Floods increasing in Buenos Aires Salado River Basin. González, Marcela Hebe and Fernández, Adriana Elsa

Transcription

1 Floods increasing in Buenos Aires Salado River Basin González, Marcela Hebe and Fernández, Adriana Elsa Sea and Atmospheric Research Centre - CONICET Department of Atmospheric and Oceanic Sciences - University of Buenos Aires gonzalez@cima.fcen.uba.ar Publicado en Environmental Change and rational water use Eds. Scarpati y Jones, Orientación Gráfica editora, junio Abstract The objective of this paper is to analyse the water excess in the soils during the last twenty years in order to improve the knowledge of the behaviour of the principal hydrological variables in the Salado River basin. In this region, rainfall and soil water storage have to be monitored in order to minimise the negative impacts of the intense floods that have been taking place more frequently since Rainfall, water balance and the runoff mean annual cycle during the period , using the Thornwaite and Matter method (1955), were analysed, detecting maximum rainfall in January and minimum in August. However, soil water storage reaches a peak in winter in a great part of Buenos Aires Province (Argentina). An example of a relevant flood event is shown in the paper. The flood in 2000, which affected the northwest of Buenos Aires and lasted from May to December, was used to compare the observed state of soils with the results derived from some hypothetical experiences. The main conclusion is that rainfall greater than normal in autumn causes important runoff and the area will probably be flooded the rest of the year because of small evaporation even though rainfall decreases in winter. Keywords Floods water rainfall soils - runoff 1

2 1. Introduction The Salado River is located in Buenos Aires province, in the east of Argentina and runs into the Atlantic Ocean (fig 1a). It is an almost flat region with a mean gradient of 0.25 m/km from west to east, located in an extended region known as the Pampas Plain. The maximum altitude of the plain is only 108m above sea level. The steeper slope is in the north-western part of the province being 3.5m/km from northwest to southeast. Rainfall, consequently, spreads easily over lower areas. On the east side, the gradient decreases considerably. There is an area with a mean gradient of 1.1m/km where rainfall does not run freely. Finally, there is a large region with gradients under 0.35m/km, where it is very difficult for rainfall to runoff. This area also receives the water flowing from the west. There, the vertical transference (evaporation-infiltration) is greater than horizontal running making floods probable. There are only two mountain systems: Tandilia and Ventania. The former, in the central part of the province, is 300 km long and its highest peak is 500m high. The latter is 80km long, its highest peak is 1200m high and it is located in the central south of Buenos Aires from northwest to southeast. There is plenty of humus in the soil of the Pampas Plain. It is composed of animal and crop detritus. Known as black soil, it has great water retention capacity and is good for growing of excellent crops. It is, therefore, considered one of the richest regions in Argentina. Most of the wheat, corn and soybean production comes from this area and it is, of course, important for the economy of a country where agricultural activities are one of the major economic resources. There are many rivers and streams in the plain; some of them end in the Paraná River and others in de la Plata River. The Salado is the most important river in the area, it starts in the south of Santa Fe province, runs along 700km and ends in the Samborombón bay, in the Atlantic Ocean. Its principal tributary rivers are: Las Flores, Vallimanca and Saladillo streams and its basin is km 2. An important characteristic of the basin is that it is more densely populated and land use more intense in the high basin, while the lower basin is used more extensively and thus less contaminated towards the Atlantic Ocean. 2

