Remote Sensing and Hydrogeology

on the Khorat – Plateau / NE - Thailand

 

 

Robert Faber

Contact: robert.faber@univie.ac.at

 Institut für Geologie, Geo-Zentrum der Universität Wien

Althanstraße 14,  A-1090 Wien

 

Keywords:

·Hydrogeology

·Groundwater

·Faults

·Remote Sensing

·Modelling

·GIS

 

Abstract:

Synoptic interpretation of different data sources is used for a hydrogeological characterization of an area near Khon Kaen / Thailand.

Landsat TM, Jers-1 and digital elevation data (DEM) allow a detailed lithological classification and fault pattern analysis of the working area. Database analysis of borelogs and groundwater chemistry data leads to a distinction of three groundwater types. Field work is used for verification and to improve the accuracy of the developed groundwater model.

 

1. Introduction:

Shortage of useable water is one of the limiting factors for the economic development of Northeast Thailand (Khorat Plateau). This lack is caused by low precipitation, high evapotranspiration and the appearance of salinar formations in shallow depth.

 

1.1 Methods:

The applied methods were chosen to meet the requirements of the investigated area:

·flat geomorphological relief – lack of outcrops

·size: about 2000 squarekilometers

·little vegetation

 

 

 

1.2 Workflow:

 

 

Fig. 1: Workflow

 

 

 

1.3 Geology:

 

The outcropping sediments of the Khorat – Plateau are build up by two Groups:

·Khorat – Group (Upper Jurassic to lower Cretaceaous fluviatile sediments)

·Phong Hong – Group (Cretaceous to lower Tertiary)

 

 


Fig. 2: Geological map of the study area

 

These two Groups are separated by a mid Cretaceous tectonic event, which lead to a relief inversion. The lower part of the Phong Hong – group – the so called Mahasarakham - Formation – is built up by three salt layers. The Tertiary is on the whole an erosive period caused by tectonic uplift. 

           


Fig. 3: Stratigraphy of the Khorat – Plateau (Lovatt Smith & Stokes,

1997; changed)

 

2. Remote Sensing

 

Due to the lack of outcrops remote sensing data was used for fault pattern analysis. Lineaments were mapped from LANDSAT TM, JERS-1 and DEM Data. Four sets of lineaments were distinguished:

 

· NW - SE

· WSW - ENE

· SSW – NNE

· N – S (only in some subareas)

 


Fig. 4: Lineament Pattern and Statistic

            A: distribution off all lineaments

            B: lineaments of the Khorat – Group

            C: lineaments of the Phong Hong – Group

            D: spatial variation in lineament distribution

 


Lineaments are only an indication for the existence of faults. If these image elements correspond to real faults a distinction has to be made if there is any influence on groundwater flow or not.

 

Fig. 5: Major Lineaments (whole LANDSAT – TM image)

 

The LANDSAT – TM data was used for lithological mapping. This step was important to get an impression of  the geometry of the geological units. Principal Component Analysis (PCA) and recombination of PCA band 3 and 4 made it possible to distinguish very accurately between several sandstone formations. PCA band 5 was used for the mapping of Tertiary and Quaternary gravels.

 


Fig 6: Lithological mapping based on LANDSAT – TM

 

During the dry season large areas in the Khon Kaen province are covered with salt crusts. The reflectance of this surface is much higher than the normal soil reflectance. Therefore LANDSAT – TM data was used for an unsupervised image classification. The classified image data displayed two characteristic patterns of possible soil salinisation:

 

A) Parallel to the stratigraphic lower border of the outcropping Mahasarakham – Formation

B) Corresponding to the drainage pattern, topographic lower than A

 

No indication for fault related salinisation was found in the distribution of saline affected soils.

 

3. GIS / Database Analysis

 

Due to the collaboration with Thai institutions (DMR, Institute of Geotechnology / Khon Kaen university) data from previous projects was available to use within this work. This data included:

 

· 578 wells (219 of them with borelogs): groundwater chemistry (analogue data, DMR)

· time series data: groundwater level, geochemistry from about 30 locations (Institute of Geotechnology / Khon Kaen university)

 

These datasets were imported to the project database and checked for their reliability. Ionic balances of older datasets are affected by large errors in their ionic balance.

 

 

First step of data analysis was the visualization of the ionic concentration data.

These plots correlated very well with the unsupervised image classification of the LANDSAT – TM data.

 


Fig. 7: left:      Spatial distribution of wells with high saline water  

            right:    unsupervised image classification based on LANDSAT TM data

                        with spots: possible salt crusts

 

Southeast of the working area the salt dome of Bora Bú is located, which is a good evidence for salt tectonics in this area. At this site the overlying Mahasarakham shale dips up to 70° and compact salt was detected with refraction seismic in shallow depth (30m under surface).

Borelogs of deeper wells within the study area – some of them drilled for solution mining - (37 wells with drilling depths greater than 80 m) were used for the creation of several cross sections to quantify possible salt diapirism.

 

 


Fig. 8: Cross Section Khon Kaen City, positions of litho logs are topographic

corrected, local dipping is not sufficient to explain the high position of the

uppermost salt layer. 

 

These cross sections and well descriptions suggests the existence of salt pillows (Fig. 8, well K58) and deformation of the overlying strata, but there is certainly no salt diapirism of the same magnitude like in Bora Bú. 

