Evolution of the erosive process after a watershed …...Evolution of the erosive process after a watershed fire: An example: Prinos torrent, Thassos island, Greece. D, Emmanouloudisl,

Evolution of the erosive process after awatershed fire: An example: Prinos torrent,Thassos island, Greece.

D, Emmanouloudisl, I. Takosl &I. Spanos21Department of Forestry, Technological Educational Institute of Drama,Greece.2N.AG,RE.F. Forestry Research Institute, Thessaloniki, Greece.

Abstract

North Mediterranean areas, like Spain, S, France, Italy and Greece areconsidered semi-arid regions and suffer from both extensive and intensiveerosive phenomena, which are more obvious in areas that have been previouslyburned out. These phenomena cause serious floods and damages in areas close totorrent beds.

The present paper focuses on such erosive phenomena, particularly in twoburned mountainous watersheds in Thassos Island of northern Greece. Theerosive process was initially monitored after the first heavy rainfall that followedthe fire (15/8/1 989) and then repeatedly during the next decade (1989-2000). Theresults show that during this period there was a decrease in erosion valuesprobably due to two reasons:

On the one hand, the soils that are exposed immediately after the fire in therunoff force present more intensive erosive phenomena, These soils are stirredand with low grade of compression due to the effect of the rooting system of theplants. This phenomenon is more obvious in shrublands. The soils pass in arelative equilibrium three or four hydrologic periods after the fire, the erosionvalues are decreased and the danger of floods is minimized,

On the other hand, the existence of natural reforestation leads to theadditional stabilization of the ex-bare soils.

Finally, in this paper the very few cases in which there is no immediatedecrease in erosion values are examined, In these cases, urgent actions are

© 2002 WIT Press, Ashurst Lodge, Southampton, SO40 7AA, UK. All rights reserved.Web: www.witpress.com Email [email protected] from: Risk Analysis III, CA Brebbia (Editor).ISBN 1-85312-915-1

266 Risk Analysis III

1 Introduction

The appearance of fires in extended areas, especially in the summer months, is acommon phenomenon, not only in the broader area of North Meditemanean zone,but also in other regions with semi-arid climate in our planet. The fires oftencreate irreparable damages in the forest vegetation as well as in the ecosystem ofthe area as a whole, In the last two or three decades this phenomenon has sofrequently appeared that some researchers (Spanos et al. [1]) consider it anintegral part of the Meditenanean environment. However, most of the times theworst of all is the appearance of serious floods in the nearby areas during the firstwinter after the fue. These floods come from watersheds previously burnt andthe y are obvious in the plain areas which are flowed by torrents.

Numerous researchers (Kotoulas [2], Emmanouloudis [3], Pavlidis [4], etc),point out a high relationship between the summer fire and the autumn or winterflood in a given area in such a way that we could talk about a “pair” of naturaldestruction: fire - flood (Emmanouloudis [3]), or about accompaniedphenomena of the type: lightning – thunder, If one considers that this “pair”correspond to half of the known as natural destructions (fire, floods, typhoon,earthquakes) the severity of the problem can be understood. Areas of Spain, S.France, N. Italy and Greece have repeatedly suffered from destructive floods forthe last two decades in which people were lost and damages of millions Eurowere made (Rodriguez [5], Emmanouloudis [6]). For this reason, groups ofresearchers of all the above mentioned countries study the causes and thedestructive results of fires and floods, present estimation models of theecosystem and suggest restoration methods.

In the present paper the researchers have made an effort in order to explainthe immediate appearance of flood according to time in an area recently burntwith the help of a case study. The effort is made having taken the erosion as afactor into consideration which, according to the international references, has asignificant role in the creation of floods. Then, taking erosion and reforestationinto account, the researchers try to explain the lack of relevant flood phenomenain the same area during the next decade, although there was intense precipitation,

2 Field area

The field area is located in NW Thassos, which is the northwest Aegean Island,12nm south of the Macedonia coast. The research concerns a small watershedapproximately 20Km2 that is divided into two separate sub watersheds (Fig, 1),These two sub watersheds, as shown in this figure, join on the edge of Prinosvillage. In August 1989, NW Thassos and subsequently the watershed weretotally burnt, An extremely heavy rain, which fell three months afterwards(7/1 1/89), caused all the eight torrents in the area to flood but the Prinos broughtabout the most extensive and serious damages (Photo 1).