3 When the gradient becomes very flat, the river speed diminishes and more sediment settles on the river bed. The climate of the region is temperate with yearly rainfall increasing from southwest to northeast and raining more in summer (Gonzalez and Barros, 1996, 1998). Alternatively wet and dry periods are observed, as Ameghino (1954) reported more than a hundred years ago; for example, the wet period beginning in 1860 and ending in 1920 and the present one which began in 1970 and still ongoing. Although short humid periods have been appearing since 1900, the situation seems to be changing in recent times and heavy floods are detected. During the last forty years, rainfall has increased an average of 20% in the plain, yearly rainfall has risen from 600mm to more than 1000mm in the west of Buenos Aires and the agronomic limits have shifted about 200 km towards the west (Castañeda and Barros, 1994; Minetti and Vargas, 1998; Minetti et al, 2001) and associated changes in circulation were detected (Vargas et al, 1995). Some authors have studied the occurrence of wet and dry periods in the Pampas Plain. For example, Vargas (1979) evaluated the humidity excess and deficits over the period in this region to make mean field of the principal hydrological variables. Vargas and Alessandro (1990) studied the rainfall and temperature extremes in Corrientes and Minetti and Sierra (1989) did the same in Cuyo. Krepper et al (1987) studied the rainfall pattern in wet and dry years in Buenos Aires Province. The term floods refers to extended inundations, longer than a month, when soils remain saturated of water. Floods frequency has considerably increased during the last twenty years in Buenos Aires province. Penalba and Vargas (2001) observed that positive rainfall anomalies tend to concentrate since 1960 when they studied rainfall during the last century. (6) Wet conditions with some flooded areas occurred in the north-west of Buenos Aires Province in 2000, 1999 and 1997, over the Salado river basin in 1996, 1995, 1993 and 1985, in the west of the province in , in the central part of the Salado River basin in 1991, in the central-south of Buenos Aires in 1982 and in the north of the province in 1986 and When these floods occur, roads are covered with water, some towns are isolated and of course, thousands of animals die and crops are damaged. Floods during wet periods form temporary lagoons and as a result the ground water rises. The province authorities then built draining canals to reduce 3

4 accumulated water and to lower the water table. Besides, in the Pampas Plain there is more water stored in the soil than usual when an ENSO (El Niño Southern Oscillation) is in its mature phase (Forte Lay and Spescha, 2001) and rainfall is already affected (Grimm et al, 2000). Probably these facts add to the magnitude of floods in the Salado basin but the study of all of them is beyond the scope of this paper. The problem of the increase of flood frequency in the Salado basin requires a quick and definitive solution because of the economic importance and the population density of this area. As the last and most important wet period began in 1970 approximately and continues, there is a great social need for improving the knowledge of flood development, but data are still scarce. As a matter of fact, it is very difficult to determine the most convenient structural and non-structural decisions in order to minimise the negative impact of this climate variability effect. This paper is a first approach to understand the situation in the Salado basin during the last twenty years of the twentieth century. The principal objectives are to analyse mean rainfall and water balance in Buenos Aires province and to study the principal features of the flood occurred in Therefore some experiences are supposed in order to confirm that the flood risk increases when positive rainfall anomalies take place in autumn. The data and methodology will be described in section 2. The mean evolution of precipitation and water balance will be discussed and the analysis of relevant flood in 2000 is presented in section Materials and Methods Monthly rainfall and temperature from 27 selected stations (fig.1b) from the National Meteorological Service (SMN) gauge, during the period were used to carry out this study. The station data were considered if the record had only few missing data. In some cases, missing data could be completed using an homogeneous station and the rates traditional methodology (World Meteorological Organisation, 1990). 4