 

Regarding the content of total dissolved solids (TDS) and the position of the groundwater sampling, three types of groundwater were distinguished:

 

·        groundwater with no contact to the salinar layers, TDS shows small increase with well depth

·        groundwater which was influenced/mixed with high salinar water (moderate increase with well depth)

·        high saline groundwater (up to 80000 mg/l tds) with no dependency of depth

 

4. Field Survey

 

4.1 Outcrops

 

As mentioned above the topographic relief of the working area is very flat, therefore more or less all of the outcrops, which could be found were located at the western and northern margin of the investigated area. Measurements within these sandstone formations of the Khorat – Group resulted in dip values up to 38°, whereas the average dipping was in the range of 5 to 10°. The inclination of the strata as suggested by the deep wells in the basin seems to be much less (0 to 2°, effects of salt pillows neglected). 

 


 

 

1                   Phu Phan-Formation             8          Phra Wihan-Formation

2                   Phra Wihan-Formation                     9          Mahasarakham-Formation

3                   Phra Wihan-Formation                     10        Phu Phan-Formation

4                   Phu Tok-Formation                           11        Mahasarakham-Formation

5                   Khok Kruat-Formation                      12        Sao Khua-Formation

6                   Khok Kruat-Formation                      13        Phra Wihan-Formation

7                   Phra Wihan-Formation

 

Fig. 9: Outcrop locations

 

Fault and fissure orientations measured in the outcrops were characterized by very steep dipping, one maxima was always orientated 90° to the dipping direction of the strata at the nearest point of the ridge, the second maxima was more or less parallel to the dipping. Data density is very low especially in the central and eastern parts of the study area. The first maxima shows the same directions like the lineament data derived from LANDSAT TM and DEM data.

 

 

4.2 Geophysics

 

Since local geologists (Kriengsak, S., 1994) supposed a fault controlled groundwater flow system, fault verification was the first step during the field survey. Spatial distribution of soap holes were mapped and compared with the position of lineaments and the directions of faults, which were known from nearby outcrops. Geophysical methods (geoelectric and electromagnetic measurements) were engaged to prove fault related upwelling of saline groundwater. Either soap hole distribution not geophysics gave a clear indication for fault influence on the groundwater flow.

 


 

Fig. 10: Locations of Geophysical surveys

 

Especially the area of Ban Nong Kai Nun was surveyed in detail, since it was the key area of the previous groundwater flow model. Very high saline groundwater was only found in deeper wells (~ 60 m), which may be interpreted as an indication that the uppermost salt layer is in a very shallow depth at that position. This interpretation is supported by the extrapolation of the salt layer depth of the surrounding wells, which were drilled for solution mining.

 

 

4.3 Groundcheck of LANDSAT TM data

 

A ground check of the LANDSAT TM data interpretation was carried out to identify salt crusts and to compare their locations with the positions suggested by the unsupervised classification. To recognize different types of soil salinization a small electromagnetic device (Geonics EM-38) was used. Two types of soil salinisation were recognized:

 

 

The first type was found near dried up channels of the drainage system, the second type dominated in areas like Phra Yun or south of the Nong Lup Range, where upwelling groundwater was expected.


 

Fig. 11: Salt crusts at Phra Yun

 

5. Modeling

 

Outcrop data and lithological logs were used to generate a 3D model of the regional geology around Khon Kaen. Remote sensing data was used to extrapolate these point location data to a complete volumetric model.

 


Fig. 12: 3D Model of the geology near Khon Kaen

 

Groundwater Modeling

 

Groundwater flow is defined by several parameters:

·        Water volume (precipitation, evapotranspiration, surface runoff)

·        Lithology: Permeability of different rocks

·        Structures: Faults (discontinuities in rocks and their open volume)

·        Topography: flow is compensation of potential differences

 

 


Fig. 13: Influence of the topography on groundwater flow. If the slope is constant,

there will be only one recharge and one discharge area (right). If there is a

decline in slope – caused by different rock properties – the system will

develop other additional smaller re- and discharge areas.

(compare Fig. 14: Numerical Modeling: Re- and Discharge Areas)

 

To calculate a groundwater model data about these parameters is needed and due to the fact that the lateral extension of the modeled area is limited boundaries of the model must be defined. This “boundary value problem” has a deep impact on the results.

 

If the system is closed, in other words if the boundaries are set to be “inactive cells” or are inactive because of their topographic high position, the groundwater is forced to move up at the lowest position of the model. This might be especially a problem if the lowest area is at the same time a border of the model. In this case the groundwater has no possibility to leave. Similar results are achieved if no flow boundaries are placed in topographic low areas.

 

In the investigated area there is such a critical situation in the east, where the Chi river flows further to east. Because of the information from the well logs and the analysis of the LANDSAT TM image, it was evident that there has to be the possibility for the groundwater to leave the model area to the east.

 

 


 

Fig. 14: Numerical Modeling: Re- (blue) and Discharge (red) Areas.

  The Topographic effect dominates distribution of re- and discharge of

   groundwater (precipitation and kf – value differences are rather small

   within the modeled area)  

 

Input data for the numerical model:

 

A four layer model (Khorat – Group, Lower Mahasarakham – Formation, Upper Mahasarakham – Formation, Tertiary/Quatanary) was calculated first and compared later with a very simple one layer model.  Due to the fact that the kf – values are very similar the differences between the two models were very limited.

The distribution of re- and discharge areas was more or less identically and showed that the dominating factor is the topography (Fig. 14).

 

As this distribution pattern is in good accordance with the other data (LANDSAT TM, well data, water chemistry), there seems to be no need to add a fault driven groundwater flow system.