Didier Richard(1), Pierre Usselmann(2).
(1)Cemagref Snow avalanches engineering and torrent control Research Unit, Grenoble, France. Mail: Tlf: +33 / 4 76 76 27 73 Fax: +33 / 4 76 51 38 03
(2)CNRS UMR 5651 ESPACES / GEMS, Montpellier, France. Mail:
After the catastrophic event of December 1999 in Vargas it is of course of prime importance, in order to conceive prevention countermeasures, to asses the probability that a same event could occur again. It is especially crucial in the Vargas case because a similar event already affected the Vargas coast in February 1951. The fact that the 1999s event could have been even more extreme than the 1951s one (at least it has been undoubtedly much more murderous) reinforce the importance of the question: under which conditions is such an extreme event like the Vargas 99 floods repeatable?
The authors have only realized a one-week mission in Vargas in February 2000. Therefore they can of course not pretend to give a definitive answer to such a complex question. Nevertheless some tracks can probably be advanced that can maybe provide elements on this topic.
After a short summary of the most remarkable characteristics of the 1999s event, a list of possible investigations is drawn up, as they have been identified at the date of the mission.
On December 15 and 16, 1999, exceptional rainfalls fell down on the northern coast of Venezuela, especially affecting the northern slope of the El Avila Mountain. Furthermore these rainfalls occurred after more than two months with rainfalls significantly higher than the mean values. The following data, recorded at the Maiquetia Airport, are mentioned in a report compiled by PNUD (2000): 122 mm in October 1999, compared to a mean inter-annual value of 56 mm, 290 mm in November 1999, against 54 mm, and of course 1204 mm in December 1999, against 54 mm. The rainfalls of December 15 and 16 occurred thus probably on saturated soils, or at least soils much more wet than usually at this season.
All the torrents of the northern slope of El Avila, more or less perpendicular to the coast, knew exceptional floods, burying almost all their alluvial fans under huge volumes of sediments.
The damages were considerable, both for what concerns human losses and on the material or economic plan. Many houses were swept, buried or filled by sediments. The road giving access to the littoral localities remained impracticable for a long time. The water supply and other networks were hard affected, as well as the La Guaira seaport. The torrential flows often pushed their sediment transport quite far in the sea (around 100 meters at some places).
The most affected zone is located between Maiquetia at the West and Anare at the East; about fifty torrents are involved, from which about twenty were very touched. The concerned watersheds are all coastal catchments located on the northern slope of El Avila Mountain, which culminates at more than 2700 m (Pico Naiguata). The crest of El Avila is more or less parallel to the coast, at less than 10 km from it. Thus the slopes are very steep. The El Avila Mountain is rather strongly forested. Many very large trunks have been found in the flood deposits.
Some characteristics of the events of December 999 can be pointed out here, as they seem to be particular enough.
The volumes mobilized, transported and deposited on the alluvial fans are considerable. Between La Guaira and the eastern part of Caraballeda, sediments have been deposited almost everywhere along the littoral road, burying the houses and buildings with a thickness of more or less one stage. Similar deposits are also present more in the East, mainly at the outlets of the torrents.
Size of the transported boulders:
In some cases, sediments arriving on the alluvial fans contain very coarse boulders. For instance, in Caraballeda at the outlet of Rio San Julian, many boulders have diameters of more than 2 m, some of them reaching the coast. Transported vegetation debris (trunks in particular) also reach considerable dimensions, in diameters and lengths. For instance trunks of more than one meter in diameter and 20 meters long have been found.
An important variability in the situations:
Despite a certain general homogeneity of this northern side of El Avila, each quebrada functioned with its own characteristics, which result in very variable situations from one catchment to the other. The following illustrations can be given as examples:
Between the 14th and 16th of December 1999, the rainfall on the Vargas State reached more than 900 mm, which is more or less the average annual rainfall in Caracas. It rained already a lot at the beginning of December, especially the 2nd and 3rd (about 200 mm). That probably prepared the saturation of the soils. Data compiled from the Maiquetia Airport station (PNUD, 2000) mention rainfalls of 293 mm from December 1 to 13. Furthermore, as mentioned above, November and October 1999 had already known rainfalls significantly higher than the inter-annual averages.
Therefore a very detailed attention should be paid, in any hydrological analysis of the phenomenon, to possible characteristics of the December 1999 event with respect to duration. Rainfall durations of one day, or even shorter (15/12), 3 days (14/12 to 16/12) 15 days (first half of December), three months, should be analyzed.
There are probably no discharge data. Even if there were measuring sites, they surely did not resist to the December 1999 floods.
The usual way to qualify the occurrence frequency of a flood event is to carry out a statistical analysis on the recorded hydrological data that can be gathered on the concerned zone. This kind of hydrological analysis usually provides a suitable assessment of the exceptional character of a given flood event, through the well-known and quite pertinent concept of return period.
