Z. Tessler, CCNY

Global Environmental Change and Delta Risk

Zachary TesslerCharles VörösmartyCity College of New York

Z. Tessler, CCNY

Relative Sea Level Rise is a major concern in deltas

Approx. Global Mean

Newest estimates put rate of global mean SLR at 2.6-2.9+-0.4 mm/yr over satellite altimetry era (Watson et al. 2015)

Z. Tessler, CCNY

But, consequences of RSLR are not equal across deltas

Coastal development:

vs.

Z. Tessler, CCNY

But, consequences of RSLR are not equal across deltas

vs.

Coastal development:

Coastal defense:

vs.

Ganges Delta (photo: Irina Overeem)

Z. Tessler, CCNY

But, consequences of RSLR are not equal across deltas

vs.

Coastal development:

Coastal defense:

vs.

Ganges Delta (photo: Irina Overeem)

Different climo-meterologic conditions:

Z. Tessler, CCNY

But, consequences of RSLR are not equal across deltas

vs.

Coastal development:

Coastal defense:

vs.

Ganges Delta (photo: Irina Overeem)

Different climo-meterologic conditions:

RSLR will impact the communities living

on these deltas very differently

Z. Tessler, CCNY

Goal: Estimate impact of RSLR on delta risk, globally

Z. Tessler, CCNY

Delta Flood Risk Model

● Model risk as expected loss

● Loss from a single hazardous event, h0

● E is exposure, the number of people exposed to hazardous conditions by h0

● V is vulnerability, is the average harm or loss endured by those exposed

L=Eh0∗V h0

Z. Tessler, CCNY

Delta Flood Risk Model

● Total risk or expected loss, R, is the sum of losses over all possible hazardous events, weighted by probability H(h):

R=∑h

H (h)E (h)V (h)

Z. Tessler, CCNY

Delta Flood Risk Model

● Simplify range of hazards down to single representative hazard class:

● How can we estimate H, E, and V for each delta?● How does this relate to RSLR?

H EV

Z. Tessler, CCNY

Estimating H, E, V

● Direct estimation:– From historical event database, estimate exposed population

and total loss with respect to event intensity

– But, for many deltas, historical data is inadequate

● Indirect estimation:– Identify variables that are both indicative of risk factors, and

also measured or estimated at the global scale

– Sample from global datasets to obtain consistent estimates for each delta

– Normalize and weight indicators

Z. Tessler, CCNY

Delta Extent Maps

Indus

KrishnaGodavari

BrahmaniMahanadi

Ganges

Irrawaddy

Chao Phraya

Mekong

HongPearl

Yangtze

Yellow

Mahakam

Z. Tessler, CCNY

Delta Extent Maps

Population: GRUMPv1(based on administrative data and night-light remote sensing)

These maps will soon be available, website is under construction now

Z. Tessler, CCNY

Hazard Frequency and Intensity (H)

● Indicators:– Tropical cyclone

intensity and frequency

– M2 tidal amplitude

– River discharge 30yr return value (standardized)

– Wave energy 30yr return value (standardized)

Z. Tessler, CCNY

Hazard Frequency and Intensity (H)

● Indicators:– Tropical cyclone

intensity and frequency

– M2 tidal amplitude

– River discharge 30yr return value (standardized)

– Wave energy 30yr return value (standardized)

Z. Tessler, CCNY

Vulnerability (V)

● Many variables have been used in local/regional studies to estimate social vulnerability

● Here, restricting ourselves to datasets available at the global scale– Belmont GDVI efforts to develop a more comprehensive index will

be very helpful here!!

● Vulnerability of a delta varies at the household scale (strength/quality of housing) and the delta scale (coastal infrastructure)

● Indicators:– Per capita GDP (household vulnerability)

– Aggregate GDP (Capacity for infrastructure investment)

– Government effectiveness index (Capacity/will for preparedness)

Z. Tessler, CCNY

Exposure (E)

● Challenges estimating population exposure:– (Topography + Population Distribution) is not

enough due to coastal protections, data resolution

While low-lying areas in the Mekong and Irrawaddy are flooded, <0m elevation land in the Pearl is protected by coastal and channel barriers (From Syvitski 2009)

Z. Tessler, CCNY

Exposure (E)

● ΔE is a more tractable measure than E● RSLR results in lower elevation, increased

population exposed (for a given hazard)● Estimate RSLR from available indicators● Obtain estimate for changing flood risk due to

