Between 14 and 16 December 1999 the mountain front above Venezuela’s central coast shed thousands of shallow landslides that gathered into debris flows and ran out onto the alluvial fans where most of Vargas state lived. Neighborhoods were buried under meters of boulders and sand, the coastal strip was cut off from Caracas, and survivors were taken out by ship and helicopter over the following weeks. Published death tolls for the event range from roughly one thousand to roughly thirty thousand, depending on who is counting and what they are counting. This piece covers the terrain, the evacuation that never happened at scale, the maritime lift, and the relocation that followed.

The mountain front and the fans it built

The Cordillera de la Costa rises out of the Caribbean immediately behind Venezuela’s central coast, and it rises fast, with Pico Naiguata reaching about 2,765 meters within roughly ten kilometers of the shoreline. Slopes on that front are commonly steeper than thirty degrees and in places much steeper, the soil mantle over the metamorphic bedrock is thin, and the drainage consists of short ravines, the quebradas, that fall from the ridge to the sea over a very small horizontal distance. Caracas sits on the inland side of the same range at about 900 meters, connected to the coast by one mountain highway.

What people call the coastal plain in Vargas is a string of alluvial and debris fans built at the mouths of those quebradas, coalescing into a narrow apron of flat ground between the mountain and the water, rather than a plain in any geological sense. The San Julian fan carries Caraballeda and Los Corales, the Uria fan carried Carmen de Uria, and the same arrangement repeats at Macuto, Tanaguarena, Naiguata, Camuri Grande and along the rest of the coast. Every buildable flat surface on that shoreline was manufactured by the process that would run over it in December 1999.

By the late 1990s that apron carried Venezuela’s principal international airport at Maiquetia, its principal seaport at La Guaira, mid-century resort and apartment development including reinforced concrete towers of ten stories and more, and a very large stock of informal housing that had spread across the fan surfaces and climbed the banks of the ravines themselves. Venezuelan census figures around that period put the population of Vargas state near 300,000. In several places structures had been built inside the channel, on the bed the flow uses when it comes.

Two weeks of rain and then three days

Rain fell across the coastal range through the first two weeks of December 1999, saturating the soil mantle without producing much visible trouble, and then an extraordinary burst arrived on 14, 15 and 16 December. The gauge at Maiquetia airport recorded roughly 911 millimeters across those three days, with something on the order of 380 millimeters on 16 December alone, figures reported in the U.S. Geological Survey work on the event by Matthew Larsen and Gerald Wieczorek and their Venezuelan colleagues. The same sources give the site’s mean annual rainfall as roughly 520 millimeters, so the three-day total was close to twice what the location normally receives in a year.

The failure sequence is worth stating in mechanical terms because it governs what warning is possible. Antecedent rain brought the thin soils on the steep slopes close to saturation, the intense rain then triggered thousands of shallow slips that detached soil and vegetation down to rock, and those slips fell into steep channels already loaded with sediment from previous storms and previous events. Once a slip enters a channel it bulks, picking up stored boulders, sand and water until it moves as a slurry with something like the density of wet concrete, traveling at several meters per second and carrying blocks the size of vehicles in suspension. It stops when the channel loses confinement at the fan apex, and it deposits its load across the fan, which is where the towns were.

The explanation that circulated afterward blaming deforestation for Vargas is largely a myth, and it needs to be labeled as one, because much of the upper catchment lies inside the protected area then known as El Avila National Park and the slides were mapped on forested slopes within it. Root strength does not prevent shallow failures at that rainfall intensity on that gradient. Human land use mattered enormously in this event, but it mattered on the fans and in the channel bottoms where the housing was, rather than on the ridges.

The record in the ground and the record on paper

Damaging debris flows and flash floods on this coast are documented well back into the colonial period, with events in 1798, 1912, 1938, 1944, 1948 and 1951 among those that appear in Venezuelan hazard studies, and the February 1951 event killed people in the same towns that were hit in 1999. The interval between large events on any single fan is long enough that no living resident had seen one, and short enough in geological terms that the landform is actively growing. That combination produces exactly the settlement pattern Vargas had, because the flat ground looks permanent to anybody who arrives between events.

The hazard was also on paper. A national inventory of geological hazards published in Venezuela in 1983 by Andre Singer, Carlos Rojas and Miguel Lugo, produced through the country’s seismological research foundation FUNVISIS, compiled the historical record of these coastal debris flows and identified the settlements on the fans as exposed to them. I would send anyone who needs the primary document to FUNVISIS rather than rely on my summary of it, but the essential point does not depend on the details: sixteen years before the disaster, Venezuelan scientists had written down which towns sat on which fans and what those fans were made of.

Beyond the literature, the evidence was lying in the streets. Boulders several meters across sat on the fan surfaces before 1999, some of them incorporated into walls and foundations, and their size records the size of the flow that delivered them, because water alone does not move a hundred-ton block across flat ground. When you can see boulders on a fan that no local flood could have carried, you are looking at the deposit of a debris flow, and the depth of the deposit tells you how much material the last event delivered.

