precipitation variability and regionalization over the southern ...8 corque 3929 18º 21’ 67º...

16
Precipitation variability and regionalization over the Southern Altiplano, Bolivia Ramiro Pillco 1,2 , Cintia Bertacchi Uvo 1 and Lars Bengtsson 1 1 Department Water Resources Engineering, Lund University, Sweden 2 Institute of Hydraulic & Hydrology, San Andres Major University, Bolivia Contact author: Cintia B. Uvo Dept. Water Resources Eng./Lund University, Box 118, 221 00 Lund, Sweden Phone: +46 46 2220435, Fax: +46 46 2224435, Email: [email protected] Abstract: This is a first attempt of a regional precipitation analysis over the Bolivian Altiplano. The study region is the semi-arid Poopó and Uru-Uru lake basin, in the southern tip of the Altiplano. Fifteen rain gauges and three wind stations are located within and around this basin were the source of data used to perform a regionalization of the precipitation within the basin. The regionalization evidences four main zones of homogeneous precipitation within the basin: a) the mountainous region at the east of the basin, where precipitation is enhanced by the moist continent to the east; b) the northernmost tip of the basin, where precipitation is influenced by the larger amounts of precipitation from the region around Lake Titicaca; c) the central and southern zones, that are part of the wide flat area of basin. This is a very dry area which precipitation seems to be influenced mainly by westerly winds; and d) the north-western part of the basin which precipitation seems to be largely influenced by the presence of Lake Poopó, despite being located at about 50 km from it. Keywords: Precipitation variability; precipitation regionalization; Altiplano; Lake Poopó; Lake Uru-Uru Introduction The climate of the Altiplano, the wide region in the Central Andes, has been the interest of several studies in the latest two decades. Its precipita- tion is concentrated during the austral summer months (December to March), when about 70 % of the precipitation occurs (Garreaud et al., 2003). Characteristically, the Altiplano presents strong temporal and spatial precipitation variability, droughts and floods are common resulting on high

Upload: others

Post on 27-Aug-2021

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

Precipitation over Southern Altiplano 149

Precipitation variability and regionalization over the Southern Altiplano, Bolivia

Ramiro Pillco1,2, Cintia Bertacchi Uvo1 and Lars Bengtsson1 1 Department Water Resources Engineering, Lund University, Sweden

2 Institute of Hydraulic & Hydrology, San Andres Major University, Bolivia

Contact author: Cintia B. UvoDept. Water Resources Eng./Lund University, Box 118, 221 00 Lund, Sweden

Phone: +46 46 2220435, Fax: +46 46 2224435, Email: [email protected]

Abstract: This is a first attempt of a regional precipitation analysis over the Bolivian Altiplano. The study region is the semi-arid Poopó and Uru-Uru lake basin, in the southern tip of the Altiplano. Fifteen rain gauges and three wind stations are located within and around this basin were the source of data used to perform a regionalization of the precipitation within the basin. The regionalization evidences four main zones of homogeneous precipitation within the basin: a) the mountainous region at the east of the basin, where precipitation is enhanced by the moist continent to the east; b) the northernmost tip of the basin, where precipitation is influenced by the larger amounts of precipitation from the region around Lake Titicaca; c) the central and southern zones, that are part of the wide flat area of basin. This is a very dry area which precipitation seems to be influenced mainly by westerly winds; and d) the north-western part of the basin which precipitation seems to be largely influenced by the presence of Lake Poopó, despite being located at about 50 km from it.

Keywords: Precipitation variability; precipitation regionalization; Altiplano; Lake Poopó; Lake Uru-Uru

IntroductionThe climate of the Altiplano, the wide region in the Central Andes, has been the interest of several studies in the latest two decades. Its precipita-tion is concentrated during the austral summer months (December to March), when about 70 % of the precipitation occurs (Garreaud et al., 2003). Characteristically, the Altiplano presents strong temporal and spatial precipitation variability, droughts and floods are common resulting on high