The assessment, even rough, of the return period of past events is always of the highest importance. One will of course not consider in the same way a current event likely to reproduce on average every 10 to 20 years, as an exceptional event, which should repeat only on average every 500 to 1000 years.
The December 1999 events were surely exceptional, at least by the number of victims. Nevertheless the assessment (even coarse) of their return period remains to specify. In a work published by the PNUD (2000), the return period of the December 15 and 16, 1999, is estimated at 1000 years, without much more precision. A recent report of Franceschi (2000) presents this 1000 years estimate like the extrapolation from a Gumbel diagram proposed by Ayala (1978) in a quite old hydrological study.
Unfortunately it seems that in spite of the quite large region affected by the 1999 event, the available data are crucially poor. Apparently the only working station at this moment was the Maiquetia Airport climatic station, that is at the sea level, and on the western margin of the affected zone.
The situation in terms of hydrologic and climatic network should surely be improved in a region so regularly affected by heavy rainfalls and floods, and now so vulnerable because of the urbanization density.
As rare as can be the 1999 episode, it must also be considered that an already serious event of the same type occurred less than 50 years ago. Effectively, a similar event occurred at the same place, in February 1951. The climatic process could be the same one, and uninterrupted rainfalls of about 529 mm in 60 hours are mentioned (PNUD, 2000). There had been "only" 700 dead along the Litoral Central, but the railway and the highway of Caracas had been already cut at several places. In fact, the density of the urbanization on this coast was much lower (only the roadway systems preceded the current density of urbanization).
Moreover it seems that a National Inventory of the Geological Risks exists, containing several references of events of this type between 1740 and 1983.
At least a comparison as fine as possible between the 1951s and 1999s events should be carried out in order to precise why, or according to which criteria, the 99 event was more extreme than the previous one. For instance the following points can be investigated:
In default of the possibility to carry out a rigorous hydrological analysis, an historical approach could maybe give some results, as many events seem to have been recorded in the past since the 18th century. If these events are enough documented, maybe an empirical return period could be estimated from this analysis.
As the 99 event is obviously impressive by the huge amounts of transported sediments, a quite detailed estimation of these volumes should be achieved. Such an estimation is necessary to achieve comparisons between the volumes mobilized on the slopes and the volumes respectively deposited on the alluvial fans and pushed away in the sea.
The evaluation of deposited volumes is essential for designing adapted countermeasures. It is moreover not very complicated in very urbanized zones where many reference level marks are present. The aerial photographs recorded in the days following the catastrophe can of course contribute to this evaluation.
For instance the affected zones in Caraballeda have been estimated from this aerial cover to 100 ha ("severe damage or total destruction" - source: El Nacional - 15/01/2000). If we consider an average deposits thickness from 1,5 to 2 m (a little less than one stage), we obtain in a first approach an evaluation of the deposited volume about 1,5 to 2 millions m3. The use of image processing techniques could allow refining this evaluation.
It would be extremely interesting to provide comparison between these volumes and volumes known, evaluated or estimated, at the time of former events. The aerial photographs recorded after the February 1951 event should allow an evaluation of the volumes of this event as precise as that for the 1999 event. For other older events, a thorough work, with more random results, would be necessary.
A map at the scale 1:100 000 was worked out (FUNVISIS) from helicopter overflights carried out between 26/12/99 and 5/01/00, distinguishing the following features:
This map does not allow appreciating the respective contributions of slopes and of the beds of the torrents in the sediment production. One can think however, in spite of locally very significant scouring phenomenon, that the major contributions are due to slope processes.
The question of the relative contributions of the various sources of sediments is obviously interesting for what concerns the understanding of the phenomenon, and therefore for what concerns the assessment of future floods. This question however remains open. For instance the very numerous scarps that took place at the basis of the catchments are of an orange-brown color, very different from the grey color of the sediments accumulated in the beds of the downstream reaches and on the fans.
No element either to date allows to make comparisons in term of volumes, on the dynamics of sediments according to their size or their origin. Indeed, a significant, but not quantified, part of the sediment supply from the December 1999 floods reached the sea, either as deposits, which modified the coast shape, or transported by the marine currents.
The evaluation of the volumes exported from the slopes would be useful, although certainly rather delicate. On the other hand, contributive surfaces could certainly be rather easily deduced from the processing of aerial photographs or satellite images. A first estimate of volumes could be deduced from it, assuming an average thickness mobilized by the different types of slope movement, deduced for instance from field observations.
A survey of the slope processes providing sediments to the torrents is thus certainly necessary to estimate in more details the part of the volumes mobilized on the slopes. A geomorphological and geotechnical approach of these processes, characterizing the failure conditions of the involved soils, is possibly a way to precise the intensity and dynamics of the phenomenon.