What arrived on the fans on 15 and 16 December

The large flows came through during the night of 15 December and into 16 December, when almost everyone on the coast was indoors, and they did not arrive as one wave. Surges came down the quebradas repeatedly over hours as successive landslides fed the channels, so a building or a street that survived the first pulse was hit again, and the sediment that stopped in the channel raised the bed and pushed later surges out across new ground. People who tried to move during the night were moving across a fan that was being resurfaced while they crossed it.

Caraballeda and Los Corales, on the San Julian fan, produced the images that traveled around the world, with boulders reported up to roughly ten meters across and weighing in the hundreds of tons deposited in the streets and driven through the ground floors of reinforced concrete apartment towers. Deposition across the fan ran to several meters in places, buildings were filled with sediment to the second floor, and the shoreline advanced seaward as the fan grew. Carmen de Uria was buried, and the town was subsequently abandoned rather than excavated, which matters directly to the counting problem discussed further down.

The infrastructure failures were of the kind that convert a local disaster into an isolated one. The coastal road crosses every quebrada, which meant it was cut at every crossing, and the mountain highway to Caracas was blocked by slides and washouts during the same storm. Mud and water reached the airport at Maiquetia, and the port at La Guaira sustained damage. Venezuelan reconstruction reporting generally puts destroyed housing on the order of 8,000 units with tens of thousands more damaged, and those figures come from post-event government assessment rather than from any independent audit.

The evacuation that did not happen at scale

Warnings existed in the general sense that civil defense authorities were aware of the rain and that news coverage reported flooding, and some local, small-scale movements took place. What did not exist was a rainfall threshold tied to a defined action, a mapped evacuation zone matching the fans, identified routes and receiving sites, and an authority with both the legal power and the transport to move a population of that size before the flows arrived. Under those conditions an alert reaches people as information rather than as an instruction, and information about rain in a place where it has been raining for two weeks does not move anyone.

The geography made the decision expensive even for someone willing to make it. One coastal road strung along the fans and crossing every ravine, one mountain highway to Caracas that ran through the same failure zone, and the Caribbean at the back gave roughly 300,000 people very little egress capacity, and both land routes were destroyed by the event they would have been used to escape. Vertical movement inside the concrete towers was ambiguous as well, because going upstairs is a reasonable response to water and a poor response to a flow that removes the ground floor and scours the foundation.

Venezuela held a national constitutional referendum on 15 December 1999, in the middle of the rainfall, and both official attention and news coverage that week were substantially occupied by it. I am not going to argue that the referendum explains the absence of an evacuation, because the absence of any debris flow evacuation planning for those fans is a sufficient explanation on its own, and a competing national event only tells you how thin the margin was.

The death toll, and why the range is so wide

The high end comes from Venezuelan government and civil defense statements in the weeks after the event, which gave figures around 30,000, with some early official comments running as high as 50,000. That 30,000 figure entered the international disaster databases, including EM-DAT, maintained by the Centre for Research on the Epidemiology of Disasters at UCLouvain, and it is the number most often quoted in general reference works. A figure of about 15,000 appears widely in technical and scientific literature on the event, including publications associated with the U.S. Geological Survey work on the debris flows.

The low end comes from Rogelio Altez, a Venezuelan historian and anthropologist at the Universidad Central de Venezuela, who examined the documentary record, the recovery of remains and the population accounting and argued that the number of dead was on the order of a thousand or fewer, and that the larger figures were never supported by bodies, records or any reconciled list. These sources disagree, and they disagree about method rather than arithmetic, because one end of the range is an official estimate of people presumed missing from settlements with no reliable population baseline, while the other end is a count of deaths for which some documentation exists. Averaging them would produce a number that nobody stands behind.

Several specific conditions made an accurate count impossible. The informal settlements on the fans and ravine banks had no dependable population register, occupancy was partly transient, an unknown number of bodies went into the sea, and an unknown number remain under deposits that were never excavated, Carmen de Uria being the clearest case. Survivors were then dispersed to shelters across the whole country within days, which broke the link between a missing-person report and a community that could confirm it. When I use this event in a briefing I give the range, name who publishes each end, and say in the same sentence that the sources disagree.

Getting people off the coast by sea and air

With the coastal road cut at every ravine and the highway to Caracas blocked, the Vargas coast became an island for practical purposes, and the evacuation of survivors ran over water and by helicopter for weeks. The Venezuelan Navy, merchant vessels and private boats moved people out through La Guaira and off beaches, while helicopters lifted people from rooftops and from cleared ground and staged them toward Caracas and other receiving points. Published totals for people displaced vary widely, from tens of thousands into figures well above one hundred thousand, and I would not present any single one of them as settled.

A maritime evacuation at that scale demands more than boats. It needs a usable quay or a beach that a landing craft can work, medical triage at both the loading point and the receiving point, crowd control at a place where a large number of frightened people are converging on a small opening, a way to keep families together, and above all a manifest. The absence of a manifest in Vargas fed directly back into the counting problem, since a population scattered to shelters in a dozen states without documentation cannot be reconciled against a list of the missing afterward.

