affected by earlier flooding. The combined impacts were severe: 1.1 million people were affected and nearly 200,000 homes destroyed. This use case applies a flood-attribution workflow to assess how climate change may have influenced the event’s severity.

TC Freddy affected 2.7 million people, causing 876 deaths, massive displacement, widespread infrastructure damage—including roads, schools, and health facilities—and major agricultural losses. Flooding also triggered a cholera outbreak, worsening health risks and disrupting essential services across Mozambique, Malawi, and Zimbabwe.

Impact summary

Thoko Chikondi/ActionAid

Fot. Thoko Chikondi/ActionAid

Affected people

2,700,000

Fatalities

876

Injuries

2,925

Displacements

916,000

Houses destroyed

appx. 200,000

Source: Southern Africa: Snapshot of Tropical Cyclone Freddy’s Impact (February - March 2023) | OCHA

Mathias Thole-SAPP

Fot. Mathias Thole-SAPP

ha

Affected agricultural land

Affected livestock

Source: Southern Africa: Snapshot of Tropical Cyclone Freddy’s Impact (February - March 2023) | OCHA

Lost tourism days due to the flooding in Somerset

Image captured Mar 7, 2023, by the VIIRS instrument aboard the joint NASA/NOAA Suomi NPP satellite

Geographical context

TC Freddy impacted Mozambique, Zimbabwe and Malawi. Freddy started south of Indonesia on February 2nd, 2023, and moved westward across the Indian Ocean. It took 19 days to make its first landfall in Madagascar After this, Freddy re-intensified over the Mozambique Channel, making landfall in Mozambique and dissipating on 24 February 24th. However, on March 2nd, Freddy regenerated over the Mozambique Channel and changed course toward Madagascar before turning back toward Mozambique. After making landfall, it finally dissipated on March 13th after making landfall. 

Our use case analysis focuses on the second landfall of Freddy on the mainland of Mozambique and is based on the compound flooding and resulting direct damages. We do not consider direct damage from wind or waves, although these drivers may have contributed as well to the impacts of Freddy.

01

COMPOUND HAZARd

Tropical Cyclone Freddy 

Mozambique and neighboring southern African countries are highly vulnerable to tropical cyclones due to their location along the southwest Indian Ocean cyclone corridor and the presence of extensive low-elevation coastal zones and inland flood basins. The region’s history of destructive storms (including Cyclones Idai and Kenneth in 2019) is compounded by systemic vulnerabilities: high poverty levels, food insecurity, weak infrastructure, political fragility, and social inequality. These systemic risks are amplified by repeated cholera outbreaks and constrained health systems during disasters. Moreover, river basins such as the Licungo and Zambezi have experienced repeated seasonal flooding, exacerbated by deforestation and land-use change.

Freddy impacted the coastal areas of Mozambique through a combination of high precipitation, storm surges and, in some areas, high river discharges, making it a multivariate compound flooding event. In addition, the two landfalls in different areas of the country could be defined as a spatially compound event.

Tropical Cyclone Freddy developed under oceanic and atmospheric conditions that were more favorable than historical averages. Sea surface temperatures exceeding 27°C, weak vertical wind shear, high relative humidity, and strong TC heat potential enabled repeated intensifications. Freddy accumulated the highest recorded cyclone energy globally and became one of the longest-lasting systems in history. After forming near Indonesia, Freddy crossed the Indian Ocean for 19 days before its first landfall in Madagascar. The cyclone then re-emerged over the Mozambique Channel, where warm waters and high moisture flux allowed it to re-intensify before making its first landfall in Mozambique on 24 February.

This first landfall brought torrential rainfall (250–300 mm in three days) triggering widespread flooding in central provinces, many of which had already experienced flooding due to heavy early-February rains along river basins such as the Licungo. Though coastal storm surge reached only about 0.3 m, the cyclone also contributed to flooding in low-lying coastal cities such as Beira.