Page 2: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

150 Ramiro Pillco, Cintia Bertacchi Uvo and Lars Bengtsson

socio-economic costs. Spatially, precipitation decreases from 1100 in the northern parts to less than 200 mm in the south and from east to west (Figure 1). This spatial variation of precipitation is related to the topogra-phy and to the availability of moisture, provided by the Titicaca Lake in the north and by the continental source of moisture in the east. (Garreaud et al. 2003; Pillco and Bengtsson, 2006) The timing and intensity of the rainy season over the Altiplano has been associated by several authors to large-scale circulation such as the Atlantic Intertropical Convergence Zone – ITCZ (Ronchail, 1995, Garreaud et al. 2003), the El Niño- Southern Oscillation – ENSO (Thompson et., 1984; Acei-tuno, 1988; Lenters and Cook, 1999; Vuille et al., 2000a; Arnaud et al., 2001; Francou et al., 2003), the Bolivian High (Vuille, 1999; Lenters and Cook, 1999; Garreaud, 1999, Garreaud et al. 2003), and the South Atlantic Convergence Zone – SACZ (Lenters and Cook, 1995, 1999). Intra-seasonal variations of precipitation are frequently associated to the availability of moisture in the region, brought by the zonal wind aloft over the mountains, from the conti-nent to the east (Garreaud and Aceituno, 2001; Garreaud et al., 2003).

Figure 1. Isohyets for aver-age annual precipitation (TDPS, 1993) at the Titi-caca-Desaguadero-Poopó-Salinas (TDPS) system with location of the Poopó/Uru-Uru Lakes basin in the south-east of the region.

Page 3: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

Precipitation over Southern Altiplano 151

The rainfall mechanisms over the Altiplano as a whole, have been thor-oughly studied before and are reviewed by Garreaud et al. (2003). Few studies, however, have been developed for sub-regions within the Altiplano, due main-ly to the lack of a dense regional data network over the region. The objective of this work is to de-fine the rainfall spatial and temporal variability within the Poopó/Uru-Uru Lake basin, located in the southeastern tip of the Altiplano (Figure 1). This basin has a total area of 24013 km2. A network of 15 rain gauges and three short-term wind stations within and around the Poopó/Uru-Uru lake basin were available for the development these analyses.

The Poopó/Uru-Uru lake basinThe Poopó/Uru-Uru lake basin is joined to the eastern mountain ranges of Cen-tral Andes, between 18o 30’ and 19o 15’ S, and 66o 50’ and 67o 15’ W with an aver-age elevation of 3700 m a.s.l. (Figure 2). The mountains occupy about 40 % of the basin area and are mostly located in the eastern part of the basin. The remaining area is flat, and a part of it is occupied by two shallow lakes: Lake Poopó and Lake Uru-Uru that occupy about 15 % of the total basin during their maximum lake-level. Lake Poopó has a surface area of about 3000 km2 at spill-over and less than half of that during the dry season. It may dry out after consecutive drought years as observed from 1994 to 1997. The main sources of water to this lake are the River Desaguadero, which carries the outflow from Lake Titicaca, regional intermittent rivers and rainfall. Lake Uru-Uru is smaller and intermittent. At its maximum depth (about 0.75m) its surface area is

Figure 2. The Poopó and Uru-Uru lake basin with topography and location of the rain gauge network.

Page 4: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

152 Ramiro Pillco, Cintia Bertacchi Uvo and Lars Bengtsson

about 300 km2. During wet periods it may happen that the lakes join into a single one (Pillco and Bengtsson, 2006). Figure 2 shows the topographic map of the basin together with the distribution of the available rain gauge stations. The basin extends along the eastern ranges of Andes, where the mountainous formation is the most enhanced. Three topographic structure zones are distinguished over the basin:

a) The eastern mountain region. This is a part of Central Andes where elevations can reach up to 5400 m a.s.l. It is the origin of most of the regional intermittent rivers that flows into the lakes. Most of precipita-tion within the basin falls in this region due to orographic effect.

b) The small western mountain region that reaches up to 4800 m.

c) The flat area of elevation around 3700 m a.s.l., extending from north to south, where the Poopó and Uru-Uru lakes are located.

