Microsoft Word - 9-2719_s1_ETASR_V9_N3_pp4120-4124 Engineering, Technology & Applied Science Research Vol. 9, No. 3, 2019, 4120-4124 4120 www.etasr.com Japitana et al.: A Geoinformatics-based Framework for Surface Water Quality Mapping and … A Geoinformatics-based Framework for Surface Water Quality Mapping and Monitoring Michelle V. Japitana School of Engineering, University of San Carlos, Cebu City and College of Engineering and GeoSciences, Caraga State University, Butuan City, Philippines mvjapitana@carsu.edu.ph Marlowe Edgar C. Burce School of Engineering, University of San Carlos, Cebu City, Philippines mcburce@yahoo.com Chul-soo Ye Image and Vision System Laboratory, Department of Aviation and IT Convergence, Far East University, Gamgok-myeon, Korea csye@kdu.ac.kr Abstract—The management of water systems must be sustainable, something that requires systematic and comprehensive monitoring. However, the attempts to obtain a comprehensive water quality monitoring (WQM) program in developing countries are challenged due to the lack of an integrated framework and limited resources. At present, the Philippines has no systematized technical and operational monitoring approach, poor coordination and data collection system, and weak law enforcement. On the other hand, Geoinformatics promises a more convenient and cost-effective WQM to complement the traditional method. The goal of this study is to demonstrate how to maximize the use of Geonformatics in developing methods and models to address the lack of spatial trends, timely monitoring, and integrated monitoring framework that could lead to a sustainable WQM system. This study employs remote sensing and GIS technique combined with ground-based water quality data as Geoinformatics-based framework to derive water monitoring and assessment information. Results of this study showed that satellite images can be utilized to derive empirical models to estimate WQ parameters. The validation results showed that the estimated WQ values using the RS-based models have no significant difference when compared with the actual WQ values. Also, the WQ maps derived using Geographic Information System (GIS) were proven useful in providing better representation and analysis of spatial and temporal information that can provide a comprehensive and cost-effective reference for WQ monitoring and assessment. Keywords-landsat; water quality monitoring framework; water quality modeling; water quality trends I. INTRODUCTION Water sustainability is of high importance, as water is one of the most vital resources. Water and its quality are one of the vital elements in the Sustainable Development Goals (SDGs) of the United Nations (UN) Development Program. Water serves as a common link among other SDGs [1], but the world is not yet on track in achieving water-related SDG indicators. In fact, the attempts to obtain a comprehensive water quality monitoring (WQM) program in developing countries are challenged due to the lack of an integrated framework and limited resources. It is, therefore, reasonable to give full attention to how emerging technologies can aid in transforming current water management and water quality monitoring practices, especially in less-developed countries. Geoinformatics emphasized in a formal approach to handle geoinformation and was described as an integrated approach to GIS, photogrammetry, remote sensing, and cartography [2]. Integration of remotely-sensed data, GPS, and GIS technologies provides a valuable tool for monitoring and assessing waterways [3]. There is a high feasibility of integrating Geoinformatics technologies in the existing water quality monitoring framework. Remote sensing technology has allowed measurements on a global scale over long periods and is now proving useful in monitoring coastal waters [4], estuaries [5], and lakes and reservoirs [6, 7]. Biological oxygen demand (BOD) is among the most common water quality parameters measured using remote sensing [8] while RS-based model development studies for water pH are very limited. The pH level is a measure of acid content in water, thus water containing a great deal of organic pollution will generally tend to be acidic [9], while BOD is a measure of the amount of oxygen that bacteria will consume under aerobic conditions while decomposing organic matter [8]. The goal of this study is to develop a Geoinformatics-integrated water quality monitoring framework that can provide comprehensive and cost-effective water quality (WQ) assessments. This study aims to demonstrate the use of remote sensing and GIS technologies for WQ model development and validation, WQ estimation and mapping, and in WQ spatial and temporal trends analysis. II. STUDY AREA Tubay River is one of the classified Class “A” water bodies in Agusan del Norte, Philippines. Its headstream is the main outlet of Lake Mainit and the river traverses the municipalities of Jabonga, Santiago, and Tubay. There are two major tributaries of the river, the Aciga River is at its upstream portion and the Kinahiluan River at the mid-downstream part. Tubay River is situated in a 22,000-hectare catchment area with diverse land uses that include mining, irrigation and other agricultural uses, fishery, livestock production [10], and mining. The Environmental Management Bureau (EMB) pointed out that Tubay River is a receptor