ReseaRch PaPeR Journal of Agricultural and Marine Sciences Vol. 24 : 44– 50 DOI: 10.24200/jams.vol24iss1pp44-50 Reveived 25 Nov 2018 Accepted 15 Mar 2019 Evaluation of salinity intrusion in arable lands of Al-Batinah coastal belt using unmanned aerial vehicle (UAV) color imagery *Sawsana Hillal Al-Rahbi, Yaseen Ahmed Al-Mulla, Hemanatha Jayasuriya *Sawsana Al-Rahbi ( ) s.h.rahbi1@gmail.com, Department of Soils, Water and Agricultural Engineering, ollege of Agricultural and Marine Sciences, Sultan Qaboos University, Box 34, Al-Khod 123. Introduction The Sultanate of Oman is considered as an arid country with an average annual rainfall of about 100 mm. Although agriculture occupies about 5% of the total area of Oman distributed around eleven governorates (regions), the agricultural sector consumes more than 93% of the total water demand. Al-Batinah North governorate represents the largest area of Oman’s agricultural lands (Table 1) covering 24% of the total agricultural area of the country and is considered to be Oman’s most important agricultural area as it produc- es 65% of the Omani agricultural production with crops such as dates, fruits, vegetables and forage crops such as alfalfa and Rhodes grass (Choudri et al. 2015a; Choudri et al. 2013). The population in Al-Batinah has sharply increased since 2003: i.e. by more than 100,000 people within 7 years. Subsequently, the socioeconomic activities; active ports, coastal tourism projects, industrial activities, in- frastructure development, intensive agriculture and ur- banization have been rapidly taken place in this coastal zone (Choudri et al. 2015c). Such activities are related to population growth and have led to increasing pressures on natural resources including groundwater, agriculture and land use (Lawley et al. 2016). It also has resulted in some environmental challenges such as seawater intru- sion, water and soil salinition and desertification (Chou- dri et al. 2015a). The management plans to mitigate the environmental challenges are constrained by the shortage of informa- tion about the interaction between the development ac- tivities and the environment (Rishi and Mudaliar 2014). In general, lack of information about the global, national and local land resources may lead to management plans تقييم تسرب امللوحة يف احلزام الزراعي يف ساحل الباطنة ابستخدام الصور امللونة امللتقطة من طائرة بدون طيار سوسنة بنت هالل الرحيب* وياسني بن أمحد املال وهيماناثا جاياسوريا Abstract. Salinity by seawater intrusion due to excess groundwater pumping for irrigation is a major environmental challenge in the coastal areas of the Sultanate of Oman. Increasing salinity levels moving inward the arable lands is happening in a rapid manner. Thus, salinity needs to be evaluated and quantified using a fast and accurate method. The objective of this study was to estimate salinity intrusion in Al-Batinah coastal belt using color aerial imaging. The study was conducted in five randomly selected sites at increasing distances from the seashore of Al-Suwaiq area in Al-Batinah region of northern Oman. Color aerial images were acquired for each site with an Unmanned Aerial Vehicle (UAV). Images were enhanced by orthorectification in ENVI software. A Green Leaf Index (GLI) was obtained from each site image using Matlab software. Image analysis results were compared with the results of analyzed soil and water samples taken for ground-truth verification. There was a strong negative correlation between the distance from the seashore and the soil EC of each site (R = -0.95). Similarly, the mean value of GLI increased as the salinity levels decreased, R= -0.96 and -0.92 for soil EC and water EC, respectively. We demonstrated the possibility of the use of color images taken by a UAV to accurately quantify the effect of soil salination on vegetation along the costal belt. Keywords: Seawater Intrusion; Salinity Dynamics; UAV; Image Processing; GLI. املســتخلص: متثــل امللوحــة حتديًــا بيئًيــا كبــرًا يف املناطــق الســاحلية يف عمــان. حيــث