Figure 1 presents the yearly precipitation field over the Titicaca, Desagua-dero, Poopó and Coipasa Salar region, which comprises about 75 % of the Bolivian Altiplano (TDPS, 1993). Over the Poopó/Uru-Uru basin it is evi-dent a negative gradient of precipitation from north to south and east to west, changing from 500 to 200 mm. Pillco and Bengtsson (2006) describe two climate zones over the basin: semi-arid in the northernmost and central parts of it and arid in the southern parts .

Data Within and around the Poopó/Uru-Uru lakes basin, 15 time series of monthly precipitation are available from the Servicio Nacional de Hidrom-eteorología de Bolivia (SENAMHI). Five gauges are located nearby the lake shore, eight along the flat area of the basin; and two at high altitudes (Table 1, Figure 2). The series were homogenized by the SENAMHI. Missing values were filled based on simple arithmetic mean, taking into account the precipita-tion observed on two or three adjacent rain gauges of similar altitudes. The original time series of Pazña and Challapata were extended to the period 1960 – 2001 by means of linear regression. For Pazña, the extension was made based on the Oruro series and for Challapata, on the Quillacas one. The period of October 1975 to September 1985 was chosen for the anal-ysis of rainfall distribution and for regionalization analysis, as, following

Page 5: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

Precipitation over Southern Altiplano 153

suggestion from Lau and Sheu, (1988), during this period all stations had less then 10 % missing data. All time series were standardized previous of use in multivariate analysis. Monthly data were subsequently grouped in totals for hydrological years (October to September), totals for the wet season (December to March) and totals for the dry season (April to November). The highest yearly pre-cipitation corresponds to Uncia with 592 mm, placed at the eastern moun-tains, and the lower to Uyuni with 201 mm, placed in the flat and arid zone (Table 1, Figure 2). Hourly wind measurements are available from Oct/01/2006 to Dec/31/ 2006 for three sites: Oruro, Salinas and Turco, the latest one located 50 km west from Corque, at similar latitude. The ENSO phenomenon was represented by the Oceanic Niño Index (ONI) and by the NINO3.4 Index, both provided by the Climate Predic-tion Center, USA (http://www.cpc.noaa.gov/data/indices).

Table 1. Location and altitude of available precipitation stations used during this work. Information about total annual precipitation during hydrological years (Oct. – Sep.), precipitation during wet (Dec. – Mar.) and dry (May – Apr.) seasons, as well as years of record are also provided.

Precipitation

N Stations Altitude Latitude Longitude Annual wet- dry- Record m a.s.l. (º) (º) season season years

mm mm mm

1 Condoriri 3750 17º 31’ 67º 14’ 442 312 121 70-952 Eucaliptus 3728 17º 36’ 67º 31’ 461 342 108 76-013 Caracollo 3770 17º 38’ 67º 13’ 434 323 109 76-864 Oruro 3702 17º 58’ 67º 04’ 478 351 122 60-025 Pazna 3710 18º 36’ 66º 56’ 484 356 120 60-956 Challapata 3720 18º 53’ 66º 47’ 397 299 93 60-957 Uncia 4420 18º 30’ 66º 05’ 592 439 128 70-848 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-979 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-9510 Andamarca 3740 18º 46’ 67º 30’ 244 198 40 60-9511 Orinoca 3780 18º 58’ 67º 15’ 367 303 64 75-9512 Quillacas 3749 19º 14’ 66º 57’ 277 215 56 60-0213 Salinas 3860 19º 38’ 67º 41’ 274 251 25 75-8514 Uyuni 3653 20º 27’ 66º 43’ 201 178 21 75-0215 Potosi 4070 19º 35’ 65º 45’ 393 299 83 75-02

Page 6: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

154 Ramiro Pillco, Cintia Bertacchi Uvo and Lars Bengtsson

Analyses and MethodsSpatial Rainfall DistributionThe spatial interpolation of the 15 rain gauges over the Poopó/Uru-Uru basin was inferred by means of Kriging. This method is considered as a flexible interpolation method, where weights are selected based on how the data fluctuation varies in space. In hydrology, this method has been widely applied for rainfall interpolation (Tabios and Salas, 1985; Creutin and Obled, 1982; Bastin et al., 1984 among others). The specific tool used in this work was the Kriging Interpolator (KI), which is an extension in Arc-View/Spatial-Analysis. KI estimates the value of the precipitation at a given point from the values at surrounding stations and from a variogram model. This method assumes the isotropic nature of the data as it uses only the distance among points, not the orientation. The radius of influenced chosen was between 50 and 100 km.

a) b) c) 

Figure 3. Spatial precipitation distribution of (a) average total annual rainfall for hydrological year, (b) average total for the wet season (Dec – Mar) and (c) average total for the dry season (Apr – Nov). Topography is included in the figure.