5 To investigate the economic and social impact of flood occurred in 2000, a detailed search was carried out, especially from local newspapers and magazines. Informative news generated by institutions like National Agropecuary Technology Institute (INTA) or National Water Institute (INA) WebPages and Climatic Bulletins from SMN, were used to enlarge the knowledge of the situation. Mean monthly temperature and monthly-accumulated rainfall during the period were calculated in order to evaluate the mean water balance. It consisted in the monthly evolution of the water storage in the soil and the deficit or excess of water, using the methodology proposed by Thornwaite and Matter (1955) (TM, hereinafter). The mean field capacity of soils in the Pampas Plain used to perform the study was taken as 200mm (Lamas, 2005) for the entire region. This is an adequate value for the kind of soil of that plain. Following TM (1955) methodology, potential evapoperspiration was first calculated for each month using the mean temperature and sunlight period. If the difference between rainfall and potential evapoperspiration was positive (negative), hence, there was water excess (deficit) in soils. The water storage depended on the field capacity and its value let calculate the actual evapoperspiration, which was equal to potential evapoperspiration only if rainfall was greater than it. The water balance for each one of the years, chronologically between 1980 and 2000 was calculated following TM (1955) methodology, too, using the monthly temperature and monthlyaccumulated rainfall during the period beginning in January 1980 and ending in December The water balance started in a month when rainfall mainly exceeded potential evapoperspiration in order to be sure that the soil was saturated. Using this methodology sequentially since 1980, the mean error tends to decrease (TM, 1955). This was done to determine the hydrological situation of soils in 2000, which would be especially studied in next section. As an indicator of the contribution to the runoff a parameter was calculated for each month: it was supposed that if there was water excess in a month period, only half of that excess would be able to run off that month and the other half would runoff during the following months. This parameter was called runoff in the text. This would be a good assumption in areas with slight slope as it was detailed in TM (1955). 5

6 3. Results The mean rainfall cycle during the period (fig 2) shows that precipitation tends to be greater in summer than in winter. January is the rainiest month and rainfall decreases from the north (more than 130mm) to the Southwest (less than 80mm). The maximum turns from the north to the Northeast and in May it is a clear Southwest-Northeast gradient but with values lower than in January all over the area. August is the least rainy month and rainfall increases from the west with less than 30mm to the east with only 50mm next to the coast. Then, rainfall increases and, in October, a north-south gradient is the feature again, announcing the beginning of summer conditions. Monthly mean temperature and rainfall series in each of the stations were used to calculate the water balance (fig 3). Two effects influence water soil excess (positive values) and deficits (negative values). First, heavy rainfall provides more water to the soil than in normal cases and sometimes the soils saturate. Second, in summer, high temperature lets water evaporate meanwhile in winter, although rainfall is less than in summer, low temperature inhibits evaporation. The result of the combination of these two effects is that all the area has water deficit in January (fig. 3a). As summer ends, temperature decreases but rain is already high, and so, soil begins to retain water until in April, a maximum excess can be found especially in central part of Buenos Aires province, with excess of more than 60 mm (fig.3d). In winter, rainfall and water excess decrease, reaching a minimum in August (fig 3h). As spring develops, rainfall season increases again, meanwhile temperature remains mostly low, generating more water excess in soil with maximum values in October (fig.3j), although not so high as in April. Then, temperature increases favouring evaporation and the south western part of Buenos Aires has more than 20mm-water deficit in December. The water storage in the soil increases from South to North of the studied area. For example a great part of Buenos Aires province is saturated in June (fig.4). Therefore, the water storage reveals the consequences of the presence of saturated soils in winter and when rainy season begins in Spring there are high probabilities of floods. 6

7 In spite of the main water soil condition, flood frequency has considerably increased during the last twenty years in Buenos Aires province. The interannual variability of rainfall and water storage is analysed too performing sequential water balances since 1980 to More frequent flooding events are detected and corroborated with newspaper information. Therefore, great floods occurred in 1980, since October 1982 to April 1983, since October to December 1985, since June 1991 to June 1992, since January to December 1993, with great severity in Autumn, since August to October 1997, during 1999, since May to December 2000 and from January to September Sometimes they covered the complete basin and other only an area. Dry events seem to be less frequent. Figure 5 shows a sequence of the main flood and dry events registered in the area of study. An analysis of one of the most recent and important flood in the area of Salado river basin is presented. The flood extended all over the Northwest of Buenos Aires province where an area of Ha was flooded in June This is one of the four floods, which occurred in this area between 1990 and 2000, and this fact determined a very difficult situation to revert, with a great amount of economic losses and social associated problems. The flooded area covered Carlos Tejedor, Lincoln, General Pinto, Florentino Ameghino, Trenque Lauquen, Pehuajo, General Villegas and Rivadavia districts (see fig 1). Pehuajo and Nueve de Julio districts were the most affected with 80% surface flooded and Trenque Lauquen with 35% surface. In November 2000, 3600 km of roads were non usable and nine towns, located inside the affected area, were isolated. More than 200 people have to be evacuated and some of important national highways, as number 226, 86, 70 or 33 were non-usable. At the end of the year 2000, Ha were flooded, almost 10% of the Pampas Plain total extension ( Ha). The agricultural losses were 400 thousand dollars and 50 thousand in livestock (source: local newspapers). Ventimiglia et al (2003) quantified losses in Nueve de Julio district. Corn losses reached Ha with a performance of 7.000Kg/Ha and soy losses Ha with 4500Kg/Ha of performance. They evaluated that the total losses were equivalent to a decrease of 2.2 % of Domestic Gross Income. 7