أن حركــة امللوحــة بإجتــاه األراضــي الداخليــة الصاحلــة للزراعــة تســر بســرعة عاليــة، لــذا فهنــاك حاجــة ماســة للكشــف عــن حركــة التملــح وتقييمهــا باســتخدام طريقــة ســريعة ودقيقــة. اهلــدف مــن هــذه الدراســة هــو تقييــم حركــة امللوحــة داخــل حــزام الباطنــة الســاحلي باســتخدام تقنيــات التصويــر اجلــوي امللــون. وقــد أجريــت الدراســة يف مخســة مواقــع خمتــارة بطريقــة عشــوائية تبعــد مبســافات خمتلفــة عــن شــاطئ واليــة الســويق. مت إلتقــاط الصــور اجلويــة امللونــة لــكل موقــع بواســطة طائــرة بــدون طيــار. ومت حتســني الصــور هندســيا كخطــوة اوىل يف عمليــة حتليــل الصــور اجلويــة. مث مت حســاب مؤشــرإخضرار األوراق )GLI( املســتنبط مــن صــورة كل موقــع. بعــد ذلــك متــت مقارنــة حتاليــل الصــور مــع حتاليــل الرتبــة واملــاء يف عمليــة التحقــق وربــط املســتنبط بالواقــع. كان هنــاك ارتبــاط ســليب قــوي بــني املســافة مــن شــاطئ البحــر وملوحــة الرتبــة لــكل موقــع )معامــل إرتبــاط =0.95-(. وباملثــل، زادت قيمــة )GLI( مــع اخنفــاض مســتويات امللوحــة، معامــل إرتبــاط = 0.96- مللوحــة الرتبــة ومعامــل إرتبــاط = 0.92- مللوحــة امليــاه. أوضحــت نتائــج هــذا العمــل البحثــي إمكانيــة اســتخدام الطائــرات بــدون طيــار مثبتــة بكامــرا ملونــة لتقديــر وحتليــل تأثــر البعــد عــن شــاطئ البحــر علــى مســتويات امللوحــة يف الرتبــة وامليــاه، وكذلــك علــى حالــة الغطــاء النبــايت يف األراضــي الصاحلــة للزراعــة مبنطقــة الباطنــة. الكلمات املفتاحية: تسرب امللوحة، ديناميكية التملح، الطائرات بدون طيار، حتليل الصور، مؤشر إخضرار األوراق 45Research Article Al-Rahbi, Al-Mulla, Jayasuriya without environmental concerns (Mulder et al. 2011). Thus, there is a necessity for accurate, cost-effective and timely monitoring method to update the information on the status changes in the arable lands of coastal area (Mishra 2014), in order to develop a framework for the decision makers to manage the environmental problems. Bajjali (2003) has conducted a study to assess the ground water quality in Oman by analyzing 20,000 wells across different regions. The study indicated that Al-Batinah coast is the most affected area with ground- water salinity in Oman, where the water salinity ranges from 5 to 44 dS/m (Choudri et al. 2015c). As reported by Choudri et al. (2015b), Ministry of Regional Municipal- ities and Water Resources collected salinity data from 18 different wells in Al-Batinah region during the years 1991, 1993, 2005 and 2010 (Table 2). The collected data suggested that water salinity has increased gradually in all examined wells within the last two decades. Further- more, water salinity is an important factor in soil salinity (Al-Belushi 2003; Hussain 2005). Approximately 52% of Al-Batinah lands are affected by soil salinity (Al-Mulla et al. 2010). Between the years 2000-2005, the percentage of the agricultural lands affected with soil salinity has in- creased by about 7% (Al Barwani and Helmi 2006). In addition, soil salinity is considered as one of the main reasons of desertification in arid and semi-arid regions and so in Al-Batinah coast particularly (Al-Belushi 2003; Choudri et al. 2015b). On the other hand, soil salini- tion is considered as one of the main reduction factors of Omani dates exportation which decreased by 2,000 MT within a 5-year period (2007-2011). Similarly, pro- duction of date palm in Al-Batinah region has steadily declined within the last few years mainly due to ground- water salinity (Al-Yahyai and Khan 2015). Although there are many studies investigated the salinity levels in Al-Batinah region using the tradition- al field visits and lab analysis, there is no documented evidences on evaluation of salinity change inward the coastal belt, and particularly using areal imaging tech- nique. Therefore, the objective of this study was to an- alyze salinity change inward Al-Batinah coastal belt us- ing images collected from an unmanned aerial vehicle (UAV) combined with color imaging techniques. Table 1. Area of the agricultural lands in each governorate of Sultanate of Oman (feddan) Governorate