Page 7: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

Precipitation over Southern Altiplano 155

The rainfall distribution of the Poopó/Uru-Uru basin, inferred by krig-ing, is presented in Figure 3. Figure 3a shows the spatial distribution of the total precipitation for hydrological years. The precipitation decreases from 520 to 240 mm following the northeast-southwest direction. The decreas-ing gradient north from Lake Poopó is about 50 mm/1o, while south from the lake, it is two times higher. The mountainous region, in the eastern part of the basin presents higher totals as a result of the orographic effect. Figure 3b shows precipitation totals during the wet season. As expected, the field is similar to the one for yearly totals, considering that about 70 % of the precipitation falls during this period. However, this field emphasizes the effect of the altitude in increasing precipitation locally. Totals of precipitation during the dry season are presented in Figure 3c. During this season larger amounts of rainfall are restricted to the eastern mountainous region extending from north to south. Similar maps of spatial distribution of precipitation (not shown) were drawn for the hydrological years Oct 82- Sep 83 and Oct 73- Sep 74, which are, respectively, an El-Niño and a La-Niña year. In both cases, the spatial distribution was similar to the ones in Figure 3, however, the spatial average precipitation during the El-Niño year was 150 mm below average and dur-ing the La-Niña one, 50 mm above average.

Rainfall regionalizationCluster Analysis was used for grouping rainfall stations and determining rainfall areas of high similarity within the basin. Standardized precipitation data were used so that the method could group stations of similar precipita-tion variability. Cluster analysis methods are used to develop dendograms that are tree-like hierarchical diagrams showing the relations among all variables in a given set (Krumbein and Graybill, 1965). It is a common practice to use orthogonal solutions as input for cluster analysis so that the original data is filtered and kept only the main modes of variability. In this work, the input data used for the Cluster Analysis were the five first eigenvectors from a Principal Component Analysis – PCA (e.g. Jackson, 1991) applied to the correlation matrix of the standardized monthly precipitation. Together the eigenvectors explain 77.8 % of variance of the original data. This method was applied by other authors in similar tasks before (e.g. Bonell and Pier-sol, 1992; Uvo and Berndtsson, 1996).

Page 8: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

156 Ramiro Pillco, Cintia Bertacchi Uvo and Lars Bengtsson

Principal Component AnalysisFigure 4 presents the three first PCA modes of the standardized monthly precipitation over the Poopó/Uru-Uru lake basin. The data set used for this analysis was 15 time series of monthly precipitation from October 1975 to September 1985. The first mode (Figure 4a), explain 47.3 % of the total variance of pre-cipitation. This mode represents the influence of ENSO over the precipita-tion variability. The correlation between this mode time series and the se-ries of NINO3.4 Index (Figure 4b) is 0.3 (statistically significant at >95 %). It is possible to notice that ENSO influences the whole region in a similar way, so that El Niño years are related to lower than normal precipitation and La Niña ones to higher than normal. This mode also indicates that precipitation at the northern and southern tips of the basin is more affected by ENSO than the precipitation in the central part. This lower effect of ENSO in the central basin could be related to the presence of the Poopó Lake in the region. The possibility of a lake-effect phenomenon in that region is further explored later in this paper. Figure 4c, represents the eigenvector of the second PCA mode, which explains 9.1 of the total variance. This model evidences the semi-arid north and eastern parts of the basin from the arid west and south as described by Pillco and Bengtsson (2006). Figure 4d shows the eigenvector of the third PCA mode, that explains 8.5 % of the total variance. This mode evidences the orographic effect on the precipitation variability and differ the north-west area from the eastern parts of the basin. Modes four and five explain together 11.9 % of the total variance; however, as no physical association was found in this case, they are not shown here.

