An Integrated Approach for Saltwater Intrusion Monitoring based on Remote Sensing combined with Multivariable Analysis: A Case Study of Coastal Zone in Southern Vietnam45229

An Integrated Approach for Saltwater Intrusion Monitoring based on Remote Sensing combined with Multivariable Analysis: A Case Study of Coastal Zone in Southern Vietnam Quoc Huy Nguyen (1), Tien Yin Chou (2), Mei Ling Yeh (2), Thanh Van Hoang (2), Xuan Linh Nguyen (1), Huyen Ai Tong (3), Quang Thanh Bui (4) (1) PhD program in Civil and Hydraulic Engineering, Feng Chia University, Taichung, Taiwan R.O.C GIS Research Center, Feng Chia University, Taichung, Taiwan R.O.C (3) Space Technology Institute, Hanoi, Vietnam (4) Faculty of Geography, Hanoi University of Science, Hanoi, Vietnam * Correspondence: st_huy@gis.tw (2) Abstract: Saltwater intrusion is a basic concern in many parts of Vietnam relative to long-term dependable water supplies It affects many sides of human life and the ecosystem Remote sensing is a useful tool for saltwater intrusion monitoring In this study, we proposed an integrated approach to estimate EC (Electrical Conductivity) value from multi-temporal optical remote sensing data for monitoring saltwater intrusion of coastal zone in southern Vietnam Multiple variable analysis was used to discover the relation between EC and different indices groups which were extracted from LANDSAT satellite images, including: original bands group, PCA (principle component analysis) group, brightness group, vegetation group, salinity group, ratio group and combined group All results were validated by field survey data This research indicated that group of combined indices from LANDSAT (EC6) had the highest correlation to EC index with R2 = 0.77 and could be used for multi-temporal saltwater intrusion monitoring A set of maps from 2005 to 2019 were established for Ben Tre province where is one of coastal zones in southern Vietnam to support policy manager to make decision for reducing damage from saltwater intrusion Keywords: saltwater intrusion; remote sensing; multiple variable analysis; Mekong Delta; Vietnam Introduction Saltwater intrusion is an essential problem concern for humans and the ecosystem It is a factor that changes the properties of the soil, which adversely affects the development of crops, causing damage to agricultural production Besides, it also causes human and animal health impacts and influences industrial production activities using natural water sources (Chhabra 2017) There are many definitions of saltwater intrusion In general, saltwater intrusion is a phenomenon of saltwater intruding deeply into the interior of areas when sea level rises at high tide It is the process when saltwater seeps into coastal groundwater systems and mixes with freshwater Salty land is an area where contains soluble salts at concentrations higher than usual and causes adverse effects on crops Saltwater intrusion is a common soil degradation It’s a process of accumulation of salts and dissolved alkali metals on the soil surface and upper soil layers This process usually starts from the lower soil layers and then slowly spreads to the surface In the past, saltwater intrusion monitoring was usually based on field measurement of salinity indicators but it was expensive that need to find other methods Remote sensing technology is one of the tools to solve this problem There has been a lot of research on developing correlation models to observe saltwater intrusion through extracted indices from satellite images (An et al 2016) used NIR and SWIR index data from LANDSAT7 and LANDSAT8 images to estimate Soil Salinity Index (SSI) Meanwhile, (Scudiero et al 2015) found a correlation between Canopy Response Salinity Index (CRSI) from LANDSAT7-TM to monitor