As usual arises the question of the nature of the flows. The size of the transported boulders, their position sometimes (at the level of the third floor of a building!), and certain typical forms suggest the occurrence of debris flows. The identification of typical forms, like lateral and frontal levees, is however complicated by the spreading extent, the presence of the sea, of constructions, the succession and alternance, obviously, of coarser, finer, more liquid flows...
The sights of the La Guaira port, on the contrary, if we except the presence of a few sporadic boulders, suggest a water flow, strong enough to push the containers and of course very charged in suspension, more than an abundant mud contribution in the form of debris flow.
Some marks of flow on houses built on the banks along the alluvial fans suggest also rather liquid flows, nevertheless considerable. Finally, even in some boulders deposits, structures in shape of "piles of plates" rather refer to bed load sediment transport.
A short calculation, considering about 1 000 mm rainfall on a catchment area of 20 km2 (approximate area of the Rio San-Julian catchment) and a runoff coefficient of 60 %, give a flow of about 12 millions m3. Thus the solid contribution, evaluated before at 1,5 to 2 millions m3, would represent about 15 % of the water runoff. This is much closer to an order of magnitude coherent with bed load transport than with debris flow. The maximum transport capacity Qs/Ql for bed load transport at a slope of 10 % is about 10 %. That seems to indicate that the majority of the flows of December 1999 generated a solid transport by torrential bed load.
Nevertheless, it is rather probable that failures of large gneiss plates in the intermediate stage of the basins, supplying huge quantities of sediments arriving suddenly in the steep gullies that could be pushed by strong liquid flows, have generated debris flows. These debris flows then were probably preceded by very large boulders resulting from the fragmentation of the gneiss plates, which initiated at the time of their failure when they fall in the ravines, and continued along the hydrographic channel.
The flows of December 1999 in the torrents of the Northern side of El Avila Mountain thus probably were mainly strong flows with intense bed-load transport. Some debris flows surges transporting very big boulders occurred also however, more or less at the rhythm of the rock-falls in the intermediate gneiss stage.
Some marks of strong scouring can consolidate the assumption of flows with bed-load transport. Quebrada Uria for instance, after having come out of its bed, invaded almost all the Carmen de Uria village. It deepened then until reaching the rock base, several meters under the former ground level. There is not far from 10 meters between this rock base level, and the highest traces that the flow left at the top of the second stage on white walls in houses damaged by the flow. It is also probable that partial collapses of buildings in the district Los Corales in Caraballeda are due to the weakening of foundations due to an excessive scouring during the flood.
The characterization of the failure scenarios, that a geotechnical study as mentioned above could attempt to establish, could also contribute to slice between debris flows and bed-load solid transport. Roughly speaking, sporadic failure scenarios implying significant volumes each time would be rather associated with debris flow type, while more progressive erosion processes (possibly even by slopes failures spatially diffuse and more regularly spaced in time) would be rather at the origin of bed-load solid transport.
Even if the main sediment volumes seem to have been provided by hyperconcentrated bed-load transport flows, debris flows have certainly occurred, and played an important role in the overflowing initiation in many places. Very huge boulders also have been brought probably by debris flows, most of them causing important damages to houses, buildings or infrastructures.
However the variability between the different catchments is very important, but it is still difficult to know certainly if this variability is mainly due to the rainfall conditions or to the catchments characteristics, like anthropic or vegetation differences.
For sure, flood events of a certain intensity are repeatable on the northern side of El Avila Mountain, as many such events have occurred in the two last centuries. The last one before the 99 event occurred not later than 1951. The consequences of this previous event were undoubtedly incomparable with the catastrophic and terrible result of the December 1999 floods, but the level of urbanization also was incomparably lower.
The occurrence of a similar event on the southern side is neither totally incredible. There are some marks of ancient deposits in the vicinity of Caracas, possibly representative of similar events. It is therefore essential to assess as precisely as possible the probability that a similar event occurs again in a given delay. The uncertainty of the answer will probably stay quite important for the moment, considering the uncertainty and lack of available data. But one thing is sure: even in case of a flood significantly lower than the 99 event, the risk is very important, because the vulnerability of this Northern coast has considerably increased.
It is therefore probably time already now to prepare the next floods, by engaging investigations and collecting the data that can improve the knowledge of the formation conditions of such phenomena.
Ayala, L. Estudio Hidrologico Integral de las cuencas de los rios del Litoral Central, INOS, Caracas, 1978.
Franceschi, L.E. Efectos de las lluvias de diciembre de 1999, Caracas, 2000.
PNUD, Efectos de las lluvias caidas en Venezuela en diciembre de 1999, Caracas, 2000.