Cluster AnalysisA cluster analysis was developed using the five first modes of the precipita-tion PCA. In total, the explain 76.8 % of the original variance. Five homo-genous rainfall zones were determined and are shown in Figure 5. The first homogeneous zone includes the four stations in the northern tip of the basin (squares). This zone is the closest to the wetter northern Altiplano. Precipitation on this zone may be associated to northern and north-eastern moist winds at low level. This idea is corroborated by the observed winds in Oruro during Oct. - Dec. 2006 that shows a strong com-ponent of north-easterly and north-westerly winds during late afternoon and evening.

Page 9: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

Precipitation over Southern Altiplano 157

Figure 4. Eigenvectors of the (a) first, (c) second and (d) third modes of the precipitation PCA. The numbers at left corner are the percentage of the total variance explained by each mode. The time series related to the first PCA mode (solid line) and the NINO34 time series (dashed line) are shown in (b). The correlation coefficient between them is 0.3.

(a) 47.3 (b)

(c) 9.1 (d) 8.5

Page 10: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

158 Ramiro Pillco, Cintia Bertacchi Uvo and Lars Bengtsson

A second homogenous zone is formed by San José and Corque stations (triangles). Given the limited data available it turned out difficult to draw a definitive conclusion about the rainfall characteristics of this group. Appar-ently a combination of wind direction predominance and the presence of the lakes may drive the rainfall characteristics of this group. This idea is further explored later in this paper. A third homogeneous zone (open circles) includes stations located near the east shore of Lake Poopó, close to the mountains. Except for Quillacas, precipitation at the other stations is influences by both the presence of the mountains and the lake. A fourth homogeneous region (black circles) includes stations to the west of the lake, located at the flat area of the basin. The third and fourth zones (circles) are located at the driest area of the basin with an arid climate (Pillco and Bengtsson, 2006). The identification of this dry region is in agreement with Garreaud et al. (2003) that states that during austral summer, the transition from moist easterly to dry westerly flow at 200 hPa (that drives precipitation over the Altiplano) occurs near 20oS. It is, furthermore, interesting of notice that local peasants at the south-ern tip of the lake basin (south from Challapata) affirm that “the wet season starts from the west”. This statement corroborates the idea that, at such latitudes, the moist easterly winds aloft are less frequent. What seems to divide the third and the fourth homogeneous zones is the proximity of the mountains to the east. The fifth homogeneous zone (diamonds) includes the stations located at mountainous areas, east of the Eastern Cordillera and one located south of the basin. These side of the mountainous areas are more exposed to the moisture from the east continent than the stations located at the lee side of the mountains.

Regional Precipitation variabilityAn analysis of the inter-annual variability of precipitation during the rainy season, for four of the homogenous regions (squares, circles and triangles in Figure 5) was carried out. A wet season precipitation composite for each homogeneous zone was calculated. For the northern region, represented by squares (o) in Figure 5, a composite was created with precipitation records from Oruro and Eucaliptus; for the north-west zone, represented by trian-gles (r), with records from San José and Corque. For the west shore zone (°), Quillacas, Pazña and Challapata were used; and Andamarca and Ori-

Page 11: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

Precipitation over Southern Altiplano 159

noca for the east shore zone (l). Each composite was standardized by extract-ing the mean and dividing by the stand-ard deviation and the standardized composites are plotted in Figure 6. In this figure, strong ENSO events are marked with + for El Niño and – for La Niña. As strong ENSO events were classified the ones associated to a stand-ardized ONI < –1.5 or ONI > 1.5. Using this method, the austral summers of 1970–71, 1973–74, 1988–89, 1999–2000 were classified as strong La Niña sum-mers and the ones of 1972–73, 1982–83, 1991–92 and 1997–98 as strong El Niño. From Figure 6 it is possible to see that, in general, all homogeneous zones present precipitation anomaly of the same sign during the same wet season, however, this is not always true. An ex-ample is the wet season of 1971–1972 when the o zone presented strongly negative anomaly and the ° zone, slightly positive. Situations when dif-ferent zones present different wet sea-son quality are most often during late 1970’s up to mid 1980’s and again in mid

Figure 6. Standardized wet season (Dec – Mar) precipitation anomalies for Northern zone (squares in Fig. 5), Central-Southern zone (circles in Fig. 5), and North-Western zone (triangles in Fig. 5). El Niño years are marked with + and La Niña ones with –.