8 Rainfall in year 2000 was depicted by large positive anomalies, especially in February (fig 6a), May (fig 6b) and October (fig 6c). In February rainfall was greater than normal all over Buenos Aires province, in May there was an intense gradient with important precipitation values in central and northern part, and in October the largest anomalies were observed in the Northwest, centred in Pehuajo district. The sequential water balances were done for each one of the years during the period in order to evaluate the temporal evolution of water in soils. In general, results differ in each flood event because the affected area covers different regions. For example, the annual water excess in 2000 (fig 7a) shows an important excess in the north of Buenos Aires Province, meanwhile in 1997 (fig 7b) the entire province is flooded, specially the central region, when they are compared with the mean annual water balance during The evolution of monthly water balance anomalies in 2000 reflected the rainfall feature: excess greater than normal especially in February (fig 8b), May (fig 8e) and October (fig 8j) over the flooded area. The water storage in soil reached 200mm, determining a flooded area along the Salado river in May (fig 9a), only in the east along the Atlantic coast until September (not shown) and in October, again all the Salado area was flooded (fig 9b). The runoff was calculated in all the stations and then averaged in the area with water excess greater than normal following the methodology detailed in section 2. The mayor differences were in May and October, when the most important rainfall occurred. The evolution of runoff showed values greater than normal since May to December, indicating the period when the area could be flooded. The runoff is representative of the amount of water that covered the saturated soil and so it could increase the Salado river volume or remain determining a flooded area. Therefore, it is an important indicator of flood intensity. The comparison between runoff evaluated in each station in 2000 and averaged in all stations in the flooded area, and the mean monthly values for the period is shown in fig 10. It is important to point out that runoff is maximum at the end of autumn and in spring and it is minimum in summer. In 2000, runoff exceeded its mean value especially in May and October when rainfall was important in the area. 8

9 In order to evaluate the degree of importance of rainfall in May and in October, or both, as a determinant factor to establish a flood, the forward experiences were developed. The first simulation (E1) consisted in considering that only in May would have rained normally, the second (E2) that in May and October would have rained normally and the last (E3) that only in October would have rained normally. To do that, the sequential water balance beginning in 1980 and ending in 2000, were calculated replacing May and/or October 2000 rainfall by their mean values calculated over the whole period ( ), as it was required in each of the three experiments detailed. The most important conclusion was that the annual water balance was particularly determined by rainfall in May because in E1(fig 11b) and E2 (fig 11c) the positive water balance was less than in the north of Buenos Aires than in the case of E3 (fig 11d), which showed a very similar pattern that one actually observed for year 2000 (fig 11a). The runoff was calculated as the average in all the stations located inside the area which had positive annual water balance. However, both, May and October rainfall, contributed to the important annual excess and runoff in 2000, as it can be noted in Table 1 which shows some comparison between the annual behaviour of the experiments. As the fact, E2 had values more similar to the mean ones and E3 resulted the most similar to 2000 case. Fig 12 shows the runoff evolution observed in 2000 and those resulted of the three simulations proposed. Observed runoff has a maximum of 103 mm in May and other of 72 mm in October. If it rained normally in May the runoff would be greater than the mean value just in June (49 mm), as it shows E1. But if October rainfall was normal (E3), May precipitation would be enough to produce a runoff greater than the mean value during the rest of the year, with a relative maximum (31 mm) in October. Finally, if both May and October rainfall were normal, May runoff would reach only 42 mm and it would decrease during the rest of the year (E2). Therefore, the positive rainfall anomalies in February and May, determined the necessary conditions to produce runoff greater than normal. To better understand this effect, fig 13 shows the mean evolution of potential evapoperspiration (ETP) and rainfall over the period in Junin. Rainfall has two maximums in autumn and spring, meanwhile the ETP is directly related to temperature. Therefore, rainfall is greater than ETP since March to July and water stores in the 9

