DSpace at VNU: Application of multimedia methodology for investigation of karst water in highland regions of Ha Giang Province, Vietnam

DSpace at VNU: Application of multimedia methodology for investigation of karst water in highland regions of Ha Giang Pr...

S P E C I A L I S S U E

Application of multimedia methodology for investigation of karst

water in highland regions of Ha Giang Province, Vietnam

Ngoc Thach Nguyen•Ngoc Hai Pham•

Xuan Canh Pham• Thi Thuy Hang Nguyen•

Van Lam Nguyen•Thi Thanh Thuy Duong

Received: 25 May 2012 / Accepted: 21 June 2013

 Springer-Verlag Berlin Heidelberg 2013

Abstract Ha Giang is one of the largest, northern border

provinces of Vietnam, consisting of four districts: Yen

Minh, Quan Ba, Dong Van and Meo Vac This province

features varied karst landscape of Carboniferous–Permian

limestone The region has been recognized by UNESCO as

one of the 77 geological parks in the world and the second

in Southeast Asia on 3 October 2012 In the dry season,

little or no rain is recorded; therefore, surface water is very

scarce For this reason, proper delineation and exploitation

of the groundwater resource is critical for sustainable water

supply This has been identified as an important challenge

under the scientific project KC-08-10 in the national

pro-gram KC-08 Remote sensing and GIS were used to

deci-pher the signature of karst water in the highland of Ha

Giang Information layers generated were subjected to

multi-criteria evaluation using analytic hierarchy process

for decision making to identify ideal locations for

groundwater prospecting The study resulted in delineation

of ten zones for all regions and 18 ideal drilling sites in

Tam Son Town of Quan Ba District Drilling and resistivity

soundings were performed to assess the success of the

interpretation Deep resistivity survey confirmed low

resistivity (200–300 Xm) near the identified potential sites

in Tam Son Town of Quan Ba District Further, successful

drilling at site LKTS1 with a discharge of 7–9 l/s is

observed, proving the potential of this methodology for

rapid exploration of groundwater in wascare karst ter-rains of Vietnam

Keywords Karst water Remote sensing  GIS  AHP  Hydrographic geomorphology Water resource

management Groundwater exploration

Introduction

Geographical information system (GIS) has been rapidly developed and effectively used in various fields of earth science and natural resources management Remote sens-ing and GIS have been used in explorsens-ing groundwater in mountainous areas, particularly in the limestone areas (Granados-Olivas et al 2005) Since the 1950s in many countries, aerial photograph interpretation methods have been applied to groundwater investigation with indirect signs through geomorphologic and tectonic concept Remote sensing has popular applications because of its unique advantages such as synoptic coverage, remote area access, and spectral, resolution and multi-temporal properties (Sabins1991) These properties make satellite remote-sensing data useful for application-specific use, especially for hydrographic geomorphological map-ping (Granados-Olivas et al.2005; Walvoord et al.2002) Remote-sensing data have been utilized for decades in hydrogeological investigation work and thematic research Common applications of remote-sensing analysis are stra-tigraphy mapping, geological structure analysis, fault detection and identification, and geological lineament extraction (Sabins1991) In many cases, a high density of the extracted geological lineaments is interpreted as a zone

of highly fractured rock (Babcock 1974) Hence, these zones receive, in general, first priority for prospection of

T T H Nguyen

Faculty of Geography, VNU University of Science,

Vietnam National University, Hanoi, Vietnam

e-mail: nguyenngocthachhus@gmail.com

Faculty of Geology, University of Mining and Geology,

Hanoi, Vietnam

DOI 10.1007/s12665-013-2617-3

groundwater resources (Balakrishnan 1986;

Granados-Olivas et al.2005; Nag2005; Srinivasa Rao et al.2000)

Systematic methodology in application of remote sensing

for groundwater research has been introduced by FAO in the

book, ‘‘Groundwater research by remote sensing, a

meth-odological approach’’, by Travaglia and Dainelli (2003) The

approach used in this study was a development of the

tradi-tional standard sequence of drainage, landforms, cover and

lineaments analyses, to which several improvements and

additions were made The lineament system is a major

indicator of connection between surface and deep

ground-water in the karst regions Faults and lineaments system can

be extracted automatically or semi-automatically using

digital or visual image processing on satellite data These two

structural features provide independent information,

allow-ing assessment and analysis of groundwater potentials prior

to actual drilling (Balakrishnan1986)

