Identification of Groundwater Potential Zone using Multi-Influence Factor Technique (Study Case: Brantas Groundwater Basin, East Java, Indonesia)

Brantas Groundwater Basin is the biggest groundwater basin in East Java Province, Indonesia. It is covering 22 regencies/cities which have high water need. Mostly people water need is supplied from groundwater. Identification of Groundwater Potential Zone (GWPZ) is required to ensure sustain groundwater supply for fulfilling that water need. One technique for that is Multi-Influence Factor (MIF). It considers influence factors of groundwater such as rainfall, geology, geomorphology, slope, lineament density, drainage density, soil texture, and land-use/landcover. Raster data obtained from Google Earth Engine (GEE), Aster DEM, and Geological Map of East Java were included as the research data. They had score and were estimated by using MIF technique. GEE and open-sourced GIS were used in computational raster data processing of MIF technique. Identification result of GWPZ showed that Brantas Groundwater Basin consist of zones which are very poor (2% area), poor (24% area), good (47% area), and very good (17% area). The result can assist hydrogeologist and local authorities to formulate further policy of GWPZ management.


Introduction
People water need supplied groundwater will always increase along with raising number of people and their various activities. Groundwater is used for various needs including drinking water, domestic use, farming, and industry [1]. High number people growth will cause water need increase and groundwater overuse risk [2] [3]. The further impact of that is higher pressure on aquifer and land degradation [3] [4].
Brantas Groundwater Basin is covering 22 regencies/cities in Eat Java, Indonesia. Total area of Brantas Groundwater Basin is up to 9995 km 2 (20% area of East Java Province). The fresh water need in East Java is high, but its fulfillment rate is only 60% [5]. Fresh Water Regional Company takes role in management of that fresh water fulfillment which production effectivity is up to 87.36% [6]. It manages the fresh water that comes from any sources such as river (58,61%), groundwater (20,23%), water spring (18,02%,), reservoir (0,75%), and others (2.39%) [6]. Water spring is part of groundwater [7]. The proportion of fresh water source turns to be 38,35%, if groundwater and water spring are estimated into one. Acquisition data of water source did not yet involve groundwater used by people who have not get yet access the pipelines network owned by Fresh Water Regional Company. Thus, fresh water sources for people come only from both groundwater and rivers.
Sustainable groundwater can be carried through identification of groundwater potential zone (GWPZ). A lot of data and methods are able to be applied to determine GWPZ which are controlled by lithology, geophysical method, remote sensing and GIS, and cloud-based computing (for instance Google Earth Engine) [4][8] [9]. The research objective is identifying GWPZ in Brantas Groundwater Basin, East Java. GWPZ technique applied is Multi-Influence Factor by using Geographic Information System (GIS) [8][10] [11][12] [13].

Methods
The research data come from any source, such as Google Earth Engine (GEE), Aster DEM, dan Geology Map of East Java Province. GEE is digital platform using cloud-based computing with high computational performance. GEE is broadly used in climate monitoring, water management, dan environmental protection [4]. GEE has public data archives that can be accessed easily without data downloading process. GEE provides multi-petabyte data which are ready to be analyzed. Data are only able to be accessed by Internet-accessible Application Programming Interface (API) and an associated web-based Interactive Development Environment (IDE) [14]. This research used some data obtained from GEE comprising USGS Landsat 8 Collection 1 Tier 1 and Real-Time data Raw Scenes [15], OpenLandMap Precipitation [16], OpenLandMap Soil texture class (USDA system) [17], and Global ALOS Landforms [18] . USGS Landsat 8 Collection 1 Tier 1 and Real-Time data Raw Scenes are part of Landsat 8 TIRS satellite image with 30 meter in spatial resolution. The data were processed by using CART Classifier to generate Land-use/Landcover Map. OpenLandMap satellite image was used to obtain Rainfall Map in 1000-meter spatial resolution. Soil Texture Map was acquired based on Soil Texture Class (USDA System) from OpenLandMap satellite image in 250-meter spatial resolution. Global ALOS Landforms satellite image was processed to deliver Geomorphology Map with spatial resolution of 90 meter.
Other raster data involved in this research is ASTER DEM with 30 meter in spatial resolution. ASTER DEM provides Digital Elevation Model (DEM) data which were able to obtain lineament data, channel network and slope. Quantum GIS (QGIS) software was used to process those research data. By using Inverse Distance Weighted (IDW) tools of this software, both Lineament Density Map and Drainage Density Map were created. Geology map as conventional data map published Center of Geology and Environment Research, Ministry of Energy and Mineral Resource, Republic of Indonesia was also used in this research. By using that map, any geology material information could be generated. That map was converted to raster data, then. All maps were reclassified on each subclass based on MIF provision. The next process was computing process using MIF technique as shown in Table 1. Flowchart of methodology can be seen in Figure 2.
Influence factors of GWPZ are rainfall, geomorphology, geology, drainage density, lineament density, land-use/landcover, soil texture, and slope. Inter-relationship of those factors is shown in Figure  3. The factors causing mayor impact were weighted by more than 1.0 value, whereas the factors causing