On 2 March, Freddy regenerated and stalled in the Mozambique Channel due to a combination of subtropical anticyclones and broader regional anomalies including a positive Sub-tropical Indian Ocean Dipole (SIOD) and Southern Annular Mode (SAM) phase and transitioning ENSO conditions. It looped back toward Madagascar before turning once more toward Mozambique. On 11 March, Freddy made its second landfall in east-central Mozambique, near Quelimane, bringing daily rainfall exceeding 200 mm and cumulative totals up to 1700 mm across the event period. The stalling phase near the Malawi border again intensified upstream river discharge, contributing to further inland flooding.

Riverine flooding dominated impacts, as spatial analyses show deep inundation of 4–5 m in low-lying deltas and along major channels. With large regions facing over 1000 mm cumulative rainfall, flooding persisted for weeks and creating cascading health and logistical challenges like cholera outbreaks, fragmented roads, and displaced people.

Observed impacts

Freddy produced a sequence of impacts across Mozambique, Malawi, Zimbabwe, and Madagascar

Freddy produced a sequence of impacts across Mozambique, Malawi, Zimbabwe, and Madagascar, evolving across several phases of extreme rainfall, inland flooding, and repeated re-intensification. Its first impacts quickly followed the initial landfall in Mozambique on 24 February. The 250–300 mm of rain that fell over three days rapidly saturated river systems and triggered or compounded flooding in central provinces already affected by heavy rainfall since early February. Rivers began to overflow, agropastoral lands were inundated, and communities experienced displacement as roads were cut.

Following dissipation, Freddy regenerated over the Mozambique Channel on 2 March, looped back while it re-intensified, and made a second landfall on 11 March near Quelimane. This second landfall brought extreme daily rainfall exceeding 200 mm and cumulative totals up to 1700 mm across affected regions. Flooding intensified across Zambezia, Sofala, Manica, Tete, and Niassa, with deep inland inundation reaching 4–5 m in low-lying deltas and along major drainage channels.

At its peak, an estimated 18,500 km² was underwater, and approximately 2.7 million people were cumulatively affected, with nearly 200,000 houses destroyed. Infrastructure suffered widespread losses: 5,967 km of roads were damaged, nearly 1,900 schools were impacted (with many converted into emergency shelters), and 223 health facilities across three countries were affected. Homes in river flood plains around Quelimane experienced the most severe and prolonged flooding, reflected in exposure estimates of 187,000 people facing flood depths exceeding 5 cm and 128,000 exposed to more than 20 cm.

Agricultural losses were significant. In Malawi alone, over 204,800 hectares of crops were flooded just before harvest, amplifying food insecurity. The destruction of cropland, livestock losses, and disrupted transport networks generated cascading economic impacts. A cholera outbreak in Mozambique, spurred by flood damage, compounded the crisis with the spread of disease.

Displacement surged to 916,000 people. As waters receded after mid-March, long-term impacts became evident: extensive damage to agropastoral land and more than 200,000 homes, fragmented roads, increased disease and health risks, and economic damages estimated at USD 114 million. These impacts unfolded in a context of high baseline vulnerability, meaning even incremental increases in flood depth produced outsized social and economic consequences.

Response

In response to Cyclone Freddy’s impacts in Mozambique, the government and humanitarian partners activated early warning systems and community alerts to move people to safer areas before landfall, which helped limit loss of life compared with past storms such as TC Idai. Investments in risk communication and local disaster committees facilitated these actions. Mozambique received coordinated international assistance under a joint response plan addressing Floods, Cholera, and Cyclone Freddy, with support from agencies such as the Red Cross, UNICEF and the World Food Programme providing emergency supplies, water and sanitation support, food assistance, shelter materials, and child protection services. Funding was also allocated to water, hygiene, food, and shelter sectors. Despite challenges from concurrent cholera outbreaks and infrastructure damage, ongoing efforts included hygiene promotion, distribution of learning and hygiene kits in schools, and broader multi-sectoral relief activities to support affected communities. 

Attribution analysis 

The attribution analysis for TC Freddy uses a storyline approach that compares a factual simulation of the flood event, representing present-day climate conditions, with a counterfactual simulation representing a world without human-induced warming. The analysis focuses specifically on precipitation because rainfall was the dominant driver of flood impacts in Mozambique. 