Agricultural land Area (feddan)* Percentage % Muscat 11,555.85 3.26 Dhofar 65,921.13 18.57 Musandum 3,242 0.91 Al Buraimi 16,123.21 4.54 Ad Dakhiliyah 45,732.97 12.88 Al Batinah North 85,118.27 23.98 Al Batinah South 48,984.53 13.80 Ash Sharqiyah South 15,206.87 4.28 Ash Sharqiyah North 27,523.27 7.75 Adh Dhahirah 33,295.08 9.38 Al Wusta 2,307.9 0.65 Total 355,011.1 100 *(M.A.F 2013) Figure 1. The Study Area location 46 SQU Journal of Agricultural and Marine Sciences, 2019, Volume 24, Issue 1 Evaluation of Salinity Intrusion in Arable Lands of Al-Batinah Coastal Belt Using Unmanned Aerial Vehicle (UAV) Color Imagery samples (one sample from each site) were collected from the irrigation water sources (wells) of each site and were kept in a clean plastic container and transferred to the lab for analysis. The EC of each water sample were mea- sured using EC meter. Image Acquisition Aerial images were taken by a digital color camera with 12.4 megapixels resolution. The camera was mounted on a quadcopter UAV (model: phantom-3-pro, DJI INC., China). Site images were taken from (130-275) m above the ground according to each site area. The captured images were saved in JPG format, which is a common format for realistic images and readable in different im- age processing softwares. The images were transferred to the computer to be analyzed. Image Analysis Orthorectification was conducted as a pre-processing technique in order to enhance the site images and to decrease image distortion. The Environment for Visual- izing Images (ENVI) software (version 5.0.3, Exelis Visu- al Information Solutions INC., US) was used for image Orthorectification using Ground Control Points (GCPs) and Replacement Sensor Model (RSM). The GCPs were collected using Google Earth software (Version: 7.1.7.2600, Google INC.). Several vegetation indices which depends only on color bands; Green Leaf Index (GLI), Visible Atmo- spherically Resistant Index (VARI) and Triangular greenness index (TGI). GLI has been commonly used in thresholding the green vegetation in aerial images of canopy scales (Chianucci et al. 2016; Hunt Jr et al. 2013; Macfarlane and Ogden 2012). Thus, GLI was computed (Eqn. 1) to determine canopy attributes within each site Materials and Methods Study Area The study was conducted in Al-Suwaiq area (23° 50’ 58” N, 57° 26’ 19” E). It is located at the south part of Al-Batinah North governorate (Fig. 1). The climate of Al-Suwaiq is characterized as dry with average annual humidity of 32% and high evapotranspiration rate. The average air temperature of the coastal area is 28.5 °C and 17.8 °C in the mountain area. The average rainfall rate in Al-Suwaiq (as a part of Al-Batinah region) is 50 mm/year, varying in time and places within the region (Kwarteng et al. 2009) . Sites Selection Five sites were randomly selected within 0.3 to 6 km in- land distance from the seashore of Oman through the agricultural land within the study area. Samples Collection and Analysis In each of the five randomly selected sites, five locations were selected randomly to collect soil samples. A Glob- al Positioning System device (Garmin eTrex Legend Cx GPS, USA) was used to register each location coordi- nates. At each location, three soil samples were collect- ed from three different depths; 5 cm, 20 cm and 50 cm. Around 500 g of soil were taken with an auger to repre- sent each depth. Each sample was kept in a clean plastic bag and annotated separately. A total of 75 soil samples were collected representing 5 sites × 5 locations × 3 depths. The saturation method was used to obtain soil extract from soil samples. Each soil extract was investi- gated for electrical conductivity (EC) which is expressed by deci-Siemens per meter (dS m-1). In addition, water Figure 2. The locations of the selected sites 47Research Article Al-Rahbi, Al-Mulla, Jayasuriya using MATLAB software (Version: 9.0.0.341360, Math- work INC., USA). GLI= (2G-R-B)/(2G+R+B) (1) Where G, R and B are the digital values (0-255) of the