International assistance arrived quickly and included aircraft, supplies and offers of engineering support. In January 2000 the Venezuelan government withdrew its consent for a planned United States military engineering deployment that was already at sea, an episode that was widely reported at the time and that anyone who needs the detail should read in that contemporaneous reporting. The general planning point for anybody who builds international or interstate assistance into a concept of operations is that consent is granted by a political authority and can be revoked by one, so the plan needs an alternative that does not depend on it.

Relocation, sediment dams, and what came back to the fans

Displaced families went into shelters in military installations, schools and public buildings across the country, and the government moved substantial numbers to other states, with reporting at the time naming destinations including Barinas, Apure, Guarico, Anzoategui and Zulia, along with newly built settlements. Some families remained in temporary shelter for years, and others occupied buildings in Caracas. Relocation over that kind of distance separates people from their work, their schools and their extended families all at once, and a significant share of relocated households returned to Vargas over the following years because the port, the airport, the tourism economy and their relatives were all on that coast.

Venezuela created a state corporation for the recovery and development of Vargas, CORPOVARGAS, in 2000 to run reconstruction, and the engineering program that followed centered on sediment retention structures and channel works in the main quebradas, including San Julian, an approach drawn from Japanese sabo practice with Japanese technical cooperation involved in the study work. These structures do real work by capturing coarse sediment upstream of the fan apex and by reducing the bulking that turns a landslide into a large flow. Their capacity is finite, they fill with sediment and must be excavated on a schedule, and a maintenance program that lapses converts a mitigation asset into a full basin sitting above a town.

Settlement returned to the fan surfaces, and in February 2005 heavy rain on the same coast produced further debris flows and flooding, with fatalities, evacuations and renewed damage in the same municipalities. Vargas was renamed La Guaira state in 2019, and anyone with a current operational or planning interest there should get present-day hazard and capability information from Venezuelan authorities rather than from a retrospective article. The transferable finding is that structural mitigation and land use control are not substitutes for one another, and that Vargas got a partial version of the first while the second largely did not hold.

What to do at your agency

• Have the emergency manager send one written request this month to the state floodplain manager or NFIP coordinator asking whether any alluvial fan flooding areas or debris flow hazard zones are mapped in your county, and file the answer in the hazard analysis section of the mitigation plan.
• Ask your GIS analyst for a map of every structure within one hundred feet of a drainage channel that leaves a slope steeper than your state geological survey’s threshold, and bring that map to the next scheduled hazard mitigation plan update meeting.
• If your jurisdiction contains a burn scar, put the post-fire debris flow products on the agenda of your next coordination call with the National Weather Service forecast office, and leave that call with the specific rainfall intensity thresholds written into your plan next to the name of the person who orders movement.
• Have the emergency manager list every community in the jurisdiction served by a single road, count the population in each, and write one paragraph into the existing evacuation annex stating the trigger for moving those communities early and the estimated hours required.
• Ask the fire chief or emergency manager to contact the Coast Guard sector, the port or harbor authority and any local marine unit, confirm in writing who can move passengers by water and from which landing points, and record it in the resource annex with a twenty-four hour contact number.
• Verify with your mass care lead that the shelter registration process produces a usable manifest, run it on paper during the next shelter drill without network connectivity, and confirm that the resulting list can be handed to a reunification function.
• Add one item to the next building official or plan review meeting asking whether the code and zoning as adopted permit habitable structures within mapped fan, channel or debris flow zones, and get the answer in writing for the mitigation plan.

Takeaways

• Rainfall at Maiquetia over 14 to 16 December 1999 is reported in the USGS work on the event as roughly 911 millimeters against a mean annual total near 520 millimeters, following two weeks of rain that had already saturated the thin soils on the coastal range.
• Thousands of shallow landslides mobilized into debris flows in the steep quebradas, bulked with stored channel sediment, and deposited boulders reported up to roughly ten meters across on the fans where the towns of Vargas were built.
• Every flat buildable surface on that coast is an alluvial or debris fan, which means the landform itself is a record of the hazard, and Venezuelan hazard studies including the 1983 inventory by Singer, Rojas and Lugo had documented the history sixteen years before the disaster.
• Published death tolls disagree sharply, with Venezuelan official statements and the EM-DAT database giving about 30,000, technical literature frequently repeating about 15,000, and Rogelio Altez of the Universidad Central de Venezuela arguing from the documentary record for a figure on the order of one thousand or fewer.
• The count cannot be settled because the informal settlements had no reliable population baseline, bodies went to sea or remain under deposits that were never excavated, and survivors were dispersed nationally within days without manifests.
• No large-scale evacuation occurred because there was no rainfall threshold tied to a defined action, no mapped evacuation zone matching the fans, and no route out that would survive the event, since both the coastal road and the mountain highway to Caracas were destroyed by the same storm.
• Blaming the disaster on deforestation is a myth in this case, since much of the source area lay inside a protected forested park and the failures occurred on forested slopes, while the decisive human factor was housing built on the fans and inside the channels.
• Reconstruction produced sediment retention structures in the main ravines with Japanese technical involvement, but settlement returned to the fan surfaces, and further debris flows on the same coast in February 2005 caused fatalities and damage again.