In the factual scenario, hydrological and hydrodynamic models are forced with ERA5 reanalysis data and best track cyclone information. These datasets and models produce estimates for river runoff, coastal water levels, and rainfall to compute compound flooding and associated direct damages and exposed population.

In the counterfactual scenario, long-term climate trends are removed from precipitation using the Clausius–Clapeyron relation, which assumes that extreme rainfall decreases by approximately 7% per 1°C of cooling. For this case, an 8% reduction in precipitation is applied to approximate pre-industrial climate conditions. All other flood drivers are kept constant.

Attribution results are derived by comparing flood extent, flood depth, exposed population, and economic damages between the factual and counterfactual worlds, producing an event-specific assessment of climate change influence. 

Attribution results

The attribution results for Tropical Cyclone Freddy demonstrate that climate change measurably increased the severity of flooding and associated impacts. 

Influence on Flood Hazard

The primary climate change signal appears in rainfall. The factual scenario incorporates precipitation that is 8% higher than what would be expected in a pre-industrial world. This increase leads to amplified runoff and deeper floodwaters across central Mozambique.

When comparing simulations:
Flood extent increased by 3% in the factual scenario. This indicates that climate change marginally expanded the geographic footprint of inundation. However, the spatial pattern reveals that the greatest increases occurred along tributaries of the Licungo River, where upstream basins experienced intensification of rainfall-driven flooding.

Mean flood depth increased by 0.15 m, with localized increases exceeding 0.5 m. These increases in depth are particularly important. Depth-damage curves used in the study show that even small additional depths substantially elevate damage levels and population exposure.

Flood maps show that the largest increases in depth align with low-lying inland areas and drainage channels, rather than coastal zones, implying the dominance of direct precipitation and riverine flooding over storm surge during this event. 

Influence on Flood Impacts

While hazard-level increases due to climate change were modest, the impact consequences were proportionally larger:

Total economic damages increased by USD 21 million, a 19% increase attributable to climate change. Under factual conditions, total direct damages were estimated at USD 114 million, meaning that nearly one-fifth of this loss can be linked to human-induced warming.

Number of buildings damaged increased by 8%. The analysis indicates that this increase is driven more by greater flood depths than by expansion of flood extent. In other words, buildings that would have been flooded anyway were flooded more severely in the warmer-climate scenario.

Population exposure to severe flooding increased significantly. For flood depths exceeding 20 cm (a level associated with substantial damage and mobility disruption) exposure increased by 13%, representing approximately 16,000 additional people affected attributed to climate change.

These results highlight that the influence of climate change can be more pronounced on the flood impacts than on the flood hazard, especially in regions of high vulnerability where even small increments in depth translate into large social and economic consequences.

Comparison of population exposure

Comparison of building damage

Interpretation and Context

The analysis acknowledges that this attribution is conservative, given that other studies have shown that the impact of climate change on extreme rainfall may go beyond the Clausius-Clapeyron rate, meaning the true climate change influence could be higher. Moreover, the counterfactual adjustment considers only a reduction in precipitation, other climate change-driven trends that could modulate the event (e.g. changes in cyclone track, intensity, or translation speed; sea-level rise). It is important to emphasize that the attribution analysis is focused on the inland flooding, as the result of the second landfall in Mozambique. The real-world impacts of Freddy go beyond that, since there were also major impacts in Malawi and other regions. 

Finally, the analysis notes that tropical cyclone attribution is particularly challenging in data-scarce contexts, especially since cyclones are low-probability events that are poorly modeled. Mozambique’s limited observational data increases uncertainty. Despite these limitations, the results clearly indicate that increased temperature contributed tangibly to the severity of TC Freddy’s flooding and resulting impacts.

Policy implications 

For the case of TC Freddy, COMPASS did not carry out any stakeholder engagement activities or investigate the policy context. For more about the context of Mozambique, we refer to the case of TC Idai and Kenneth.