Weightage Calculation
Each influence factor has different classification weight. Relationship of groundwater potency influencing factors is shown in Figure 2 and impact of those factors is shown in Table 2. Major effect (A) is the factor causing more impact to other factors. Less impact to other GWPZ's factors is delivered by the factor called as Minor effect (B). Value of each factor is accumulation value of relationship factor to other factors. Relative weight is accumulation A and B, whereas assigned weight of each influence factor is proportion relative weight to total weight. Assigned weight was considered as rank factor when weighted overlay was being processed by QGIS.  (Table 1), whereas r is factor weight ( Table 2).  [22]. Variation of seasonal rainfall will impact to different groundwater fluctuation [20]. Rainfall does not only influence groundwater amount, but it also influences renewed rapidly rate of water to shallow aquifer [23]. OpenLandMap Precipitation GEE was used to derive rainfall data [16]. The data was in monthly rainfall data form. Satellite image processing using GEE provides annually rainfall data (Fig.3). Brantas Groundwater Basin has various rainfall areas which are 1350 -2000 mm/year (28% area), 2001 -2500 mm/year (49% area), more than 2500 mm/year (22% area). Rainfall is weighted between 6 and 9. The more intense rainfall is in the area; the more class weight value is given (Table 1). Rainfall has 10% factor weight of influencing factor total to groundwater potency ( Table 2).

Geology
Geology data used in this research is the published data by Center of Geology Research, Ministry of Energy Mineral Resource, Republic of Indonesia. Geology materials information of Brantas Groundwater Basin shows that the basin material consists of breccia and conglomerate (4%), lava, pyroclastic, tuff, polymict (14 %), claystone material (4%), marl and limestone (2%), sandstone (18%), and alluvium (34%) (Fig. 5). Geology condition directly impacts to groundwater potency [25]. Rocks control the aquifer by its porosity and permeability. Rocks having fracture, crack, and are under high weathering rate are good aquifer [26]. Alluvial area is also good aquifer, so it makes to be easy to exploit water through well [25]. Good aquifer can also consist of limestone, sand dune and sandstone [8] [27]. Materials having high potency to accumulate groundwater were more weighted than others (Table 1). MIF technique estimation indicated that geology factor has 23% weight compared other factors ( Table  2).

. Drainage Density
Drainage density has important role in groundwater recharging process [25]. High drainage density areas showed that they have high percolation and permeability of the lithology [28]. ASTER DEM was used to derive drainage density information. By using Channel Network tool of QGIS software, DEM data was processed. Channel Network Map was used to generate Drainage Density Map. Brantas Groundwater Basin has 3 classes of drainage density which are 0 -0,5 km/km2 (30% area), 0,5 -0,1 km/km2 (60% area), and more than 0,1 km/km2 (10% area) (see Fig. 6). High drainage density has higher weight value than low one (Table 1). Compared to other influence factor of groundwater potency, drainage density has 13% weight ( Table 2).

Lineament Density
Lineament density was generated by 30 meter ASTER DEM processing. Lineament is useful to determine secondary porosity of a fracture system and or groundwater moving and storing channel [22]. Lineament is long fracture and has high groundwater potency [29]. Lineament Density Map of research area was classified into 3 classes, which are class of 0 -0.5 km/km 2 , class of 0.5 -1.0 km/km 2 , and class of 0.1 -0.45 km/km 2 (Table .1, and Fig. 7). High lineament density will increase infiltration, so it was more weighted than low one [30]. Lineament Density has factor 18% weight of all influence factors for groundwater potency determination (Table 2).