RSLR

ΔR=H (ΔE)V

Z. Tessler, CCNY

RSLR Indicators

● Delta:– Population Density

– Wetland Disconnectivity

– Impervious Surface Area

– Groundwater Depletion

– Hydrocarbon Extraction

● Upstream Basin:– Population Density

– Wetland Disconnectivity

– Impervious Surface Area

● Offshore:– Local sea level rise trend

Estimates from Ericson (2006) and Syvitski (2009)

Z. Tessler, CCNY

Change in Exposure (ΔE)

Mek

ong

Am

azon

Gan

ges

Z. Tessler, CCNY

Hazard Frequency and Intensity (H)

Mek

ong

Am

azon

Gan

ges

Z. Tessler, CCNY

Investment Capacity - inverse of Vulnerability (V)

Mek

ong

Am

azon

Gan

ges

High VulnerabilityLow

Vulnerability

Z. Tessler, CCNY

ΔRisk SpaceΔ R=H (Δ E)V

Z. Tessler, CCNY

ΔRisk SpaceΔ R=H (Δ E)V

Z. Tessler, CCNY

Low vulnerability is very important!Δ R=H (Δ E)V

Z. Tessler, CCNY

A thought experiment...

● How might vulnerability change in an energy-constrained future?– Increased energy prices

– Require stronger infrastructure to maintain constant level of protection due to future SLR

– Overall, more expensive to reduce vulnerability

● Re-weight Vulnerability Index to reduce the relative importance of GDP– (more difficult to “buy your way out of trouble”)

Z. Tessler, CCNY

Future Changes in Vuln, and Exposure Both future

vulnerability and exposure increases

Z. Tessler, CCNY

ΔRisk

Wealthy deltas currently invest in protective infrastructure, reducing risk from increasing RSLR despite geophysical setting

If this is not an option in the future, risk “rebounds” and approaches the risk state with no investment.

Seen most dramatically in wealthy deltas in hazardous locations – Mississippi, Rhine, Han, Yangzte...

Z. Tessler, CCNY

ΔRisk

Wealthy deltas currently invest in protective infrastructure, reducing risk from increasing RSLR despite geophysical setting

If this is not an option in the future, risk “rebounds” and approaches the risk state with no investment.

Seen most dramatically in wealthy deltas in hazardous locations – Mississippi, Rhine, Han, Yangzte...

Z. Tessler, CCNY

ΔRisk

Wealthy deltas currently invest in protective infrastructure, reducing risk from increasing RSLR despite geophysical setting

If this is not an option in the future, risk “rebounds” and approaches the risk state with no investment.

Seen most dramatically in wealthy deltas in hazardous locations – Mississippi, Rhine, Han, Yangzte...

Z. Tessler, CCNY

ΔRisk

ΔR

Δ RF−Δ R

Δ R

Z. Tessler, CCNY

Delta Risk Profiling Conclusions

● Krishna, Ganges, Brahmani, Godavari deltas are most sensitive to increased RSLR (per capita risk basis)

● Several wealthy deltas (Mississippi, Rhine) have low ΔR despite high hazard and high RSLR estimates due to low vulnerability

● Future changes to vulnerability due to changing economics of coastal protection will have outsized effects on these systems. These particular systems are highly sensitive to changing vulnerability.

Z. Tessler, CCNY

Delta Characterization by Environmental Stress Patterns

● Looking across deltas, are some patterns of environmental stress more common?

● What are the implications for sustainable management?

Z. Tessler, CCNY

Indicator Correlations

Indicators projected into first 2 PCA components

Z. Tessler, CCNY

Indicator Correlations

Indicators projected into first 2 PCA components

Z. Tessler, CCNY

K-means Clustering

Z. Tessler, CCNY

K-means Clustering

Blue: Mean indicator values within clusterPink: Min-Max range of indicator values

Z. Tessler, CCNY

Sensitivity to Clustering MethodColors indicate Kmeans clustering with k=8Line width: How often two deltas are clustered together in all clustering

methods (KM-6, KM-7, KM-8, KM-9, Affinity Propogation)

Z. Tessler, CCNY

Delta Clustering For Improved Management

● Clusters identify systems under similar suites of environmental stress

● First take on address “deltas as snowflakes” challenge

● Provide an apples-to-apples context for comparing delta change

● Opportunities for sharing of effective management strategies between other systems