of domestic solid and liquid wastes and other non-point sources of pollution [10]. Corresponding author: Michelle V. Japitana Engineering, Technology & Applied Science Research Vol. 9, No. 3, 2019, 4120-4124 4121 www.etasr.com Japitana et al.: A Geoinformatics-based Framework for Surface Water Quality Mapping and … III. PROPOSED FRAMEWORK Figure 1 shows the schematic diagram outlining the proposed Geoinformatics-integrated WQM framework. In the proposed framework, the use of remote sensing and GIS techniques are combined with ground-based water quality data to derive WQ monitoring and assessment information. The developed Geoinformatics-integrated framework is further proposed to be employed to any satellite sensor and WQ of interest. The different components of the proposed framework are described in this section. Fig. 1. Proposed Geoinformatics-integrated WQM framework A. Satellite Data and Water Quality Datasets Quarterly WQM data of EMB for the years 2014-2016 were available during the conduct of this study. The dates of measurements of these datasets that closely matched the time of acquisition of the satellite images were considered in the model development, model validation, and WQ estimation and mapping. Also, satellite images with different dates of acquisition were utilized for the model development and for validating the models. Table I shows the list of Landsat images downloaded from USGS Earth Explorer and employed in delineating river map and in estimating pH and BOD. TABLE I. INPUT IMAGES FOR WATER BODY DETECTION, MODEL DEVELOPMENT AND VALIDATION, AND WQ ESTIMATION. Input Image Acquisition Date Purpose Landsat 8 (bands 2,3, 5-7) March 29, 2017 Waterbody detection Landsat 8 (bands 1-7) April 19, 2015 WQ model development Landsat 8 (bands 1-7) July 26, 2016, August 27, 2016 WQ model validation for pH and BOD Landsat 8 (bands 1-7) April 19, 2015, October 12, 2015, July 26, 2016, August 27, 2016 WQ estimation/mapping B. Image Processing First, the Landsat images were pre-processed to apply radiometric and atmospheric corrections. Then, the surface reflectance bands were used to perform principal component analysis (PCA), band ratio, and multi-band water indices which were computed using (1)-(3) respectively: �� � �� ��⁄ (1) ( ) ( )6363 ρρρρ +−=MNDWI (2) ( ) 7sh 0.25 AWEI ρρρρρ ×−+×−×+= 6532 5.15.2 (3) where �� is ultra-blue (coastal/aerosol) band, � is the blue band, � is the green band, �� is the NIR band, �� is the SWIR 1 band, and �� is the SWIR 2 band of Landsat 8 OLI image. C. Waterbody Detection The MNDWI and AWEI water indices were classified using minimum distance algorithm to delineate Tubay River (Figure 2) from the Landsat 8 image. The delineated river map is employed as a mask image during river WQ maps generation. Fig. 2. Derived river map using the combination of MNDWI and AWEI water index. D. Water Quality Model Development and Validation The WQ models for pH and BOD were derived as described in [11]. The two remote sensing-based WQ models were given as follows: � � 8.339 � �0.827 � ��� (4) ��� � 0.382 � �28.746 � PC4� (5) The models were then applied to the corresponding input bands and the estimated WQ values for pH and BOD at the specific locations of the EMB monitoring stations were calculated. To evaluate the reliability of the models, t-tests were performed by comparing the estimated WQ values with that of the actual values measured by the EMB. E. WQ Estimation and Mapping The validated WQ models were applied to the whole input bands of using Band Math of ENVI 5.1 software to derive the WQ images. Then, the WQ images were loaded in QGIS 2.16 software to generate the WQ maps. Using QGIS, non-water areas in the WQ image were masked out using the river map as mask image. And finally, gradient colors and histogram Engineering, Technology & Applied Science Research Vol. 9, No. 3, 2019, 4120-4124 4122 www.etasr.com Japitana et al.: A Geoinformatics-based Framework for Surface Water Quality Mapping and … stretching were applied for better representation of the spatial and temporal distribution of pH and BOD in the study area. IV. EXPERIMENTAL RESULTS The models were validated using two sets of Landsat images, the validation results with the best performance are shown in Tables II and III, which show the estimated versus the actual WQ data and the result of the t-test for pH and BOD, respectively. The differences between the estimated and actual pH values are very small having an average value of 0.11. At EMB stations 3 and 5, the estimated pH value is very close to the actual pH value with difference values of 0.04 and 0.08, respectively. When t-test was performed to compare the two datasets for pH, the result showed that the t value is less than the critical t value resulting to a p-value higher than 0.05, hence there’s no significant difference between the two measurements. Table III shows that the average difference between the two BOD measurements is 0.32. The lowest difference values between the predicted and the actual BODs are at EMB stations 1 and 7. The t-test results showed a p-value greater than 0.05, hence the null hypothesis that there is a significant difference between the two groups of measurements can be rejected. TABLE II. ESTIMATED AND ACTUAL PH Station pH Difference t-test Estimated Actual t value p-value 1 8.12 7.94 0.18 t=2.08