10 soil. This process tends to revert only in the middle of winter (August) because of the low temperatures and in spring, rainfall increases again. If rainfall is unusually high in autumn or beginning of winter, soils are saturated and the low winter temperatures do not let water evaporate. So, normal spring rainfall, can determine a severe flood. 4. Discussion The behaviour of rainfall and water balance in soils was studied using TM (1955) methodology during the period in the Pampas Plain. The analysis of mean rainfall reveals an annual cycle with maximum rainfall in January and minimum in August and the water storage in soils reaches its maximum value in winter in a great part of Buenos Aires. This period is characterised by an increase of flood occurrence. The flood in year 2000 affected the northwest of the province and it was characterised by heavy rainfall in February, May and October, which determined an extended flooded area from May to December. May rainfall seems to determine the subsequent flood, as it can be observed in the annual runoff and water balance in Table 1 and figure 12 indicating that when rainfall is greater than normal in autumn or at the beginning of winter, conditions are favourable for significant runoff and probable flooding the rest of the year. The reason is that it rains less in winter than in summer but there is little evaporation too. Consequently, normal convective rainfall in spring can cause floods. Aknowledgement To National Meteorological Service for providing the rainfall data to do the work. I wish to thank Eng. Alvaro Lamas (Agronomy-UBA) for providing the field capacity data in the region and Dr. Vicente Barros for his comments and suggestions. This research was supported by Inter-American Institute for Global Change Research (IAI) (CRN-055) and UBACYT-264 grant. 10

11 References Ameghino, F (1954) Drought and floods in Buenos Aires Province, Honorable Representative Camera of Buenos Aires city Publication, Argentina. Castañeda, M and Barros V (1994) Rainfall trends in South America, east of Los Andes, Meteorologica. 19: Forte Lay, J and Spescha, L (2001) El Niño, Impact in underground water of Pampas Plain, Preprint IX Latinoamerican Meteorology Congress and VIII Argentine Meteorology Congress, Buenos Aires, Argentina. González, M and Barros, V (1996). Statistical features of rainfall annual cycle and its anomalies in Subtropical Argentina, Meteorologica, 21: González, M and Barros, V (1998) The relation between tropical convection in South America and the end of the dry period in subtropical Argentina, International Journal of Climatology, 18: Grimm, A, Barros, V and Doyle, M (2000) Climate Variability in Southern South America associated with El Niño and La Niña events, J. Clim, 1: Krepper C, Scian, B and Pierini, J (1987) Variabilidad de la precipitación en la región sudoccidental pampeana, Actas II Congreso Interamericano de Meteorología, Buenos Aires, Argentina. Lamas A (2005), personal communication, Department of Agronomy, University of Buenos Aires. 11