In Vietnam, approaches to hydrographic geomorphology

and geology with remote-sensing applications had been

introduced in the University of Natural Sciences and

Uni-versity of Mining and Geology since 1972 Currently,

satellite digital image scanning and digital image

pro-cessing techniques are used for assessing underground

water potential through automated data classification and

separation techniques (Nguyen 1986, 1993) With the

availability of high-resolution satellite data, geophysical

data and improved positional fidelity due to global

posi-tioning system (GPS), the accuracy of mapping

ground-water-rich zones and well locations for further groundwater

explorations and research has been improved

In karst topographic research, the remote-sensing

method is applied to determine the landform and tectonic

features, which are related to the potential of groundwater

concentration Some typical studies can be mentioned as:

Hydrogeological characteristics of a karst mountainous

catchment in the Northwest of Vietnam (Tam et al.2001);

Study on the relationship between lineaments and borehole

specific capacity in a fractured and karstified limestone

area in Vietnam (Tam et al 2004); Study of cavernous

underground conduits in Nam La (Northwest Vietnam) by

an integrative approach (Tam et al.2005); Remote sensing

and GIS-based analysis of cave development in the Suoi

Muoi Catchment (Son La-NW Vietnam) (Hung et al

2002); A multi-analysis remote-sensing approach for

mapping groundwater resources in the karstic Meo Vac

Valley, Vietnam (Tam and Batelaan2011)

The major focus of these studies is to investigate the

relationship between hydrographic geomorphological

fac-tors (lineaments and fault systems) with hydrodynamic

characteristics of a karstic aquifer The final high-yield

locations are based on the maxima of suitable conditions

The final location of the borehole is still based on

sub-the previous one by implementing analytical hierarchy process (AHP) for final suitability, thereby better reporting the subjectivity so that it can be replicable at various scales and with ease Using AHP for decision making has various advantages, including pairwise assessment of multiple factors, weights which are compensatory and choice regarding risk-averse and risk-taking decisions (Saaty

1977)

Due to successful quantification of factors and their relative importance, the method established in this study can be applied at various scales of groundwater prospecting

in areas with similar topographic and geological setting (Fig.1)

Study area

Ha Giang, the northernmost province of Vietnam, has a relatively complicated terrain It consists of high mountains and deep valleys, rising from the south to the north, divided into three main regions In the north and northeast of the province, high mountains of limestone with high slopes separated by valleys, rivers and springs mark the area The west of the province includes highland from the Chay River massif These two regions have similar climatic conditions with moderate climate of two seasons: dry and rainy The lower areas in the province include low hills, Lo River Valley and Ha Giang Town In general, the terrain of the province could be characterized by two natural regions, including the upland and the low-lying region (Fig.2a) The upland includes the rocky mountains in the north and northeast, and the highland of mountains in the west Most of this highland forms an arch or semi-arch, with many continuous mountain ranges (Fig.2b, c) The rocky mountains include Quan Ba, Yen Minh, Dong Van and Meo Vac, which are part of the Dong Van Plateau with

80 % of the area covered by limestone, with the notable Lung Cu mountain peak of 1,621 m height The western highland includes Hoang Su Phi, Xin Man partially lying

on Bac Ha Plateau with a 2,43 l m elevation and Tay Con Linh mountain peak High mountain ranges alternate with deep valleys through narrow strips of land With 40 loca-tions which have special values in terms of natural resources in karst landforms, the highland karst region of the Ha Giang Province has been recognized by UNESCO

as one of the 77 geological parks in the world and the second in Southeast Asia on 3 October 2012 The park covers four districts of Meo Vac, Dong Van, Yen Minh and Quan Ba, totalling over 2,300 km2, with nearly 250,000 residents Up to 80 % of the plateau is covered by lime-stone The park is home to nearly 20 ethnic groups, with diverse cultures and traditions, which make the plateau an

