Estimating future coastline changes along Holland coast, under different sea level rise scenarios, using a probabilistic approach

Abstract

Due to climate change and sea
level risethe coastal zones are getting exposed to increasing risks
  like coastalrecession, putting in risk human lives and coastal
infrastructure being worthbillions of dollars. Low lying countries like the
Netherlands are consideredmore vulnerable to the effects of sea level rise.
Large parts of the Dutchcoast have been eroding for centuries and nourishments
schemes of approximately12 million m3 have been implemented annually
in order to maintainthe coastline as it was in 1990. However, the future dune
erosion will further increasedue to the impacts of climate change and hence the
adaptation strategies shouldbe in line with the accelerated sea level rise and
the possible effects thatmay bring. The most commonly used method to assess
sealevel rise impacts on shorelines is the Bruun rule. However,Bruun rule’s
deterministic nature cannot align with the risk-based framework thatcoastal
zone management requires nowadays. This necessity initiated thedevelopment of a
process-based model, the Probabilistic Coastline Recession(PCR) model,
estimating the future coastal recessions in a probabilisticapproach. In this
research, the PCR framework wasapplied at eleven locations along the Holland
coast, in the Netherlands, underthree different SLR scenarios, the RCP4.5,
RCP8.5 and Deltascenario. The availabilityof coastal profile data (from 1965
until now) and coastline position data (from1843 till 1980) made the Holland
coast an ideal location to explore and extendthe applicability of the PCR
framework. The most relevant assumptions for thiscoast were identified and
explored. The recovery rate of the dune was a weakpoint of the PCR model and
Holland coast was an interesting area to be tested.Three approaches of
calibrating the natural recovery rate of the dunes werefollowed. In addition,
the alongshore sediment transport which was assumednegligible to the previous
case studies, in this work it was integrated intothe PCR model and pointed out
that its contribution is important to the PCR.  For the eleven selected coastal
profiles,20,000 simulations of 81 years (2020-2100) have been conducted and for
everysimulation the most landward position of the coastline in every calendar
yearhas been recorded. Hence, an empirical distribution of coastline recession
forevery future year has been constructed. The ranges of the expected retreats
in2100 (relative to 2020) for the different SLR scenarios
are:0.5 m-155 m (for RCP4.5), 6 m-194 m (for RCP8.5)
and18 m-172 m (for Deltascenario), corresponding to the
50 %exceedance probability values of the cumulative distribution function
of thecoastline retreat. The average values of the coastal retreat for 2100 are
61 m,73 m and 97 m for RCP4.5, RCP 8.5, and Deltascenario
respectively.The relevant average erosion volume by 2100 are 1664 m3/m,2005 m3/m
and 2665 m3/m. According to thefindings, in 2100 the relative
increase in volume loss along the entire theHolland coast is expected to be
95 %, 121 % and 173 %respectively for RCP4.5, 138 % for
RCP8.5 and 195 % for Deltascenario.Finally, the results were compared to
those raised from the Bruun rule method. Accordingto the findings, the majority
of the profiles showing an erosive trend in thepast (before the ‘hold-the-line’
policy) raised slightly more conservativeresults when implementing the PCR
model rather than when applying the Bruunrule method- especially under the
Deltascenario. On the other hand, theBruun rule method is more conservative
than PCR model for most of the accretiveprofiles. The PCR model can now be
explored tolocations where the longshore sediment transport is not negligible. Theapproach
followed in this study allows investigating the ability of the modelfor future
coastal retreat estimates when a construction of a hard defence structureor a
port may change abruptly the longshore sediment transport. Last, this
studyadvances the PCR framework and can be a valuable assistance in the course
offurther improving the model.