In Tables 1 and 2 the morphometrical features of Prinos watershed as well asthe distribution according to the elevation classes are shown:


Risk Analysis III 267

Figure 1: The watershed of Prinos torrent.

Photo 1: Serious darnages of village’s houses.


zfjg Risk Analysis III

Table 1. Morphometrical features of Prinos watershed.

DrainageMin Max Mean Basin’s Mean

Bed’saltitude altitude altitude mean bed’s

area(Hmin) (Hmax) (Hm)

Lengthslope Slope

Km’ m m % % Km

19,22 50 1082 4?0 32,1 12,1 7,74

Table 2. Distribution of the Prinos watershed area according to the elevationclasses.

Elevation classes. Area%

< 200m 2o1- 4o1- 601- 801- >1000m400m 600m 800m 1000m

22,54 23,86 22,54 19,3 11,17 0,52

The precipitation in the area is significant according to the G.M.Smeteorological station data, The above station is located 15 Km north of theresearch area, The 30 year survey gives a mean annual precipitation of 85 lrnm.Table 3 presents the mean precipitation per month.

Table 3. Mean precipitation per month (mm)

TotJ F M A M J J A s o N D al

12 84, 78, 48, 43, 34, 24, 17, 43, 91, 10 13 854,0 3 4 6 6 7 7 3 7 5 3>0 2,9 1,1

The geological settlement consists of three formations: Marbles andlimestones, amphivolitic shales and alluvial and colluvial deposits from thePleistocene period especially in the lower part of the basin. However, thedominant formation of the basin are marbles and limestones (85%), while theshales and the colluvial are met in the lower part of the basin and participatewith a lower percentage ( 15Yo).

Finally the vegetation cover before the fire was as presented in Table 4.

Table 4, Vegetation cover before the fire:

Vegetation cover according to typ e and species

Forests 60% Shrubs 20% Agricultural land 10%Pinus brutia, Evergreen broadleavesPin us nigra

Olea europeamainly Q. cocczfera

Soon after the fwe only the 10% of the previous vegetation remainedaccording to the Forest Service.



From the Tables 1, 2, 3 and 4 we can conclude that the Prinos torrent doesnot present anything significant as far as the torrential strength is concerned.According to Emmanouloudis classification [7], it belongs to group D as itpresents a rather medium torrential strength. The watershed relief is medium thegeological settlement is resistant and the precipitation, although quite high for aGreek coast, could not be considered as such. Therefore, the climate, relief andgeological settlement factors, which are crucial for the formation of the torrentialstrength (Kotoulas [8]), are characterized as medium, However, the vegetationcover, which is the only inhibito~ factor of the flood genesis (Kotoulas [8]), wasalmost absent in this case. But except for Prinos torrent, there were other similarcases concerning torrents with even lower torrential strength (Fourka,Asprovalta, Stilida) in which serious floods were observed, In all these cases adestructive fire was preceded in their watersheds. Thus, after a heavy rainfall in awatershed previously burnt we must expect a flood even if the torrential strengthis low.

3 Field investigation

In 1991, two years after the fire, no flood event was recorded although a heavyrain had fallen, which was slightly lower than the one of 7/11/89, The same wasalso observed in the other previously referred torrents (Fourka, Asprovalta,Stilida), It must be pointed out that in most of the above-mentioned watershedsthere is no natural or artificial regeneration. A first hypothesis was that theremust be another factor than vegetation that led to the genesis of floods. Thisfactor must be searched out in the presence of an enormous amount of bed load(of every size) that not only filled the beds forcing the water out, but was alsodeposited in fields, roads and villages. Obviously these materials come from thedegradation and erosion of watershed slopes, phenomena that pre-existed thefire, in spite of the presence of vegetation. On the contrary, during the years thatfollowed the floods the water transported less bed load as it was moving in thetorrent bed. This bed load was gradually reduced in spite of the intensity of bothprecipitation and water yield. Consequently, this probably means that the waterdid not have any transported materials to sweep away,

Considering all the above, the field research was organized as follows:

1. The quantities of the total bed load yield that were deposited in the gorge aswell as in the alluvial fan on 7/1 1/89 were taken into account.

II. The quantities of the total bed load yields that were deposited in the samespots after heavy rainfalls in 1991, 1994 and 1998 were measured.

111, The total bed load yield, because of erosion or degradation during anexcessive flood event was determined with the help of stochastic methods.