green, red and blue bands of each pixel . The GLI value of each pixel in the site image were calculated using equa- tion 1. The GLI values were reconstructed by applying the MATLAB function (inpaint_nans.m). Then, the GLI pixel values were averaged to get the GLI value of the whole image. Statistical Analysis Pearson correlation coefficient was calculated to compare soil and water EC of each site with site distance from seashore. Regression analysis were used to esti- mate soil and water salinity using the distance from the seashore and the value of GLI. The method followed in this paper is illustrated in Figure 3. Table 3. The correlation coefficient of each salinity parameter Salinity Parameter Correlation Coefficient Soil EC (5 cm) -0.94992 Soil EC (20 cm) -0.87461 Soil EC (50 cm) -0.7105 Water EC -0.48239 Table 2. Location of salinity monitoring wells in Al-Batinah region with the observed salinity (ppm) in 1991, 1993, 2005 and 2010 Well ID Location (E) Location (N) 1991* 1993* 2005* 2010* N-101 578701 2621460 1504 1632 2112 3072 N-92 582083 2620091 839 833 835 849 T-52 584162 2622750 1606 2214 8262 12288 N-79 585655 2617956 800 931 1280 1798 B-49 586184 2622605 7379 8896 8979 9126 T-30 591740 2621062 9280 9421 10682 14784 N-107 575993 2623943 1187 1112 5114 10432 B-70 571376 2627585 5440 6573 6144 11520 B-73 572962 2627174 7571 7424 8800 9728 B-83 568271 2628276 8410 6298 9600 12160 T-46 585991 2621972 6720 7507 13120 16576 N-63 590404 2619803 1382 1312 5133 14656 B-31 594298 2620548 4032 4902 11494 11514 N-53 591385 2616842 1344 672 1293 1792 N-71 587411 2619729 1427 1267 1958 3590 N-111 568658 2623958 774 833 1760 1837 N-66 588832 2617526 2138 1760 1978 2323 T-85 569008 2627063 3994 3610 6278 8896 *Observed salinity concentration (ppm#) #ppm = dS/m x 640 (EC = 0.1 to 5 dS/m), ppm= dS/m x 800 (EC > 5 dS/m) Soil and Water Sample Collection Aerial Image Acquisition (UAV) Image Enhancement (Orthorectivication) Color Band Separation (R, G, B) Mean value of Enhanced GLI Regression Analysis Salinity vs GLI Salinity Analysis (EC) Figure 3. Soil EC (dS m-1) of each site (1-5) at different soil depths (5, 20 and 50 cm) 48 SQU Journal of Agricultural and Marine Sciences, 2019, Volume 24, Issue 1 Evaluation of Salinity Intrusion in Arable Lands of Al-Batinah Coastal Belt Using Unmanned Aerial Vehicle (UAV) Color Imagery Results Soil and Water Analysis The level of soil EC of the collected samples at different depths of 5, 20 and 50 cm were decreased with the in- crease of the soil depth, as shown in Figure 4. Effect of Seashore on Salinity Levels The effect of site location from seashore on salinity levels is illustrated in Figure 5. As the distance between the seashore and the selected sites increased, the water EC and soil EC decreased gradually. To investigate the effect of site location on the salinity levels, the water EC soil EC at different depths (5, 20, 50 cm) were correlated with the distance from the seashore as shown in Table 3. Water EC had the lowest correlation with the distance from the seashore (R = -0.48). On the other hand, the EC of the soils in 5 cm depth had the highest correlation. Regression analysis was done to estimate the soil EC in 5 cm depth by knowing the distance from the shore using Eqn. 2 (Fig. 6). y= -2.7671 x+22.643 (R2 =0.902) (2) Where y is the EC (dS/m) of the top layer of soil and x is the distance from the seashore (km) to the selected site. Image Analysis The image of site 2 is shown as an example in Figure 7. The averaged values of GLI had a strong negative cor- relation with soil EC (R= -0.96) and water EC (R= -0.92). The GLI value of site image can be used to estimate site soil EC (Eqn. 3) and water EC (Eqn. 4) (Fig. 8). y= -2.0737 x+16.15 (R² = 0.9128) (3) Where y is the soil EC (dS/m) and x is the mean value of GLI of the site image. y = -2.3241x + 13.887 (R² = 0.8429) (4) 0 5 10 15 20 0 2 4 6 8 Distance from the sea (km) E C (d S /m ) Variable SoilEC WaterEC Figure 5. Soil and water EC (dS m-1) of each site (1-5) and site