saltwater intrusion And many other studies have demonstrated benefits in combining indicators extracted from remote sensing images to monitor multi-temporal saltwater intrusion (Wu et al 2020; Elhag et al 2017; Fan et al 2015; Liu et al 2016; El Harti et al 2016; Nawar et al 2015; Peng et al 2019) However, the results of this researches are not applicable to all areas because of different local conditions in each region In this study, we proposed an integrated approach for saltwater intrusion monitoring in Mekong Delta Vietnam base on using multiple variable analysis to discover relation between EC and indices groups which are extracted from LANDSAT satellite images All of the results will be verified by collected sample data in our research area Methodology 2.1 Study area Ben Tre is selected for our study (Figure 1) Its area is near 2360 km2 and formed by islands (An Hoa, Bao, Minh) where were deposited by alluvial from branches of Mekong river (Tien, Ba Lai, Ham Luong, Co Chien) The location of Ben Tre ranges from 9°48' to 10°20' latitude and from 105°57' to 106°48' longitude and it’s in the coastal zone in southern Vietnam This region has natural factors which are directly affected the coastal zone Especially when water level in the river is low, saltwater can be pushed back into the river and canal system to create a salinization area with different concentrations because of flows is not strong enough to prevent salty water from sea with high tide Figure The location of Ben Tre province in the Mekong Delta Vietnam 2.2 Materials LANDSAT5-TM and LANDSAT8 satellite images used for this study because of advantages in time series analysis for regional with an average spatial and temporal resolution, the medium of spectral bands and free also Table Description of LANDSAT5-TM and LANDSAT8 Band Wave Length (μm) LANDSAT8 LANDSAT5TM Resolution (10m) LANDSAT8 LANDSAT5TM Ultra Blue (coastal/aerosol) 0.43 - 0.45 30 Blue 0.45 - 0.51 0.45 - 0.52 30 30 Green 0.53 - 0.59 0.52 - 0.60 30 30 Red 0.63 - 0.67 0.63 - 0.69 30 30 Near Infrared (NIR) 0.85 - 0.87 0.76 - 0.90 30 30 Shortwave Infrared (SWIR) 1.56 - 1.65 1.55 - 1.75 30 30 Shortwave Infrared (SWIR) 2.10 - 2.29 2.08 - 2.35 30 30 Panchromatic 0.50 - 0.67 15 Cirrus 1.36 - 1.38 30 10 Thermal Infrared (TIRS) 10.6 - 11.2 11 Thermal Infrared (TIRS) 11.5 - 12.5 10.40-12.50 100 120 100 A field salinity survey was conducted by taking soil samples in the study site to verify results One sample is designed in a 90m x 90m cell with positions (Figure 2) Figure Survey locations and Design of sample locations EM31-MK2 device was used to measure EC data with 556 samples over different land uses It will be standardized by reference to soil samples analysis Table Mean of EC over Land Uses Land Use Number of Samples EC Mean Salty Level Salty Land 18 19,3739 High Mangrove 17,8200 High Sand Coastal 67 12,7965 Medium Aquaculture 48 12,0464 Medium Crop Land (1c) 49 9,0753 Medium Crop Land (2c) 103 8,1309 Medium Crop Land (3c) 96 7,5258 Low Vegetation 77 3,8513 Non-Salty Plant 91 3,4208 Non-Salty 2.3 Data processing To discover the relation between indices group of LANDSAT and EC, we divided them into physical indices groups including: original index, principle component analysis index, brightness index, vegetation, salty index and ratio index See in Table for more details Table Indices Group which are extracted from LANDSAT Indices Group Formula Original Index B1, B2, B3, B4, B5, B6 PCA Index PCA1, PCA2, PCA3, PCA4, PCA5, PCA6 Brightness Index BI = sqrt (G2 + NIR2) (Yoo et al 2018) BI2 = sqrt (R + NIR ) (Gadal et al 2019) INT = (G + R) / (Rahmati et al 2017) NDVI = (NIR - R) / (NIR + R) (Silvia et al 2019) Vegetation Index Salitiny Index Ratio Index Reference SAVI = (1 + L) * (NIR - R) / (NIR + R + L) (Ren et al 2018) EVI = G * (NIR - R) / (NIR +C1R - C2B + L) (Li et al 2019) GDVI = (NIRn - Rn) / (NIRn + Rn) (Avola et al 2019) SI1 = sqrt (G * R) (Touhami et al 2019) SI2 = sqrt (G2 + R2 + NIR2) (Alexakis et al 2018) SI3 = sqrt (G2 + R2) (Abdullah et al 2019) SI4 = sqrt (((NIR * R) - (G * B)) / ((NIR * R) - (G * B))) (Samiee et al 2018) SI5 = (R - NIR) / (R + NIR) (Samiee et al 2018) SI6 = sqrt (R * NIR) (Samiee et al 2018) B/G B/NIR B/R B/SWIR1 B/SWIR2 G/R G/NIR G/SWIR1 G/SWIR2 R/NIR R/SWIR1 R/SWIR2 NIR/SWIR1 NIR/SWIR2 SWIR1/SWIR2 After calculated, these indicators will be extracted base on measured EC locations to establish in regression analysis model where EC is dependent variable and others are independent variables Criteria(s) for regression assessment include Sig coefficient (Pvalue), Pearson correlation coefficient (R), Root Mean Square Error (R2) and Adjusted Root Mean Square Error (R2adj) Variables which have Sig > 0.05 and R ≈ will be removed If R2 and R2adj are in (0, 1) range and values moved to near that means model fits data set Typically, model can be applied when R2 > 0.5 (Abdullah et al 2019) Multiple variable regression model applied for separate indices groups to identify which group has the highest correlation with EC One combined group of all indices will also be used in this progress to assessment 70% samples data will be used for model input and 30% for validation Figure Logical framework of study Results 3.1 Relation between EC and Indices Group from LANDSAT The results demonstrate that a combined group has the highest correlation with EC (R = 0.783 and R2 = 0.6126) and will be used to estimate EC and establish salinization map for saltwater intrusion monitoring Lowest correlation with EC is PCA group (R = 0.675 and R2 = 0.4839) The brightness indices group removed because of having a reserve correlation with EC Table Regression models from indices group Indices Group Estimation Model R R2 R2adj F-Static DF Original Index L_EC1 = 11,609 + 40,220B3 82,156B5 + 87,971B6 0.750 0.5545 0.559 137.955 3:388 PCA Index L_EC2 = 13,223 - 30,522 * PCA1 + 7,836 * PCA2 + 50,068 * PCA3 0.675 0.4839 0.451 97.042 3:388 Vegetation Index L_EC3 = 18,822 - 13,088 * GDVI 0.696 0.4839 0.483 307.650 1:388 Salty Index L_EC4 = 16,816 + 8,475 * SI5 65,088 * SI6 + 533,248 * SI1 314,669 * SI3 + 0,594 * SI4 0.741 0.5367 0.544 76.716 5:388 Ratio Index L_EC5 = 17,970 + 22,547 * (G/NIR) - 2,638 * (G/R) 10,087 * (SWIR1/SWIR2) + 2,304 * (NIR/SWIR2) - 3,879 * (B/G) 0.767 0.588 0.583 95.958 5:388 All Combined Index L_EC6 = -6,489 + 20,492 * (G/NIR) + 0,383 * T + 28,615 * B6 + 6,247 * (NIR/SWIR1) 20,053 * SI2 - 3,505 * (SWIR1/SWIR2) 0,783 0,6126 0,607 75,305 6:388 Figure Correlation of surveyed EC and Indices group 3.2 Validation 30% samples data are used for validating EC estimation model from LANDSAT The correlation between the estimated EC and validated EC is R2 = 0.77 It’s not really high but can be accepted This model is applied in EC estimation from LANDSAT from 2005 to 2019 with classification levels of EC values by Ministry of Agriculture and Rural Development Table Salty Level Classification Level Description EC (dS/m) Non-Salty 15 Figure Correlation of estimated EC and surveyed EC 3.3 Saltwater intrusion Mapping After validation, saltwater intrusion maps were established by using LANDSAT and L_EC6 model to estimate EC value In particular, salty regions are mainly distributed in coastal districts over mangrove and aquaculture areas including Binh Dai, Ba Tri and Thanh Phu Figure Map of saltwater intrusion in Ben Tre in 2005 Figure Map of saltwater intrusion in Ben Tre in 2010 Figure Map of saltwater intrusion in Ben Tre in 2015 Figure Map of saltwater intrusion in Ben Tre in 2019 Conclusions and discussion The general trend