Deltas on the move; Making deltas cope with the effects of climate c

KvR report 001/2006Figure 8:population flooded in 2100 (Nx1000/y)40302010Population at risk and land lost in 2100 as a function of two scenariosaccording to DIVA simulations.Stocks at riskThe DIVA indicators of coastal population density and economic activity value (asGDP) show substantial variability among the deltas covered (Figure 7). Lifeexpectancy appears to be extrapolated straight from GDP in DIVA, since theexplained variance of our subset was 100%. Dike height and tourist arrivals (in 1995)show clear co-variability with GDP (and hence life expectancy). Whereas estimatedland loss in 2100 correlated well to land loss in 2000, the estimated populationflooded and hence people estimated to suffer serious losses did not covary wellacross time: overall, less people were estimated to be flooded in 2100, probablybecause DIVA assumes that economically feasible measures will be taken or landwill be evacuated when submerged in sea. The overall pattern when comparing thetwo scenarios is that under A2 sea level rise will be more serious and hence allconsequences as well. Overall, it can be concluded that the following socioeconomicvariables from DIVA offer useful variability among deltas: coastalpopulation, GDP, dike height and possibly tourist arrivals. The distinction in futuredevelopment according to scenarios also makes a sensible difference.Figure 9:00 40 80 120 16040000population flooded in 2000 (Nx1000/y)A2B1direct land loss in 2100 (km 2 /y)00 0.2 0.4 0.6DIVA indicators that can be used to assess the potential for softengineering measures1.61.20.80.415000A2B1direct land loss in 2000 (erosion, km 2 /y)<strong>Deltas</strong> on the movePotential for soft engineeringSeveral possibly useful indicators have been sorted to allow a detailed check-up (seebelow). Obvious variables were sediment supply and accommodation space(SEDSUP and ASPACE), but only accommodation space was found to differentiatebetween deltas (Figure 5). Other variable should be the area of low coastal plains(for example < 1 m above sea level), and the slope across the coastline. Both displaysubstantial variation among deltas (Figure 9). Steep coastal slopes generally haveless coastal plains, but the scatter remains substantial. The highest number ofpeople potentially flooded occurred in Bangladesh (Figure 9 right panel, the outlier)in a comparatively small coastal plain.DIVA- indicators combinedA principal components analysis of all DIVA indicators combined for these 34 deltassuggested that three major axes together explained a significant part (52%) of thetotal variation. The first axis correlated to storm surge indicators, and explained 21%.The second axis (additional 19%) was related to lowland areas in the coastal plaincharacteristics, and the third axis related to the population that was potentiallyflooded (additional 12%). Further components added little to the explained varianceand were not strongly correlated to any of the underlying individual variables.Notably, the population that would actually be flooded in 2100 under scenario A2correlated strongly to PCA-axis 1. Thus, physical (axes 1, 2) and socio-economicindicators (axis 3) do span across different dimensions in the DIVA database. Thisjustifies a separate inclusion of these types of indicators and supports our initialdistinction in three categories of indicators.Single linear correlations among the many variables used here support the overallpattern of the PCA: several clusters of variables exist. Table 2 only reports the mostoutstanding and significant (p>0.05) correlations. Most projections for scenarios A2and B1 also correlate well with the values estimated for 2000. This confirms thepattern of Fig 8. Lowland areas available in deltaic coastal segments correlate well,so when there is a large area below 1 m AMSL, then the same holds for 2, 3 and 4 melevations. Since scenario-based indicators of stock at risk were found to covaryclosely, we propose to use the estimates for 2000, and maybe use the B1 and A2scenario outcomes as an indicator of possible band with of future developments. It isnoteworthy, that indeed the B1-scenario, geared towards a sustainability-orientedworld, would lead to lower mortality estimates and lower sea level rise.coastal plain (area lower than 1 m ASL)30000200001000000 0.2 0.4 0.6 0.8 1people potentially flooded(x1000, in 2000, prescenario run)10000500001 10 100 1000 10000 1000003 Combining WDD and DIVA indicatorsWDD gives an accurate physical description of deltas and the processes that arebuilding and continuously changing their form. DIVA incorporates and produces amuch larger set of variables. Where the two datasets provide similar variables (meanannual discharge, tidal regime), these correlate well, which is a comfortingobservation. We chose to keep such pairs of variables, as well as most others thatmight be useful for potential future users. Therefore the indicator database is rich invariables.coastal slope (degrees perpendicular oncoastline)coastal plain (area below 1 m ASL)9293