distance from the seashore 0 5 10 15 20 25 0 2 4 6 Distance from the sea (km) E C (d S /m ) Depth 5cm 20cm 50cm Water Figure 6. The regression analysis between the sites’ dis- tance from the seashore and their water EC and soil EC in different depths (5,20,50 cm) 0 5 10 15 20 25 5 25 50 Soil Depth (cm) E C (d S /m ) Location Site1 Site2 Site3 Site4 Site5 Figure 4. Steps followed to estimate soil and water salinity using GLI 49Research Article Al-Rahbi, Al-Mulla, Jayasuriya Where y is the water EC (dS/m) and x is the mean value of GLI of the site image. Discussion Soil and water salinity decreased as the site is located farther from the seashore (Fig. 5). Site (3) showed rela- tively unexpected increase in salinity levels, which could be due to the farming practices in the site. All other sites showed clear negative correlation between the distance to the site from the seashore and the salinity levels. On the other hand, an excellent correlation was observed while investigating the EC of soils from different depths with the distance from seashore (Table 3). The top soil layers showed the highest values of soil salinity, where that could be due to salt accumulation on the soil sur- face as reported by Herrero et al. (2003). It also had the strongest correlation (R= -0.95) with distance from the seashore. The GLI mean values of the images ranged from -1.2 to 6.8. The positive value of GLI was assigned to the green leaves or stems while the negative value was for non-green site objects like; soils, buildings, woods and other non-living items (Louhaichi et al. 2001). In this study it was found that the lowest mean value of GLI was -1.2 for site 1 with the highest salinity level. In gen- eral, the results proved that the soil and water salinity had strongly affected the vegetation quantity and quality (greenness), where the mean green value (GLI) declined as the salinity increased. Vegetation Soil Salinity Index (VSSI) were used by Tran et al. (2018) to estimate salin- ity intrusion from Landsat 8 images with R2 = 0.6957. The salinity levels can be estimated by the mean value of GLI with relatively strong values of coefficient of deter- mination, compared to other vegetation indices. Conclusion This research proved the effect of salinity intrusion on site location from the seashore. The five randomly se- lected sites within the agricultural land belt with differ- ent distances from the seashore showed a decline in sa- linity levels as the site become far from the seashore. The effect of distance on soil salinity could be represented as a regression model. Mainly, this research demonstrated the possibility of using UAV with affordable digital cam- era to estimate the vegetation cover. The results showed Figure 7. Orthorectification process of site number 2 0 5 10 15 20 −2.5 0.0 2.5 5.0 7.5 GLI (Average) E C (d S /m ) Substrate Soil Water R2 = 0.91R2 = 0.84 Figure 8. The regression analysis between GLI mean value of each site image and salinity levels of the site 50 SQU Journal of Agricultural and Marine Sciences, 2019, Volume 24, Issue 1 Evaluation of Salinity Intrusion in Arable Lands of Al-Batinah Coastal Belt Using Unmanned Aerial Vehicle (UAV) Color Imagery a strong negative correlation between salinity levels and GLI as an indicator of vegetation status. Salinity assess- ment using UAV colour images is coast efficient, time- less and more accurate in relative to field and satellite assessments. Nevertheless, more image processing tech- niques may strength the possibility of aerial images in estimation of salinity effects on vegetation cover. Acknowledgement The authors thank The Research Council (TRC) of Sultanate of Oman for funding this study (Project No. ORG/EBR/13/004 - Coastal Ecosystem Management for Sustainable Development: A Case study of Al-Batinah Region, Oman). A special thanks for Dr. B. S. Choudri as he was the principle investigator (PI) of the project. References Al-Belushi A. 2003. Desertification in Al-Batinah plain, Sultanate of Oman. PhD