The lower sub-region or lower land includes the

remaining area of the province in the southeast, expanding

from Bac Me District, Ha Giang Town, Vi Xuyen to Bac

Quang close to Tuyen Quang Province The terrain here

mostly consists of low hills, with evergreen forests

alter-nating with wet rice fields and alluvial deposits along two

river banks Many kinds of crops can be seen in this region

Ha Giang is a mountainous province characterized by

distinguished tropical monsoon climate from surrounding

lower lands and midlands with two main seasons: rainy and

dry In 1999, the average temperature in the province was

28.1C (Ha Giang Station), 28.3 C (Bac Quang Station)

and 27.35C (Bac Me Station) The highest temperature is

recorded in June or July, while the lowest is recorded in

January at 1.56C (Hoang Su Phi Station) The differences

between day and night temperatures in valleys are more

notable than in the delta region The rain regime in this

province is quite diversified The yearly rainfall is

2,860 mm The number of rainy days ranges between 180

and 200 days per year In the dry season, the highland

sub-region of Ha Giang is seriously deprived of water, espe-cially in the northeastern part of the province where karst terrain is dominant Supplying water is most difficult in the karst highland areas with elevation from 700 m and above Rivers in Ha Giang are unequal in depth and have high slopes with many waterfalls and rapids Compared to the lowland, the rivers and streams in the upland have low drainage density In other words, the mountain ranges separating the river system in the upland area are very high, e.g., the bed of Nho Que River is 400 m deep from the flat area of human habitation

With support from the government, many small lakes have been constructed for various purposes, including storage of rainwater and supplying the same in the dry season However, these lakes are not adequate to meet the increasing water demand, especially in highly populated areas The most critical of these areas are the ones between elevations of 100 and 700 m, where most ethnic minorities live Tam Son Town located in the southeast has similar characteristics (Fig.2)

In consideration of the seriousness of this situation, a

sub-project under the National Research Program KC-08

‘‘Pursu-ing on prevent‘‘Pursu-ing and mitigat‘‘Pursu-ing natural disaster’s damages’’

was established This project (No KC 08/06-10) was

imple-mented between the year 2008 and 2010 The main objective of

this sub-grant was to rapidly assess potential zones of

groundwater for augmenting the water supply in the upland of

Ha Giang Province Further objectives were to establish a

quantitative method which can be rapidly applied to other areas

with similar geological and topographical settings

Given the objectives and time frame, remote sensing for

rapid identification of hydrographic geomorphological

structures and GIS for quantitative decision making were

found to be the most suitable techniques The results were

further verified by geophysical testing and drilling This

study area was spread in four upland districts of Ha Giang

Province, namely, Meo Vac, Dong Van, Yen Minh and

Quan Ba Field-based validation was conducted at Tam

Son Town of Quan Ba District

Materials and methods

Study process

To apply AHP-based decision making to establish a

quantitative method to decipher groundwater prospecting

location using several indirect signatures, the following steps were undertaken The scientific approach for studying the karst areas of the Ha Giang is also summarized as a flowchart (Fig.1)

Secondary data collection

Hydrogeological data which were the product of the mapping project conducted in the region since 1968 to present were collected Major features of the hydrogeology

in the highland area of Ha Giang Province can be described

as follows:

– Geological formations resulted in aquifers of limestone, dolomite–limestone, interbedded siltstone and shale from Ordovician (O), Ordovician –Silurian (O–S), Silurian (S2), Devon (D1) and Carboniferous–Permian (C3–P1) age The thickness of the aquifer averages about 40–50 m; it is covered by late continental formations

– Hydrogeological structures are formed and controlled

by a major fault in the NW–SE direction (Fig.3) Due

to tectonic movement, high densities of lineaments in other directions are formed

– After a long history of tectonic movement and weathering process, landforms in the area show various shapes such as bell, tower, funnel, cave and

underground cave These features indicate that

groundwater is channeled to deeper locations

con-trolled by the predominant structures The secondary

information about formations and lineaments were digitized to be amenable to further GIS and remote-sensing analysis