of saltwater intrusion in Ben Tre province showed an increase of salty area and a decrease of the non-salty area which is mainly concentrated in coastal districts The conversion of land use from rice to aquaculture is one of the leading causes of saltwater intrusion in soils Within 13 years from 2005 - 2019, a total of non-salty areas decreased by nearly 15824.84 while salty area increased where the lowest was 8169.98 and the highest was 10314.19 During 2010 - 2015 periods, high salty area expanded fastest with 6861.46 (ha) due to the inefficient production of rice and aquaculture which was converted to industrial shrimp farming Another reason is the alluvial land between large rivers such as Ham Luong, Co Chien and Tien has also been changed from plant to aquaculture The leading cause of saltwater intrusion in Ben Tre province is an extension of aquaculture area to the land Saltwater which is taken from canals will lead to saltwater intrusion deeper from the sea Moreover, canal system is intertwined while irrigation system is not closed and makes some places become saltwater holes in the dry season Therefore, saltwater intrusion occurs every year along the river and agriculture land in Ben Tre province Saltwater intrusion is a severe worldwide problem and directly affecting the natural environment, agriculture, and food security It is important to establish saltwater intrusion map which provides useful information about salty area and can be useful for land use planning and management Results demonstrated that indices group is very important in estimating EC from optical satellite images In this study, we reviewed remote sensing applications related to analysis and assessment of saltwater intrusion using indices group from LANDSAT which integrated with field survey data, spatial analysis and statistical methods However, they are not really suitable for coastal zones such as Ben Tre because of cloudy In the future, radar satellite images may be considered to solve this problem Acknowledgments The data in this article was supported by project VT-UD.03/16-20, entitled: “Studying, Assessing, and Zoning Soil Salinity by Using Multi-Temporal Satellite Imagery: A Case Study in Ben Tre Province”, which belongs to the national program on Space Science and Technology (2016 - 2020) References Chhabra, R (2017) Chapter 1: Origin and Distribution of Salt-affected Soils In Soil salinity and water quality, 2nd Edition Routledge, Taylor & Francis Group, 1-25 An, D., Zhao, G., Chang, C., Wang, Z., Li, P., Zhang, T., & Jia, J (2016) Hyperspectral field estimation and remote-sensing inversion of salt content in coastal saline soils of the Yellow River Delta International Journal of Remote Sensing, 37(2), 455-470 Scudiero, E., Skaggs, T H., & Corwin, D L (2015) Regional-scale soil salinity assessment using Landsat ETM+ canopy reflectance Remote Sensing of Environment, 169, 335-343 Wu, W., Muhaimeed, A S., Al-Shafie, W M., & Al-Quraishi, A M F (2020) Using radar and optical data for soil salinity modeling and mapping in Central Iraq In Environmental remote sensing and GIS in Iraq (pp 19-40) Springer, Cham Elhag, M., & Bahrawi, J A (2017) Soil salinity mapping and hydrological drought indices assessment in arid environments based on remote sensing techniques Geoscientific Instrumentation, Methods and Data Systems, 6(1), 149-158 Fan, X., Liu, Y., Tao, J., & Weng, Y (2015) Soil salinity retrieval from advanced multispectral sensor with partial least square regression Remote Sensing, 7(1), 488-511 Liu, G., Li, J., Zhang, X., Wang, X., Lv, Z., Yang, J., & Yu, S (2016) GIS-mapping spatial distribution of soil salinity for Eco-restoring the Yellow River Delta in combination with Electromagnetic Induction Ecological engineering, 94, 306-314 El