Figure 5. Homogeneous rainfall regions over the Poopó Lake/Uru-Uru basin, identified by cluster analysis. Shades represent topography so that darker colors are higher altitudes.

Page 12: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

160 Ramiro Pillco, Cintia Bertacchi Uvo and Lars Bengtsson

1990’s. The reason for these regional differences may lie on the different mechanisms causing precipitation in the different zones as elaborated in the previous section. It is interesting to remark that all zones follow the same tendency during years when strong ENSO events occur. This indicates that if a strong large-scale phenomenon may overcome regional precipitation differences, driving regional anomalies to the same direction (but not nec-essarily the same intensity).

Lake-effectThe increase of precipitation in the northern Altiplano associated with the presence of Lake Titicaca is well known. TDPS (1993) reported precipita-tion values as high as 1200 mm in the central part of Lake Titicaca, simi-larly to Hahnemberger et al. (2003) that found an increased precipitation with the proximity and mostly over the islands of this lake. Lake Poopó is located 300 km southward from Lake Titicaca, but differ-ently from this, Lake Poopó is shallow and may dry up under certain con-ditions. Pillco and Bengtsson (2006) showed that between 1976 and 1990, the average surface area of Lake Poopó was about 2300 km2. After this period, the lake-level decreased and the surface area retracted to 500 km2 in 1994. During the period 1994 to 1997, the lake was completely dry. The decrease of precipitation during this second period is noticed over most of the Altiplano and is associated with inter-annual precipitation variability (Garreaud et al., 2003). A comparison of precipitation at stations around the lake during the pe-riod 1976–1989 and 1990–1995 was made with the intention of determining a possible lake-effect. Average annual precipitation was calculated for the two periods, representing the period when the lake had a large surface area and when the lake was almost dry. It is hypothesized that during the first period the presence of the lake could influence the precipitation at those sites. Analyzed stations are divided in five groups:

a) Corque and San José, located at about 50 and 55 km west from the lake shore, respectively. They form the r homogeneous zone in Figure 5;

b) Challapata and Pazña, both located near the lake east shore (zone °) ; c) Quillacas, located near to the lake south shore (zone °);d) Andamarca and Orinoca, west of the lake (l zone);e) Oruro, near the north shore; and Eucaliptus and Condoriri, farther to

the north of the basin (o zone).

Page 13: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

Precipitation over Southern Altiplano 161

Table 2 shows the average precipitation of each of the above stations for the period 1975–1989 and for 1990–1995, as well as the ratio between them. During the period 1975–90, the average precipitation at Corque was 483 mm (Table 2). After 1990, when the lake surface area is considerably smaller, the average precipitation decreases to 314 mm, i.e., the average pre-cipitation at Corque during 1990–1995, is 0.62 of the one during 1975–1989. Similar characteristics were found at San Jose where. precipitation average decreases from 438 mm (1975–1989) to 304 mm (1990–1995), i.e., 0.69. These rates give an indication that the presence of Lake Poopó influences precipitation in these stations. In order to better support this evidence, some available wind measure-ments were analyzed. Hourly wind speed and direction were measure over the Poopó/Uru-Uru basin from October to December 2006, i.e., during the rainy season. Three automatic stations were installed at Oruro, Salinas and Turco. Salinas is located over the flat central region of the basin and thus, it better represents the average wind on the region. Turco and Oruro are located near to the mountains so that winds are strongly influenced by the mountain-valley circulation. The wind rose at Salina station (not shown) evidences that, at day time, south-easterly winds predominate, that may bring moisture from the lake to the stations locate at the north-west side of the lake. This mechanism could partially explain the large reduction of precipitation over Corque and San José during the period the lake was dry. At Challapata and Pazña average precipitation after 1990 is 0.71 and 0.77, respectively, from the average before 1989. The wind analysis at Salinas show that during late evening, south-westerly winds predominate that

