Monitoring Shoreline Change In Playa Jayuya, Fajardo Using .

Monitoring Shoreline Change in Playa Jayuya, FajardoUsing Remote Sensing TechniquesGWENDELYN MONGE ACEVEDODepartment of Geology, University of Puerto Rico, Mayagüez CampusP.O. Box 9017 Mayagüez, Puerto Rico, 00681-9017ABSTRACT- Coasts are dynamic environments were a high percent of the world’s populationare concentrated. In this study shoreline change is analyzed by distinct remotely sensed methodsfrom ENVI and QGIS. A combination of aerial photography from 1930, 1950, 1987 and asatellite image of Landsat ETM of 2000 were selected for the shoreline change analysis ofplaya Jayuya. Edge detection was used to evaluate changes in playa Jayuya. Results showed thaterosion increased rapidly since 1950 to 2000. Better methods are required for accuracyestimations of shoreline changes.KEYWORDS- coastal geomorphology, remote s constantly changing. Coastalareas are subject to a variety of phenomena,such as sea level variations, storm surges,wave energy, tidal inundation, tectonics andland subsidence, sediment budget changes,human activities that continually modify andplay fundamental roles in coastaldevelopment and exposed to erosion (Aielloet. al. 2013).A recent study by Jackson et al.(2012), stated that to be able to gain a betterunderstanding of the processes that impactcoastal resources and determine theirvulnerability there needs to be a focus onmapping and quantifying the movement ofthe shoreline. In Puerto Rico almost 61% ofthe population lives in coastal areas,however most of shoreline researches in theisland are either from the north or westcoast. Studying this site is important notonly because it is constantly changing due tonatural process, but because it has anarcheological site that has valuable historicalinformation of Puerto Rico. Pre-historicarcheological (600 B.C,) materials such asceramic vessels and bones are found in thesand of playa Jayuya. Since coastal erosionis increasing, the archeological material isbeing lost or is trapped in the beach rock.Even though coastal erosion is a naturalprocess, it is a risk factor in this particulararea, because of the valuable historicalinformation it contains.To evaluate shoreline changedifferent techniques are used which includeshoreline tracing of aerial photograph andother remote sensing analysis to identifyshoreline change on island and atoll island(Ford, 2011). Maiti (2009), used satelliteimages to measure and predict shorelinechange, these technique are challenging dueto different time ranges or meteorologicalconditions that cause problems during dataacquisition. On the other hand themorphological characteristics provided bysatellite images facilitate the identificationof the predominant process causing theevolution of the beach. Shoreline changesstudies use several proxies in order todetermine changes in the coastline, such as

MONITORING SHORELINE CHANGElow or high water level, dunes, structures orvegetation.Ford (2011) used the edge of thevegetation as proxy because it is relativelyeasy to identify in imagery. Shorelineposition can be study using geographicinformation techniques such as DigitalShoreline Analysis System (AppeaningAddo, 2008; Ford, 2011), remotely sensingdata analysis and by creating models thatcan estimate coastal erosion (Corbella,2012). The main purpose of this project is toidentify and measured temporal changes,such as coastal erosion or accretion from1930 to 2000 using different techniques onEnvironment Visualizing Images (ENVI)and Quantum geographic informationsystem (QGIS).MATERIAL AND METHODSStudy AreaPlaya Jayuya is located on theeastern coast of Puerto Rico in themunicipality of Fajardo and is a naturalreserve under the administration of thePuerto Rico Conservation Trust. The studyarea is located in the northeast igneousprovince and is mostly composed of beach,swamp deposits and Hato Puerco formation(fig 1).Image ProcessingShoreline changes were analyzedusing a combination of aerial photographyand satellite imagery. Aerial photos from1930, 1950 and 1987 were used in thisstudy. Satellite image from 2000 wereacquired from the EarthExplorer website.(Table 1)Aerialphotographywasgeoreferenced with QGIS using controlpoints extracted from the satellite image. Atleast five control points were used for theprocedure. A subset of aerial photographyand image was performed using the resizetool from ENVI. Images were geometric andradiometric corrected. The Landsat ETM image was sharpened from 30meters to15meters for the final analysis. Techniquesfrom ENVI and QGIS were selected toidentify shoreline change to see which hadthe most accurate and precise results.Table1. Characteristics of aerial photographyand satellite image used for the studyDateImage type2000Landsat 7 ETM 2/25/1987Color aerial11/25/1950B/W aerial1/16/1931B/W aerialShorelinechange:classificationFigure1. Map of the geology of PlayaJayuya (Fajardo, Puerto Rico)2NeuralNetworkNeural network is a supervisedclassification from ENVI. To determineshoreline change the ROI tool was used to

MONITORING SHORELINE CHANGEcreate two different classes, land and sea.Neural Net was used for 1930 and 1987aerial photography and 2000 satellite image.For each of the images 10 training iteration3was selected. To determine rate of erosion ordeposition for each of the images with theRegion of Interest (ROI) tool a polygon wasbuilt to determine the area in square meters.abcdFigure 2. Subset of Playa Jayuya (a)1930 (b)1950 (c)1987 (d)2000-Satellite image