Dissertation. Jordanian Uni- versity, Jordan (in Arabic). Al-Mulla Y, Ahmed M, Al-Rawahy SA, Hussain N. 2010. Salinity mapping in oman using remote sensing tools: Status and trends. Published in the Monograph on Management of Salt-Affected Soils and Water for Sustainable Agriculture (Mushtaque A, Al-Rawa- hi SA, Hussain N (eds)). Sultan Qaboos University, Oman. 17-24. Al-Yahyai R, Khan MM. 2015. Date palm status and per- spective in Oman. Date palm genetic resources and utilization. Springer. p.207-240. Al Barwani A, Helmi T. 2006. Sea water intrusion in a coastal aquifer: A case study for the area between Seeb and Suwaiq, Sultanate of Oman. Journal of Ag- ricultural and Marine Sciences [JAMS]. 11: 55-69. Bajjali W (2003) Evaluation of the groundwater salinity throughout Sultanate of Oman using GIS. Available at: http://frontpage.uwsuper.edu/bajjali/proj/oman/ o5.htm), last consulted on 23 nov. 2009. Chianucci F, Disperati L, Guzzi D, Bianchini D, Nardino V, Lastri C, Rindinella A, Corona P. 2016. Estimation of canopy attributes in beech forests using true co- lour digital images from a small fixed-wing uav. In- ternational journal of applied earth observation and geoinformation. 47: 60-68. Choudri B, Baawain M, Al-Sidairi A, Al-Nadabi H, Ahmed M. 2015a. Relative vulnerability of coast- al wilayats to development: A study of Al-Batinah North, Oman. Journal of Coastal Conservation (Springer Science & Business Media B.V.). 19(1): 51- 57. Choudri BS, Al-Busaidi A, Ahmed M. 2013. Climate change, vulnerability and adaptation experiences of farmers in Al-Suwayq wilayat, Sultanate of Oman. In- ternational Journal of Climate Change Strategies and Management. 5(4): 445-454. Choudri BS, Baawain M, Ahmed M. 2015b. Review of water quality and pollution in coastal areas of Oman. Pollution Research, Ref 34(2): 229-239. Choudri, B.S., Baawain, Mahad and Ahmed, Mushtaque (2016). An overview of Coastal and Marine Resourc- es and their Management in Sultanate of Oman, Journal of Environmental Management and Tourism, (Volume VII, Spring), 1(13): 21-32. DOI:10.14505/ jemt.7.1(13).02. Herrero J, Ba A, Aragüés R. 2003. Soil salinity and its distribution determined by soil sampling and elec- tromagnetic techniques. Soil Use and Management. 19(2): 119-126. Hunt Jr ER, Doraiswamy PC, Mcmurtrey JE, Daughtry CS, Perry EM, Akhmedov B. 2013. A visible band in- dex for remote sensing leaf chlorophyll content at the canopy scale. International Journal of Applied Earth Observation and Geoinformation. 21: 103-112. Hussain N. 2005. Strategic plan for combating water and soil salinity in Sultanate of Oman for 2005-2015. Ministry of Agriculture and Fisheries, Oman. 117. Kwarteng AY, Dorvlo AS, Vijaya Kumar GT. 2009. Anal- ysis of a 27-year rainfall data (1977–2003) in the Sul- tanate of Oman. International Journal of Climatolo- gy. 29(4): 605-617. Lawley V, Lewis M, Clarke K, Ostendorf B. 2016. Site- based and remote sensing methods for monitoring indicators of vegetation condition: An Australian re- view. Ecological Indicators. 60: 1273-1283. Louhaichi M, Borman MM, Johnson DE. 2001. Spatial- ly located platform and aerial photography for doc- umentation of grazing impacts on wheat. Geocarto International. 16(1): 65-70. M.A.F. 2013. Oman agricultural census. Muscat, Oman: Ministry of Agriculture and Fisheries pp. Macfarlane C, Ogden GN. 2012. Automated estimation of foliage cover in forest understorey from digital na- dir images. Methods in Ecology and Evolution. 3(2): 405-415. Mishra DR. 2014. Coastal remote sensing. GIScience & Remote Sensing. 51(2): 115-119. Mulder VL, De Bruin S, Schaepman ME, Mayr TR. 2011. The use of remote sensing in soil and terrain mapping — a review. Geoderma. 162(1–2): 1-19. Rishi P, Mudaliar R. 2014. Climate stress, behavioral ad- aptation and subjective well being in coastal cities of India. Am J Appl Psychol. 2(1): 13-21. Tran PH, Nguyen AK, Liou Y-A, Hoang PP, Thanh H. 2018. Estimation of salinity intrusion by using landsat 8 oli data in the mekong delta, Vietnam.