Geostatistical Assessment of Groundwater quality in Sarchahan
1. H. Hossainipour Kooveei 2. J. Ghayomian 3. A.S.M. Gieske
1: Agricultural and Natural Resources Research Center of Hormozgan Province
P.O.Box 79145-1577 Bandar Abbas - Iran
2. Soil Conservation and Watershed management Researches Center, Tehran, Iran
3. International Institute for Geo-Information Science and Earth Observation, Enschede, The Netherlands (ITC)
Abstract
Assessment of spatial correlation in hydro chemical variables is an important tool in the analysis of groundwater chemistry. This paper investigates the spatial correlation of the Sarchahan data sets using variogram analysis and Kriging methods. Spatial assessment of groundwater chemistry is important to reveal the correlation between location and the hydro chemical variables. First, the influence of the Sarchahan diapir on the groundwater quality in the northern part of the aquifer will be investigated and secondly, the trend of the groundwater quality along the ground-water flow path in E-W direction will be studied. ILWIS 2.23 was used to determine variograms and kriged maps. Examples are given of the type of variograms found in the study area, while Kriged maps illustrate the spatial relationships found. The variogram analysis in the geostatistical assessment revealed that for the data set the spatial correlation has a relatively short range. The sill is difficult to analyze because of a spatial trend in the data, leading to an increase of the semi-variance almost immediately after the sill has been reached. From the kriged maps of the EC and Cl distributions the groundwater flow paths are clearly visible. The geostatistical analysis suggests possible infiltration of saline surface water from the Shur River in the southwestern part of the aquifer and the influence of the Sarchahan diapir on the groundwater quality in the northern part of the aquifer.
Introduction
Assessment of spatial correlation in hydro chemical variables is an important tool in the analysis of groundwater chemistry. This paper investigates the spatial correlation of the Sarchahan data sets using variogram analysis and Kriging methods. Examples are given of the type of variograms found in the study area, while Kriged maps illustrate the spatial relationships found
Study area:
The study area is in Hormozgan province, about 120 km north of Bandar-Abbas, and South of Haji -Abad about 25 km. It is located between longitudes 55° 35´ -56° 05' and latitudes 27° 02´-28° 02
Data Sets
There are two data sets for the study area:
1. The agricultural wells are monitored by the Ministry of Power (Water Organization). The data available are for 122 wells during a ten year period from 1988 until1998. The water quality analysis was done for anions, CO32-, HCO3-, SO42-, Cl- and cations Na+, Ca2+, Mg2+, and occasionally also K+, total hardness, electrical conductivity EC, pH and TDS. On the other hand for some periods only Cl- and EC were determined.
A new data set was collected by author consisting of 58 water samples (5 surface water, 1 qanat and 52 wells) that were analyzed for major anions and cations by the laboratory of the Jehad Research Center (Bandar Abbas). Five groundwater samples were analyzed for nitrate by the laboratory of Water Organization (Bandar Abbas).
´
Interpretation of the resulting variograms and maps
Variograms
The semi-variograms parameters nugget, sill and range for the data sets are summarized in Tables 1 and 2, while the graphs are shown in Figs 3 -9. In general the semi-variograms can be modeled with a spherical semi-variogram up to lag distances of 4000 m. At larger distances the trend effect starts to dominate. It should be noted that the two sets of data contain different numbers of wells. This makes it difficult to compare the nuggets, sills and ranges in the two data sets. The ranges, however, are not very large. This suggests a low spatial correlation between the sample values at medium distances. There is a substantial nugget effect, indicating large variability at very short distances, which may be due to both aquifer characteristics and errors in the chemical analysis.
Table 1: Nugget, sill and range values for the average values of the
archive data set (period 1988 - 1998).
2. Table 2 Nugget, sill and range values for the 1999
field data set.
A few comments are made on the individual semi-variograms. For reasons of space only a few characteristic semi-variogram are shown.
Electrical conductivity:
The variograms for EC were fitted with a spherical model, with a range of 2000 m. Because less sample points were available for the field data than for the archive data set, the deviation of the points from the variogram model at larger lag intervals in the former case may be due to statistical fluctuations.
Chloride:
The semi-variograms for Cl show distinct difference in range and sill values. This may be due to the fact that no saline wells in the north and south were sampled in the 1999 field work. This point will be discussed further in the next section on Kriging.
Sulphate:
The semi-variograms for the old and new data are distinctly different. The 1999 semi-variogram shows an almost linear increase from short to medium distances. Therefore it could be reasoned that a linear or power model should be fitted. Fig. 4.5b shows a power model with nugget 35 (meq/l)2, a slope of 0.009 and a power of 1.0. However, the figure also shows that the spherical model (with parameters as in Table 4.2) is almost the same for distances up to 4000 m. For ordinary Kriging purposes with a search radius of 4000 m, the spherical model was chosen.