IV, The surface of the watershed slopes was thoroughly investigated with thehelp of photos, aerial photos, and visits in the area: a) before the fire b) afterthe fire and before the flood and c) after the flood, The grade of erosion inthe slopes, the bed load production and subsequently the amount oftransported materials were investigated under the same methodology,



4 Results

From I, II, III and IV stages of field research the following results wereconcluded:

I.

11,

Soon after the 7/1 1/89 flood, the prefecture of Kavala removed the depositedmaterials from the gorge and the alluvial fan (Photo 1), It is estimated that thetotal amount was 11450 m3,After the 1991, 1994 and 1998 heavy rainfalls, the total transported bed loadyields in the gorge and the alluvial fan were estimated. This was succeededwith the help of sampling cubing and weighing method of the transportedmaterial, (Kotoulas [9]). The values of precipitation (mm) and the relevantmeasured bed load yields are given in Table 5.

Table 5. Values of precipitation (mm) and the relevant measured bed load yields

111.

Dateof the event7111/89 15/12/91 23/11/94 21/12/98

Bed Bed Bed Bed

precipitation Ioa precipitation Ioa precipitation Ioa precipitation loan (mm) n (mm) n (mm) n (mm)

;’ :’ :’ :’

181 114 120 198 95 122 102 8550

Numerous stochastic methods are referred that concern the total bed loadyield which can be deposited in the alluvial fan after a flood event. In thiscase the methods Kronfellner – Krauss [10] and Hampel [11], were chosen asthey are appropriate to the torrential conditions of the Prinos torrent,According to these two methods the following calculations where made.

Hampel method

The total sediment and bed load yield of a flood event are given by the equation:

()HuG,=2,78x FxHNx Yox}-’xs-’ x(l,23~’3x3x l-—

2300

Where

G~ : the previously referred sediment and bed load yield (m3)F : the size of watershed (Km2)YO : the water intltration factor with values 0,2-0,80H~ : the max daily rainfall (mm)J : the slope of the alluvial fan (’Yo)

S : the suspended load transportation factor, values 0,66 – 0,8Hu : the altitude of the plain in the area of alluvial cone (m)

(1)



If we put the appropriate values in eqn (l), we have:

[) 50GF = 2,78 x19,2 x181/2x 0/30X 2-’ X0/8-’ X(2 -1,23~’3 X 1 -— = 893,76rn3

2300

Kronfellner – Kraus method

The total bed load yield of a flood event is given by the equation:

G,O,,=KX]XF (2)

Where:

Gtot.: the previous referred bed load yield (m3)J : the mean bed slope (angles tangent)F : the size of watershed (Krn2)K : the factor which expresses the stream torrentiality and in this case it is

K=540X&e’

If we put the appropriate values in eqn (2), we have:

[v.

GOA= 540X 4X 0,1x19,2= 881m3e’ (

With the help of photos, aerial photos, visits and the survey of the watershedTable 6 we created. It presents ~he various phases of vegetation, erosion andthe production of the transported materials in the above mentioned concreteperiods of time,

The analysis of the results that are derived from stages 1,11,111 and IV showsthat:

1,

2.

3.

The formerly observed reduction of the transported materials (during a floodevent) in the course of time was numerically ascertained (Table 5).The materials that were transported and deposited in 7/1 1/1989 were ten ormore times bigger in volume than it was expected - according to Hampel &Kronfellner – Krauss eqns (1), (2) that gave similar volume numbers oftransported materials in a single flood event. At this point, there must beemphasized that the above equations refer to the volume calculation which isproduced and deposited once in every serious and excessive flood event.Nevertheless, the transported materials of 7/1 1/89 were ten times bigger involume, which means that the event was ten times more severe than acommonly serious flood event.On the contrary, the following years (91,94,98), during which heavy rainsoccurred as well, the transported materials had far smaller volume than theones calculated by the above mentioned equations (Table 5). Additionally thedifferences in the precipitation values in the years 1989, 91, 94 and 98 do notjustify the differences in bed load volume values.