Remote-sensing data collection

Landsat TM image (Landsat 5) with a spatial resolution of

30 m for bands 1–5 and 7 was acquired A cloud-free

image of 24 November 2000 (Fig.4) was selected to be

used for digital image processing and visual interpretation

It must be emphasized that for replicating this model of

groundwater prospecting for a larger scale, higher

resolu-tion image datasets including SPOT 5 or QUICKBIRD are

recommended

Factors and constraints

To select the most suitable location for groundwater

pro-spection, AHP-based decision making was applied Factors

and constraints to the analysis were identified based on

expert knowledge and past research Hydrographic

geo-morphological structures, (TWI), lineament density,

line-ament node density, distance to lineline-ament (NE–SW

direction) and topographic wetness index were selected as

key factors

Constraint to the analysis was identified as the region

where the need for augmenting groundwater resource is

critical Since this geographic region is marked with a high

concentration of ethnic minority population They inhabit

the slopes between elevations of 100–700 m The

con-straint mask was created using DEM of the area

Structure

Using the Landsat TM data, two methods were applied for structure mapping Automated lineament extraction using Laplacian and high-pass filter were faster compared to visual interpretation Secondary information such as digital elevation model (DEM), shaded relief, slope, aspect and curvature maps was found to be useful to enhance auto-mated extraction of linear structures (Elmahdyl and Mo-hamed2012)

However, between these two approaches, a map created

by visual interpretation in combination with secondary geological information was found to be more accurate The automatic extraction of linear features alone was found to

be misleading At times, automatic extraction process captured linear features such as roads, while it was not possible to identify overburdened structures

Through the combination of primary and secondary dataset and image enhancement, the hydrogeostructural map was generated with 17 major structures, elongated mainly from the northwest to southeast direction Pre-liminary investigation of the structure map help in locating the general area of good groundwater potential, marked by blue and dark green (Fig 3) The hydrogeostructural map was a key factor in the multi-criteria evaluation process to determine the richest zones of groundwater and drilling sites for augmenting groundwater supply

(BRG) of the study area, 24

November 2000

Lineament density and node density

Linear features can be interpreted as rift, linear valleys, linear

slope breaks or linear ridgelines These features represent

pathways for groundwater accumulation and groundwater

discharge Many studies have been applied to study and

manage groundwater contamination in carbonate aquifers;

the role of lineaments in well yield and groundwater

con-tamination is well noted A high positive correlation

(r = 0.851) was found between lineament length density and

yield, especially where lineaments were cross-cutting (Sener

et al.2005; Tam et al.2005) This indicates a strong

rela-tionship between fracturing and well production

A lineament density map was created using lineament

statistic tool in ArcView 3.1 Rose diagrams of three

dominant directions, i.e., northwest–southeast, northeast–

southwest and north–south were drawn Among these, the

oldest direction is northwest–southeast, which plays a

major role in forming the major hydrogeological structures

The second lineament system divides the major structures into several sub-structures (Fig.3) Further, to integrate information related to cross-cutting lineament, nodes (intersection of lineaments) were extracted and used to calculate a node density map (Fig.5)

The study area shows a lineament density ranging from

0 to 4.1 km/km2 (Fig.5) and a node density ranging between 0 and 5 nodes/km2

Distance from lineament NE–SW

In the study area, there are three major fault systems (Fig.3) The NW–SE trending system is the oldest Part of this system has been re-activated in the subsequent geo-logical times and determines largely the orientation of the geological structure of the study area (Fig.6) The second system is an NE–SW trending system, while the sub N–S trending system was the latest to be formed Moreover, faults NW–SE play an important role in the groundwater

movement of the study area (Tam et al.2005; Tam and

Batelaan 2011) Among four lineament directions, the

authors were interested in the NW–SE direction, as this is

the main direction for collection and movement of

groundwater Water accumulation is highest at the center

of the lineament and depletes as we move away from it To

adequately represent this in a multi-criteria evaluation

model, multiple-buffer of incrementing distances from the

center of the lineaments was made with intervals at 50,

100, 150, 200 and [200 m

Topographic wetness index

The topographic wetness index (TWI) is a function of

natural logarithm of ratio of the local upslope contributing

area and slope The topographic wetness index (TWI) is

frequently used to quantitatively simulate the soil moisture

conditions in a watershed and is the most commonly used

indicator for static soil moisture content Therefore, it plays

an important role in the research of soil erosion and

dis-tributed hydrological model in watersheds and is used to

approximate the local hydraulic gradient under steady state

conditions (Ma et al.2010)