Table 2. Average precipitation at se-lected locations around the Poopó/Uru-Uru basin, for wet seasons dur-ing 1975-1989 and 1990-1995. During the former period, the Poopó Lake had a large surface area. During the later period, the area diminished dras-tically to the total dry

Average Precipitation

1975–1989 1990–1995 ratio

Eucaliptus 466.0 445.2 0.96Condoriri 438.1 389.8 0.89Oruro 455.7 339.4 0.74Corque 483.4 301.0 0.62San José 437.6 304.0 0.69Challapata 400.2 332.7 0.71Pazña 475.5 367.7 0.77Orinoca 368.9 295.8 0.80Quillacas 278.5 276.8 0.99Andamarca 281.4 276.8 0.98

Page 14: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

162 Ramiro Pillco, Cintia Bertacchi Uvo and Lars Bengtsson

could partially explain the reduction in precipitation in both sites. Similar conclusions can be drawn for Oruro. It is interesting to notice that Quillacas and Andamarca, located at the south and west shore of Lake Poopó do not present any change in average precipitation between the two periods (Table 2). This lack of precipitation change in the south of the lake, windward, supports the hypothesis that the presence of the lake and the predominant southerly winds influences pre-cipitation at stations leeside from the lake. Condoriri and Eucaliptus are located too far away from Lake Poopó and, as mentioned before, they belong to the cluster in the northern tip of the basin that has its precipitation influenced by moisture from the wetter northern Altiplano. For the sake of comparison, the precipitation reduction rate was calcu-lated for eight stations around Lake Titicaca (not shown). In average, their rate was 0.89 that indicates a smaller precipitation reduction than the one at stations near Lake Poopó. Summarizing, this analysis show evidences that moisture from Lake Poopó affects precipitation at sites up to 55 km in the north-west direction, where the topography is smooth. However, no conclusive statement can be drawn on this matter at this moment due to the small amount of data avail-able during the period Lake Poopó is dry. Further studies including hydro-logical and meteorological modeling should be carried out before a final conclusion can be reached.

ConclusionsThis work examined the precipitation variability in the southern tip of the Bolivian Altiplano, within the Poopó/Uru-Uru Lake basin. It made use of a network of 15 precipitation stations within and around the area, as well as three wind stations. The influence of the orography on the precipitation is clear, however, further characteristics could be identified in the spatial distribution of the rainfall in the region and a rainfall regionalization was accomplished. There are three main homogeneous precipitation zones within the basin.

a) The northern tip that has its precipitation influenced by the moisture from the northern, wetter Altiplano.

b) The north-western zone, which precipitation shows evidences to be in-fluenced by the moisture from Lake Poopó.

Page 15: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

Precipitation over Southern Altiplano 163

c) The central-southern tip of the basin which is considerably drier than the remaining of the basin, and which precipitation seems influenced from westerly winds rather then easterlies.

The quality of the wet season within these zones, in general, is similar; however, it can be different due to their different source of moisture for precipitation. Evidences were found that precipitation over the north-west-ern region of the basin is the most influenced by the moisture from Lake Poopó that seems to reach distances as far as 55 km.

AcknowledgementsThis work was carried out as a research cooperation between the Depart-ment of Water Resources Engineering, Lund University, Sweden and the Hydraulic and Hydrologic Institute, San Andres Mayor University, Bolivia. We would like to thank the Swedish International Development Coopera-tion Agency (SIDA) for financially supporting this research; to SENAMHI for providing the precipitation data. RP thanks Prof. Ronny Berndtsson for his guidance and fruitful discussions.

ReferencesAceituno P. 1988. On the functioning of the Southern Oscillation in the south American

sector. Part I: Surface climate. Monthly Weather Review 116: 505–524.Arnaud Y, Muller F, Vuille M, Ribstein P. 2001. El Niño-Souther oscillation (ENSO)

influence on the Sajama volcano glacier from 1963 to 1998 as seen from Landsat data and aerial photography. Journal Geophysical Research 106: 17773–170784.

Bastin G, Lorent B, Duque C, Gevers M. 1984. Optimal estimation of the average areal rainfall and optimal selection of rain gauge locations. Water Resources Research 20(4): 463–470.