MONITORING SHORELINE CHANGE4baFigure 3. (a) 2000 Landsat ETM (red:sea; green:land) (b)1987 aerial color photography(blue:sea; red:land)Shoreline change: Edge ExtractionThe Edge extraction is a tool fromthe open source QGIS that is based ondetecting edges in the image. It has differenttypes of edge extraction; in this analysis wasused edge extraction-gradient. Shorelinechange was measured for a particular area inthe subset using the vegetation line to waterlines as proxy for changes detection.RESULTThe Neural Network classificationwas performed only for three images inorder to see the differences. The 1930 aerialphotography is black and white andclassification only work with multispectralimages. In the case of the 1987 colorphotography and the 2000 satellite imagehas better results because they have morethan one band. Figure 3a show the results ofthe satellite image (15 meters) after theclassification and figure 3b showed 1987aerial photography with two classes (landand sea). Figure 4 show the differencebetween the areas (m²) from 1930 to 2000.Deposition occurred between the years of1930 to 1950, while from 1950 to 2000show high rate of erosion (table 2). Highvalues from 1987 to 2000 may be because ofthe difference in spectral and spatialresolution of the satellite image.The edge extraction techniqueshowed reliable results for 1930, 1950 and1987 aerial photography (figure 5). Edgevegetation and water line was identifiedusing the original photography as guideline.For the 2000 image the results are moredifficult to analyze due to low quality andresolution. Based on this informationchanges in the coastline were identified asrate of deposition from 1930 to 1950 of 0.15meters, while from 1950 to 1987 themeasure was -0.46 showing erosion for thissection. From 1987 to 2000 higher rate oferosion were identify, in this case theanalysis was made between a satellite imageof 2000 and the 1987 aerial photography.

MONITORING SHORELINE CHANGEFigure 4. ROI data from 1930(red), 1950(green), 1987(blue) and 2000(yellow)abcdFigure 5. Edge Extraction (gradient) of aerial photography from (a) 1930 (b) 1950 (c)1987 andsatellite image of (d) 2000 Landsat ETM 5

MONITORING SHORELINE CHANGE6 15yrsFigure 6. Photo of Playa Jayuya 2015showing an estimation of were the sea willin 15 years.DISCUSSIONEven thou Neural Net classificationis the most accurate method fordifferentiation in ENVI, in this case is morechallenging because the data used for theanalysis is a combination of aerialphotography and satellite imagery. Lowresolution of the satellite image is alimitation when creating the differentclasses; it is very difficult to identify theexact position of the shoreline. In additionthis classification works with multispectralimages and but is not going to work withblack and white aerial photography.The edge extraction shows moreaccurate results for shoreline changeanalysis such as the ROI data from ENVIthat was used in order to compare differentmethod to determine changes in PlayaJayuya. Both results (table 2) showdeposition during the years of 1930 and1950, while 1950 to 1987 rate of erosionincreased. The fact that one of the objectiveswas to compare two different techniques byusing a combination of aerial photographyand satellite image could lead us to makeunreliable estimation of the changes in playaJayuya. Based on the dataFigure 7. Vegetation edge (archeologicalmaterial) of Playa Jayuyaobtained from the analysis if erosioncontinue increasing at the same rate, theshoreline in approximately 15 years willreach the vegetation edge, consequently thearcheological area will be exposed(Figure6and 7).Table 2. Shoreline change from 1920 to2000 data from Region of Interest (m²) andEdge Extraction ErosionErosionDepositionCONCLUSIONIn general, both techniques can beused for shoreline change analysis but forNeural Net satellite image with higherresolution are required to identify changes.Edge Extraction (gradient) tool from QGISis a technique that detect changes in theimage and is also a technique that helpidentify changes throughout time.The main objectives for this projectwere to explore new remote sensing

MONITORING SHORELINE CHANGEtechniques to study, evaluate, and monitorshoreline change as well as determineerosion or accretion in Playa Jayuya. In thisstudy the data obtain shows that erosion hasincrease since the 1950. If the rate of erosioncontinues to increase eventually thearcheological material will be lost. In orderto have more accuracy more years are needto see if there is a pattern and increase theaccuracy of estimations of erosion ution are required in order to run anysupervised or unsupervised classificationand in order to have better results. Exploringother tools for shoreline change analysis isrecommended before any procedure.REFERENCESAppeaning Addo K, Walkden M., Mills J. P.,2008. Detection, measurement and prediction ofshoreline recession in Accra, Ghana. Journal ofPhotogrammetry & Remote Sensing 63, 543–558Dewidar, K., Frihy, O., 2007. Pre and postbeachresponse to engineering hard structures usingLandsat timeseries at the north western part ofthe Nile delta, Egypt. Journal of CoastalConservation 11, 133142.Ford M., 2013. Shoreline changes interpretedfrom multi-temporal aerial photographs and highresolution satellite images: Wotje Atoll,Marshall Islands. Journal of Remote Sensing ofEnvironment. 130-140Jackson C. W., Alexander C. R., Bush D. M.,2012. Application of the AMBUR R package forspatiotemporal analysis of shoreline change:Jekyll Island, Georgia, USA. Journal ofComputers and Geosciences. 199207Li, X., & Damen, M. C. (2010). Coastlinechange detection with satellite remote sensing7for environmental management of the PearlRiver Estuary, China. Journal of MarineSystems, 82, S54-S61.Maiti S., Bhattacharya A. K. 2009. Shorelinechange analysis and its application to prediction:A remote sensing and statistics based approach.Journal of Marine Geology. 2029

Neural Net satellite image with higher resolution are required to identify changes. Edge Extraction (gradient) tool from QGIS is a technique that detect changes in the image and is also a technique that help identify changes throughout time. The main objectives for this project were to explore ne