Riskibdysk 111 273

4. From the study of Table 6 we can perceive why the bed load in the flood of7/1 1/89 is 10 times or more bigger than the expected. As shown in Table 6,the bed was supplied with transported materials, not with only the ones thatcome from the usual appearance of intense erosion after the fire (Hampel andKronfellner formulated their equations for usual intense erosion conditions),but also with materials that come from massive rock fragmentation due toclimate conditions. The Q,coccz~enz rooting system, a widely-spread speciesin the area (Emmanouloudis [3]), assisted and speeded the fragmentation ofmassive rock,In stages 4, 5 and 6, as shown in the same table, we can see that the

remaining (mainly fine) materials are available to the torrent’s run off water inthe rainfalls of91, 94 and 98. The water has the dragging strength to drift andtransport materials but the available ones are much less, compared to the watertransport capacity. For this the reason the counted volume of transportedmaterials after the rainfalls in 91, 94 and 98 is lower than the expected one (asproposed by Hampell and Kronfellner ) and the absence of floods is due to thefact that there are not many transported materials.

5 Discussion and conclusions.

The Prinos torrent (as it was shown both by the facts and the research) belongs toa category of torrents that fimction in a certain way after a heavy rainfall andthey repeatedly made their presence known in Greece during the last 100 years.This does not mean that they fimctioned differently before as we do not have anyrecords for earlier periods of time.

According to the above author these torrents are called unique lavaflowtorrents (Emmanouloudis [3]) and the presence of evergreen broadleaves in theirwatershed (especially Q. coccl~era - a very common species in Greece) is theirmain characteristic, When the rooting system of the maquis and especially ofQ. coccijera compresses the massive rock - any kind of rock: limestones,granites etc – it breaks it into pieces. However, these broken pieces of rocks areinvisible because of vegetation (Photo 2),

Therefore, there is a “hidden dynamic” of materials ready to be transportedand when a fire exists and destroys the protective vegetation, it moves towardsthe sea under the effect of flood water, in large amounts and in high speeds.Moreover, during this movement, these fragmental materials drift each other likea “domino”, The materials have various diameters ranging from coarse to fine,creating a classical texture of lavaflow. In previous researches (Kotoulas [2],Emmanouloudis [3], Paulidis [4]), it has been pointed out that a torrent couldfunction as a unique lavaflow torrent regardless of its relief, its geologicalsettlement and the precipitation height, The existence of maquis or prinus landsis enough to create the previous referred “hidden dynamic”. On the other hand,when the maquis or the Quercus are not burnt, they play a completely protectiverole in the watershed, In this case there is a total absence of flood phenomena,



Photo 2: Rock fragmentation due to Q. cocczjierarooting system,

This fact causes an additional danger as everybody (forest services, civilians andauthorities) considers these torrents innocent streams and no one can expectthese streams to become dangerous lavaflow torrents after a fire, Moreover, thisis often a question posed by the authorities or even scientists: how and whenthese materials were created despite the fact that the watershed was forested.

The Prinos torrent (in Greek language Prinos = Quercus cocczfera) belongsto this category. The torrent had a watershed densely covered by pines andQuercus coccijera and till the summer of 1989 it was fimctioning as a streamrather than as a torrent, There where also beautiful recreation sites along itsbanks. Its dynamic was released after the fire and buried the village and thefields in mud in during night, The following years, in spite of the villagers’ fears,the torrent did not cause any flood event. The transported materials (mainly leaflitter, pine needles, soil and fi-agmented materials) were washed away in 7/1 1/89.Therefore, in the last decade the flood water moved in the bed and did nottransport significant amounts of materials, Moreover, because of the significantnatural regeneration, these amounts (Table 6) are more reduced as theappearance of young stand, begins to prevent the movement of the remainingsoil. For this reason one might conclude that these torrents usually functiondestructively for only a single time.

However, there are some few cases (Samos island – a couple of months ago),where a torrent functioned as a lavaflow twice after two rainfalls in a period oftwenty days. In another case (Lesbos) we recorded a torrent functioning as alavaflow twice in a period of one year. The above mentioned have obviously the



Figure 2: A special thematic map that represents the various type of erosion in awatershed, through a 3D model.

following explanation: sometimes when the intensity of the first rain after thef~e is not high or the fragmented materials that are produced are too many, thenthe ability of the total or almost total leaching in a single flood event does notexist. In these cases, 30°/0- 40°/0 or even more of the totally produced materialsremain in the slopes and the time in which the torrent will fiction as lavaflowdepends on the first heavy rain that follows. In such cases we must be veryconsidered and we must act immediately by doing works of management andcontrol after the fust rain. These works can be done with the help of the 3Dmodels of the under management watersheds, which can be constructed with thehelp of G,I.S.