12 Minetti, J and Sierra, E (1989) The influence of general circulation pattern on humid and dry years in the Cuyo Andean region of Argentina, Int. Jour.Climatol., 9, Minetti, J and Vargas, W (1998) Trends and jumps in the annual precipitation in South America south of 15S, Atmosfera, 11: , Mexico. Minetti, J, Vargas, W, Poblete, A, Casagrande, G and Acuña, L (2001) Climatic jump in Evidence of difficult process of climatic change in Argentine, Preprint IX Latinoamerican Meteorology Congress and VIII Argentine Meteorology Congress, Buenos Aires, Argentina. Penalba, O and Vargas, W (2001) Propiedades de excesos y déficits de precipitación en zonas agropecuarias, Meteorologica, 26, 1-2, Thornwaite C and Matter J (1955) The moisture Balance, Publications in Climatology, Laboratory of Climatology, Centerton, New York, 104 pp. Vargas, W, Minetti, J and Poblete A (1995) Statistical study of climatic jump in the regional circulation over South America, J. Met. Soc. of Japan, 73: , Japan. Vargas, W (1979) Atlas de excesos y déficits de humedad en la región húmeda y semiárida argentina, Ministerio de Economía, Instituto Nacional de Ciencias y Técnicas Hídricas, 140p. Vargas, W and Alessandro A (1990) Los extremos climáticos de precipitación y temperatura en Corrientes, Meteorologica, 17,

13 Ventimiglia, L, Carta, H and Rillo, S (2003) Floods: a problem of everybody, National Institute of Agronomical Technology (INTA) Homepage. World Meteorological Organisation (1990), Climatological practises Guide, Technical Note Nº 100., Chapter 5, Genever, Swisserland. 13

14 (a) Santa Fe Province -30 PERGAMINO PEHUAJO T.LAUQUEN BOLIVAR JUNIN 9deJULIO LASFLORES AZUL Samb or omb om Bay -45 C.SUAREZ PIGUE Buenos Aires Province west longitude west longitude (b) Figure 1: Location of Buenos Aires province in Argentina and Argentina in South-America. Shaded area is the Salado River basin (a). Stations used in the study (b). 14

15 (a) (b) (c) South lati tude We st l on gitu de (d) (e) (f) (g) (h) (i) South l atitude (j) (f) (l) Figure 2: Monthly mean rainfall (in mm) in January (a), February (b), March (c), April (d), May (e), June (f), July (g), August (h), September (i), October (j), November (k) and December (l). 15

16 (a) (b) (c) South lati tude (d) (e) (f) (g) Wes t longitude (h) (i) (j) (k) (l) Figure 3: Mean water balance (in mm) in January (a), February (b), March (c), April (d), May (e), June (f), July (g), August (h), September (i), October (j), November (k) and December (l). Positive values are excess of water (full line) and negative values are water deficits (dash lines). 16

17 º Figure 4: Mean water storage in soil (in mm) June. Figure 5: Main floods (full arrows) and dry events (dash arrows) occurred since

18 (a) (b) (c) Figure 6: Rainfall anomalies (in mm) in February (a), May (b) and October (c), (a) (b) Figure 7: Annual water balance anomaly for year 2000 (in mm) (a) and for 1997 (b). Dash lines are negative values. 18

19 (a) (b) (c) West l ongi tude (d) (e) (f) (g) (h) (i) (j) (k) (l) Figure 8: Water balance anomaly (in mm) in January (a), February (b), March (c), April (d), May (e), June (f), July (g), August (h), September (i), October (j), November (k) and December (l) of year

20 (a) -3 4 (b) Figure 9: Water storage (in mm) in May (a) and October (b), runoff (mm) j f mr ap my jn jl au s o n d mean runoff runoff 2000 Figure 10: Runoff evolution during 2000 (dash line) and its mean values in (full line). 20

21 Figure 11: Annual water balance (in mm) for 2000 (a) and for E1 (b), E2 (c) and E3 (d) simulations. 21

22 runoff (mm) j f mr ap my jn jl au s o n d E1 E2 E3 actual 2000 Figure 12: Runoff for the simulations and the observed for Junin (mm) month ETP Rainfall Figure 13: Mean rainfall and potential evapoperspiration for the period in Junin. 22

23 Annual accumulated water excess (mm) Annual runoff (mm) Actual E E E Mean Table 1: Annual accumulated water excess and runoff for each one of the experiments. 23

24 24