. Soil Texture
Soil texture data was obtained from OpenLandMap Soil texture class (USDA system) which are available in GEE. Soil texture classification used USDA System. OpenLandMap satellite image is equipped with bands indicating soil texture based on soil depth, so it can also be involved as research data need [17]. Soil texture is soil property related to groundwater movement such as infiltration and percolation process [31]. Soil texture consist of 3 textures, such as sandy texture which has high permeability, loamy texture which has moderate permeability, and clay texture which has low permeability [22]. Soil textures in the research area were grouped to be sandy loam, sandy clay loam, clay loamy soils, clay soils (Fig 8). Sandy texture has higher class weight than loam and clay textures ( Table 1). Estimation of MIF technique show that soil texture has factor weight of 3% (Table 2).

Slope
Slope map was processed using QGIS Software on 30 meter-ASTER DEM. Slope correlates both locally and regionally with relief and terrain. Steep slope areas has low groundwater potency, since of high runoff and low infiltration [13]. Water spring can be found over there if there are weathered rocks that make them to be high porosity and permeability [32]. Slope has low influence to infiltration capacity, if the area is located on impermeable soil [31]. Mostly research area takes place in gentle slope (0 -15o). The steep slope which is more than 15o dominates in the mountainous area and hills (Fig 9). Compared with gentle slope, it has lower weight value since of low groundwater potency (Table 1). Slope has 10% factor weight compared with other factors based on MIF technique (Table 2).

. Land-use/Landcover
Land-use/Landcover (LULC) is an important factor for determination of groundwater potency [33], [34]. LULC was derived from USGS Landsat 8 Collection 1 Tier 1 and Real-Time data Raw Scenes processed by using GEE [15]. Landsat 8 TIRS 1 satellite image has been processed by applying CART Classifier. CART applied identification and construction of binary decision tree involving well known classified training data sample [35]. CART Classifier was able to deliver good result in LULC analysis [36]. Analysis result of CART Classifier for LULC in Brantas Groundwater Basin is shown as Figure  10. LULC is divided into 4 classes which are settlement (29% area), scrubland (9% area), forest (48% area) dan plantation/agriculture (14% area). Forest is more weighted than other classes, since it can increase groundwater recharging [33], [37]. By MIF technique estimation, LULC has 10% factor weight of all GWPZ influence factors.

Identification of Groundwater Potential Zone
MIF technique was applied by using QGIS software. Determination process of GWPZ considers factor weight ( Table 2). The most GWPZ influence factor is geology, and the least one is soil texture. Result of MIF technique was grouped into 4 zones of Brantas Groundwater Basin which are very poor (2% area), poor (24% area), good (47% area), and very good (17% area) (Fig.11). Mostly Brantas Groundwater Basin is good and very good groundwater potential zone (64% area). The zones consist of IOP Publishing doi:10.1088/1755-1315/1066/1/012004 10 alluvium material, were in lower and gentle slope, and has high drainage density and low lineament density. The zones cover some areas which are: part of Trenggalek Regency, part of Tulungangung Regency, Kediri City, Kediri Regency, Jombang Regency, Mojokerto City, and part of Sidoarjo Regency. Poor and very poor groundwater potential zone (46% area) spread to areas located in upper and steep slope having low drainage density and high lineament density. The zones cover areas such as: Ponorogo Regency, part of Tuluangung Regency, part of Mojokerto Regency, part of Madiun Regency, part of Bojonegoro Regency, part of Kediri Regency, part of Gresik Regency, and Part of Surabaya City. Other areas clustered as poor and good zone are Malang Regency, Malang City, Batu City, Blitar Regency, and Blitar City.

Conclusions
Identification of groundwater potential zone in Brantas Groundwater Basin is very important to prevent water deficiency. Geospatial technology is very rapidly developed, for instance Google Earth Engine (GEE) which are applying cloud computing. GEE delivers wider coverage and provides much more data availability. Open source-based geospatial data processing has been implemented for remote sensing and GIS data processing. Multi-Influence Factor technique used 8 factors for determination of groundwater potency such as rainfall, geology, geomorphology, drainage density, lineament density, slope, soil texture, and landuse/landcover. Result of MIF technique indicate that groundwater potential zone of Brantas Groundwater Basin consists of zones which are very poor (2% area), poor (24% area), good (47% area), and very good (17% area). The research result can be useful for empirical study of groundwater resources management.