A series of special thematic maps can be drawn for these models(Emmanouloudis [12]) that present the most dangerous spots of the watershedslopes (Fig. 2), Therefore, in these spots we can immediately act with lighttechnical works and also create technical regeneration cores especially in areaswith steep slopes where natural regeneration is difficult to establish (Spanos[13]).

As a

9

9

>

general conclusion we can say that:

The unique Iavaflow torrents are a mortal danger for many areas in Greece,especially where the vegetation that covers the mountainous watershedsconsists mainly of maquis and Q. cocc$era,The Prinos torrent is a typical representative of this group, that is notcharacterized by some specific morphological features or lithologicalformation, but only by the specific vegetation type.The majority of these torrents flood only one single time, After the f~stheavy rain that follows a fue. Then, because of lack of leaching materialsand with the help of natural regeneration there is a reduction of the erosion



process as the water does not find materials to erode and leach and theremaining soil is prevented from being washed away by the young stand,

9 In some solitary cases, which must be carefi.dly examined, the torrentfunctions as lavaflow in a period of some weeks or in a period of 2-3 years.In this case we must urgently intervene with technical works and technicalregeneration trying to prevent a repetition of the flood event and stop thelavaflow in only one event, The special thematic 3D models, which havebeen worked out by the authors with the help of G. I.S., significantlycontribute to this kind of intervention,

References

[1] Spanos, I., Daskalakou, E. & Thanos, C. Postfire, natural regeneration ofPinus brutia forests in Thassos island, Greece. Acts Oecologica, 21,pp. 13-20, Elsevier SAS, 2000.

[2] Kotoulas, D, Study of the action mechanism of the torrential strength,A.U. Th Publications: Thessaloniki, Greece, 1979.

[3] Emmanouloudis, D., & Pavlidis, Th, The cause and mechanism of thehomicidal flood, in Eleftheres Kavala North Greece. Proceedings of the6’hIUFRO Congress, ed. K. Sasa, Tampere, Finland, pp. 168-178, 1995.

[4] Pavlidis, Th, Erosion after f@e, Proceedings of the 3’h PanhellenicCongress of G.F.A., pp.49-57, 1986.

[5] Rodriguez, J.L.G. La Restauracion Hidrologico - Forestal en Iascuencas hidrograficas de la vertiente mediterranea, Conseyeria deAgricultural y Pesca: Granada and Madrid, Spain, 1993,

[6] Emmanouloudis, D,, Takes, 1,, Merou, T,, & Filippidis, E. Thecontribution of watershed management to the integrated flood protectionof the Athens basin.An example: the torrent of Chalandri. Ecosud 114 eds.Y, Villacampa,C,A, Brebbia & J-L Use, Witpress: UK, pp. 269-281,2001.

[7] Emmanouloudis, D. Two-jold classification of Greek Torrents, TEL ofKavala, Greece, 2001,

[8] Kotoulas, D. Mountainous water management and control, A.U.Th.Publications: Thessaloniki, Greece, 1998,

[9] Kotoulas, D.: Orini Hydronomiki, A.U.Th. Publications: Thessaloniki,Greece, 1995,

[l O]Kronfellner- Krauss, G., Quantitative estimation of torrent erosion, Proc.of the International Symposium on Erosion, Debris Flow and DisasterPrevention, Tsukuba, Japan, 1985,

[1 l] Hampel, P. Geschiebewirtschaft in Wildbachen, Wildbach andLawinenverbaung, Jg. 41, Heft 1, 1977.

[12]Emmanouloudis, D, & Filippidis, E, Protection system of mountainouswatersheds through a quantitative estimation model of their degradation.Proceedings of Int, Conf Prot, and Restoration of the Environ., eds K.Katsifarakis,G, Korfiatis,Y,Mylopoulos, Thessaloniki, pp. 751-759,1998.

[13] Spanos, I., Radoglou, K. & Raftoyannis, Y, Site quality effects on post-fire regeneration of Pinus brutia Forest on a Greek island. AppliedVegetation Science 4, Opulus press Uppsala, Sweden, 2001,


Documents

Evolution of the erosive process after a watershed …...Evolution of the erosive process after a watershed fire: An example: Prinos torrent, Thassos island, Greece. D, Emmanouloudisl,