The simplicity of input data make TWI a tool of choice

for groundwater study especially in areas where direct

physical method to understand aquifer is not feasible TWI

is extracted using the DEM alone A flow accumulation

grid (A) was calculated in ArcGIS This was the input into

the equation (Eq.1) for TWI using map algebra

where A is the upslope area contributing water (flow accumulation grid) to the calculation point and b is the local slope gradient

Standardization

To utilize various factors to get to a decision, it is important

to standardize quantities of different type or unit into one scale and one range This is done by the process called standardization

All the above-mentioned factors were reclassified into five levels corresponding to their relationship with groundwater potential In the below-mentioned equation, 1–5 is the score for separate units of each layer

All the standardized factors were masked with the constraint layer (area of habitation between 100 and 700 m elevation in the four districts)

Pairwise weighting and AHP weights

Assigning weight to standardized criteria is important to ensure the relative importance to be used as a factor compensation to arrive at final decision This helps in simulating a real-life scenario where adjustments are made to accommodate difficult choices to meet a greater good

All the factors were compared to each other The

com-parison was done using expert knowledge and previous

researches The comparisons were performed following a

nine-point continuous scale (Saaty1977)

The AHP (Saaty 1977) is based on decomposing a complex MCDM problem into a system of hierarchies (Saaty 1977) The final step in the AHP deals with the structure of an M 9 N matrix (where M is the number of

alternatives and N is the number of criteria) This matrix is

constructed by using the relative importance of the

alter-natives in terms of each criterion The vector (ai1, ai2,

ai3,…, aiN) for each i is the principal eigenvector of an

N 9 N reciprocal matrix, which is determined by pairwise

comparisons of impact of M alternatives on the ith

crite-rion Some evidence is presented by Saaty (1977), which

supports the technique for eliciting numerical evaluations

of qualitative phenomena from experts and decision

makers

Using this method, AHP weights for the criteria were

calculated Consistency index of value \0.1 is considered

indicative of consistent comparison A consistency index of

0.058 is reported for the final pairwise comparison matrix

showing good consistency in assigning comparative degree

of preference among factors (Saaty1977)

Weighted linear combination

Weighting factors ensure compensation of factor

impor-tance while combining them using the linear combination

method The factor layers were multiplied with their factor

weight These weighted factor layers were summed and

averaged (Eq.2) The resultant map was the final

groundwater potential map (Figs.7,8)

M ¼ 1=n X

ai  Ai

where M map of groundwater potential, n groundwater

potential level, a weight of information layer i, i

infor-mation layers (from 1….m), m layer order, A assessed for

separated layer i

Using the weight from Tables1, 2, 3, the formula is

expressed as:

M = (structure *0.31 ? TWI *0.08 ? node density

*0.33 ? distance lineament NE–SW *0.17 ? lineament

density *0.1)/5

Results

Groundwater potential is depicted by a range of continuous

values Due to the application of AHP weights, the final

GWP values were in the same range as the range used for

standardization of factors The GWP map (Fig.7a) shows

the zone of groundwater potential as values ranging from

0.91 to 4.64 Higher values indicated suitable locations for

groundwater potential and thus prospection

This continuous value map was further classified into

potential zones and drilling sites to be a useful and

prac-tical tool for the local government to augment groundwater

supply This information can be further useful in planning

activities, especially new settlement sites in the upland

Validation of the high GWP zones was conducted by detailed geological and hydrogeological field surveys and geophysical measurement (Fig.8) at Tam Son Town of Quang Ba District Geophysical testing with the deep-electronic resistivity measuring technique at 131 sites was conducted in Tam Son Town of Quan Ba District Some potential locations were determined near the maximum GWP point with very low resistivity values ranging from

200 to 300 Xm in comparison to very high resistivity values (up to 4,000 Xm) in neighboring locations (Fig.8) Drilling conducted at site No LKTS1 recorded a good discharge of 7–9 l/s (Nguyen et al 2010)

By further classifying the GWP values in Tam Son

Bold values in the cross line indicate equal importance (value = 1), bold values in the bottom line indicate the total value of pairwise comparison of a factor to others

Distance lineament NE–

SW

Lineament density

Bold values indicate average weights for all criteria and these values