Fig..1a EC variogram for old data Fig..1b EC variogram for 1999 data
(1988 - 1998) .
Fig. 2a Semi-variogram for Chloride Fig. 2b Semi-variogram for Chloride
(1988 - 1998 data ) (1999 data)
Fig. 3a Semi-variogram for Sulphate Fig. 3b Semi-variogram for Sulphate
(1988 - 1998 data) (1999 data)
Kriging maps
The resulting Kriged maps are shown in Figures 4.10 to 4.16. For reasons of space only a few characteristic maps are shown.
EC distribution (average 1988-1999)
Fig.10 shows the EC distribution according to the average chemical composition over the period 1988 - 1998. Fig. 11 shows the error map of the distribution. In general there is a trend of increasing EC from the eastern recharge areas to the western discharge zone around the Shur River. From Fig. 10 it is clear that the EC of wells in the north are strongly influenced by the Sarchahan diapir. The time series of well 100, for example, showed a sharp increase in EC over the 10 year period. It can therefore be concluded that the boundary of the saline-fresh interface was slowly migrating southward. In the southwestern part of the aquifer there is also an area with high EC. This may be attributed to saline intrusion from the Shur River, coupled with increasing salinity due to irrigation return flow. It appears that a local cone of depression is created here due to the extensive irrigation by groundwater. The water from the east flows towards the west passing the Tole Sorkh diapir on both sides. The figure also shows that on the western side of this diapir, solutes are released. This diapir therefore also contributes to the deterioration of the groundwater quality, although salinization is much less in this area than in the area around the Sarchahan diapir. Fig. 11 shows that the error in the kriged map ranges from 1000 to 2000 S/cm.
Chloride distribution (1988-1998)
The chloride distribution map (Fig. 4.12) shows the same pattern as the EC distribution. The figure also shows zones of low Cl in East-West trending directions that should be interpreted as zones of preferential groundwater flow. These are therefore indicated as zones of high aquifer transmissivity, possibly old river valleys with coarse deposits.
Sulphate distribution (1988-1998)
The sulphate distribution map is similar to the EC and Cl maps, with the exception that in the southern part a zone of low sulphate content extends northwards from the southern mountain range. It can therefore be concluded that the recharge from these mountains is low in sulphate.
EC distribution (1999)
Fig. 4.15 shows the EC distribution derived from the 1999 field work in the area. Apart from the main pattern of increase in EC values from East to West, which is the same as in the average EC distribution, there is a big difference in the center of the area where the EC values have become much larger. A possible explanation for this lies in an increase in groundwater irrigated agriculture in this central part of the basin.
Sulphate distribution (1999)
Fig 4.16 also shows a large increase in concentration in the central part, supporting the conclusion of increased irrigated agriculture in this area.
Fig. 4: Kriged EC distribution map (average of 1988-1998 data)
Fig.5: Error map of the EC distribution (average of 1988-1998 data).
Fig. 6: Kriged chloride distribution map (1988-1998 average data).
Fig. 7: Kriged sulphate distribution map (average of 1988-1998 data).
Fig.8: Kriged distribution map of the EC distribution (1999 data).
Fig.9: Kriged distribution map of the sulphate data (1999 fieldwork).
Conclusion
The variogram analysis in the geostatistical assessment revealed that for the archive data set the spatial correlation has a relatively short range. The sill is difficult to analyze because of a spatial trend in the data, leading to an increase of the semi-variance almost immediately after the sill has been reached. The variograms of the field data set are less clear, probably because of a much smaller number of samples.
The geostatistical assessment supported the hydro chemical assessment by identification of the spatial relationship of variables and mapping the state of the groundwater quality. From the kriged maps of the EC and Cl distributions the groundwater flow paths are clearly visible.
- Choopani, S., 2000. Groundwater investigation in the Sarchahan plain (in the framework of the Sarchahan flood spreading evaluation). ITC Thesis.
- Freeze, R.A. and Cherry, J.A., 1970. Groundwater. First edition. Practice Hall Inc.Englewood Cliffs. NewYork, pp 604
- Gieske, A., 2000. Groundwater chemistry of the Lenjant district, Esfahan province, Iran
- Hounslow, A.W., 1995. Water Quality Data Analysis and interpretation, Oklahoma State University Stillwater, Oklahoma, Lewis Publishers. pp 397
- Liebhold, A.M., Rossi, R.E. and Kemp,W.P., 1993. Geostatistics and Geographic Information system in Applied Insect Ecology Annu. Rev. Entomal., Vol 38, 303-327
- Kitanidis , P.K.,Introduction to Geostatistics , Cambridge University press,1997.
- Meer, F. Van der.,1993 . Introduction to geostatistics (GOL 38) Lecture Note, ITC MSc Environmental Systems Analysis and Monitoring Module, ESM.