Bonell M, Piersol G.1992. Summer, Autumn and Winter daily precipitation areas in Wales, 1982–1983 to 1986–1987. International Journal of Climatology 12: 77–102.

Creutin JD, Obled C. 1982. Objective analysis and mapping techniques for rainfall fields: an objective comparison. Water Resources Research 18(2): 251–256.

Francou B, Vuille M, Wagnon P, Mendoza J, Sicart JE. 2003. Tropical climate change recorded by a glacier of the 20th century: Chacaltaya, Bolivia, 10oS. J. Geopys. Res. 108, doi: 10.1029/2002JD002959.

Garreaud R, Aceituno P. 2001. Interannual rainfall variability over the South American Altiplano. Journal of Climate 14: 2779–2789.

Garreaud RD. 1999. Multi-scale analysis of the summertime precipitation over the Cen-tral Andes. Monthly Weather Review 127: 901–921.

Garreaud R, Vuille M, Clement AC. 2003. The climate of the Altiplano: observed current conditions and mechanisms of past changes. Palaeogeogr. Palaeoclimatol. Palaeoecol. 194: 5–22.

Page 16: Precipitation variability and regionalization over the Southern ...8 Corque 3929 18º 21’ 67º 41’ 520 446 64 75-97 9 San Jose 3850 18º 36’ 67º 52’ 448 385 52 75-95 10 Andamarca

164 Ramiro Pillco, Cintia Bertacchi Uvo and Lars Bengtsson

Hahnenberger M, Douglas M, Galvez J. 2003. Summertime Precipitation Variability and Atmospheric Circulation over the South American Altiplano: Effects of Lake Titicaca and Salar de Uyuni. Report Salt Lake City, UT.

Jackson J E. 1991. A user’s guide to principal components. Wiley & Sons, Inc.Krumbein WC, Graybill FA. 1965. An Introduction to Statistical Models in Geology.

McGraw-Hill, New York.Lau KH, Sheu P J. 1988. Annual cycle, quasi-biennal oscillation, and Southern Oscilla-

tion in global precipitation. J. Geophy. Res. 93: 10975–10988.Lenters JD, Cook KH. 1995. Simulation and diagnosis of the regional summertime pre-

cipitation climatology of South America. J. Climate 8: 2988–3005.Lenters JD, Cook KJ. 1999. Summer time precipitation variability over South America:

Role of the Large-Scale Circulation. Monthly Weather Review 127(3):409–431.Pillco R, Bengtsson L. 2006. Long-term and extreme water level variation of the shallow

Lake Poopó, Bolivia. Hydrological Sciences-Journal-des Sciences Hydrologiques, 51(1): 98–114.

Ronchail J. 1995. L’ aridité sur l’ Altiplano Bolivien. Sécheresse 1(6) : 45–51. Tabios GO, Salas TD. 1985. A comparative analysis of techniques for spatial interpola-

tion of precipitation. Water Resour. Bull., 21(3): 365–380.TDPS (1993). Climatología del Sistema de los lagos Titicaca, Desaguadero, Poopó y Sal-

ares Coipasa y Uyuni (TDPS). Comisión de comunidades de Europeas-Repúblicas del Perú y Bolivia, convenios ALA/86/03 y ALA/87/23. LP, Bolivia.

Thompson LG, Mosley-Thompson E, Morales-Arnao B. 1984. El Niño Southern Os-cillation as recorded in the stratigraphy of the tropical Quelcaya ice cap, Peru. Sci-ence 226: 50–52.

Uvo C, Berndtsson R. 1996. Regionalization and spatial properties of Ceará state rain-fall in northeast Brazil. Journal of Geophysical Research 101(D2): 2421–4233.

Vuille M. 1999. Atmospheric Circulation over the Bolivian Altiplano during Dry and Wet Periods and Extreme Phases of the Southern Oscillation. International Journal of Climatology: 19: 1579–1600.

Vuille M, Bradley RS, Keimig F. 2000. Interannual climate variability in the Central Andes and its relation to tropical Pacific and Atlantic forcing. Journal of Geophysical Research – Atmosphere 105(D10):12447–12460.