Harti, A., Lhissou, R., Chokmani, K., Ouzemou, J E., Hassouna, M., Bachaoui, E M., & El Ghmari, A (2016) Spatiotemporal monitoring of soil salinization in irrigated Tadla Plain (Morocco) using satellite spectral indices International Journal of Applied Earth Observation and Geoinformation, 50, 64-73 Nawar, S., Buddenbaum, H., & Hill, J (2015) Estimation of soil salinity using three quantitative methods based on visible and near-infrared reflectance spectroscopy: a case study from Egypt Arabian Journal of Geosciences, 8(7), 5127-5140 Peng, J., Biswas, A., Jiang, Q., Zhao, R., Hu, J., Hu, B., & Shi, Z (2019) Estimating soil salinity from remote sensing and terrain data in southern Xinjiang Province, China Geoderma, 337, 1309-1319 Yoo, H., & Chung, K (2018) Heart rate variability based stress index service model using bio-sensor Cluster Computing, 21(1), 1139-1149 Gadal, S., & Ouerghemmi, W (2019) Multi-Level Morphometric Characterization of Built-up Areas and Change Detection in Siberian Sub-Arctic Urban Area: Yakutsk ISPRS International Journal of Geo-Information, 8(3), 129 Rahmati, M., & Hamzehpour, N (2017) Quantitative remote sensing of soil electrical conductivity using ETM+ and ground measured data International journal of remote sensing, 38(1), 123-140 Silvia, L., Alexander, T., Anna, K., & Polina, K (2019) Assessment of carbon dynamics in Ecuadorian forests through the Mathematical Spatial Model of Global Carbon Cycle and the Normalized Differential Vegetation Index (NDVI) In E3S Web of Conferences (Vol 96, p 02002) EDP Sciences Ren, H., Zhou, G., & Zhang, F (2018) Using negative soil adjustment factor in soiladjusted vegetation index (SAVI) for aboveground living biomass estimation in arid grasslands Remote sensing of environment, 209, 439-445 Li, W., El-Askary, H M., Qurban, M., Allali, M., & Manikandan, K P (2019) On the Drying Trends Over the MENA Countries Using Harmonic Analysis of the Enhanced Vegetation Index In Advances in Remote Sensing and Geo Informatics Applications (pp 243-245) Springer, Cham Avola, G., Di Gennaro, S F., Cantini, C., Riggi, E., Muratore, F., Tornambè, C., & Matese, A (2019) Remotely Sensed Vegetation Indices to Discriminate Field-Grown Olive Cultivars Remote Sensing, 11(10), 1242 Touhami, M K., Bouraoui, S., & Berguig, M C (2019) Contribution of Satellite Imagery to Study Salinization Effect of Agricultural Areas at Northern Eastern Oasis Algerian Region In Advances in Remote Sensing and Geo Informatics Applications (pp 235-238) Springer, Cham Alexakis, D D., Daliakopoulos, I N., Panagea, I S., & Tsanis, I K (2018) Assessing soil salinity using WorldView-2 multispectral images in Timpaki, Crete, Greece Geocarto international, 33(4), 321-338 Abdullah, A Y M., Biswas, R K., Chowdhury, A I., & Billah, S M (2019) Modeling soil salinity using direct and indirect measurement techniques: A comparative analysis Environmental Development, 29, 67-80 Samiee, M., Ghazavi, R., Pakparvar, M., & Vali, A A (2018) Mapping spatial variability of soil salinity in a coastal area located in an arid environment using geostatistical and correlation methods based on the satellite data Desert, 23(2), 233-242  ... of advantages in time series analysis for regional with an average spatial and temporal resolution, the medium of spectral bands and free also Table Description of LANDSAT5-TM and LANDSAT8 Band... and aquaculture areas including Binh Dai, Ba Tri and Thanh Phu Figure Map of saltwater intrusion in Ben Tre in 2005 Figure Map of saltwater intrusion in Ben Tre in 2010 Figure Map of saltwater intrusion. .. Pakparvar, M., & Vali, A A (2018) Mapping spatial variability of soil salinity in a coastal area located in an arid environment using geostatistical and correlation methods based on the satellite data