WaywardHouse
Sea Level, Storm Surge, and Coastal Morphodynamics
Sea level is not a fixed datum — it fluctuates with the tides, surges during storms, and is rising globally due to climate change. The total water level at a coastline is the sum of these components, and each component leaves a different signature on the shoreline. This essay decomposes relative sea level change, derives the storm surge momentum balance and the Bruun Rule for long-term shoreline retreat, models barrier island rollover, and builds the Coastal Vulnerability Index from measurable physical variables.
Prerequisites: linear superposition, slope geometry, dimensional analysis, index construction
On October 29, 2012, Hurricane Sandy made landfall near Atlantic City, New Jersey. Storm surge of 2.7 m above normal tide level flooded lower Manhattan, devastated barrier island communities in New Jersey, and caused over USD 65 billion in damage. The surge was not primarily a wave phenomenon — it was a long, slow rise of the entire water level driven by the atmospheric pressure of the storm and, more importantly, by the wind pushing water toward the coast.
Sandy’s surge sat on top of a tide. The tide sat on top of a slowly rising baseline sea level. The waves that destroyed beachfront structures sat on top of the surge. The total water level — the only thing a coastline actually experiences — is the sum of all these components. Understanding them separately is the prerequisite for predicting their combined effect.
1. The Question
What are the components of total water level at a coastline? How do we model storm surge from first principles? How much shoreline retreat does sea level rise produce? What controls whether a barrier island survives sea level rise or drowns? And how do we synthesise these vulnerabilities into a practical index?
2. The Conceptual Model
Total water level $\eta_{\text{total}}$ at the coast is a superposition of components operating at different timescales:
\[\eta_{\text{total}} = \eta_{\text{MSL}}(t) + \eta_{\text{tide}}(t) + \eta_{\text{surge}}(t) + \eta_{\text{setup}}(t) + \eta_{\text{infragravity}}(t)\]- MSL trend $\eta_{\text{MSL}}$: the long-term change in mean sea level (climate signal + vertical land movement), operating over decades to centuries
- Tide $\eta_{\text{tide}}$: gravitational forcing by sun and moon, periodic and predictable, amplitude 0.1–8 m depending on location
- Storm surge $\eta_{\text{surge}}$: wind and pressure-driven setup during storms, episodic, amplitude 0.1–5+ m
- Wave setup $\eta_{\text{setup}}$: radiation stress-driven mean water level rise in the surf zone (Essay AC1), amplitude 5–20% of offshore wave height
- Infragravity waves $\eta_{\text{infragravity}}$: long-period oscillations generated by breaking waves, amplitude 0.1–0.5 m, important for overwash and run-up
For coastal flooding and morphological change, the extreme total water level — the combination of a high tide, a large surge, and large waves — determines whether a threshold is crossed.
3. Building the Mathematical Model
3.1 Relative Sea Level and Its Components
Relative sea level (RSL) is the sea surface elevation relative to a fixed point on land. It differs from eustatic sea level (the absolute volume of ocean water) by the rate of vertical land motion (VLM):
\[\dot{\text{RSL}} = \dot{\text{SL}}_{\text{eustatic}} - \dot{h}_{\text{VLM}}\]where positive VLM means land rising (reducing RSL change) and negative VLM means land subsiding (amplifying RSL rise). Eustatic sea level rise is currently $\approx 3.7$ mm yr⁻¹ (2006–2018 satellite altimeter average). VLM ranges from +10 mm yr⁻¹ in Fennoscandia (post-glacial isostatic rebound) to -25 mm yr⁻¹ in parts of the Mississippi Delta (compaction + fluid withdrawal).
For the Fraser Delta in British Columbia: eustatic rise $\approx 3$ mm yr⁻¹, VLM $\approx -1$ to $-3$ mm yr⁻¹ (slow deltaic subsidence), giving RSL rise $\approx 4$–$6$ mm yr⁻¹.
3.2 Storm Surge: The Momentum Balance
Storm surge is driven by two mechanisms:
Inverse barometer effect: A 1 hPa decrease in atmospheric pressure $p_a$ raises sea surface by approximately 1 cm:
\[\eta_{\text{pressure}} = \frac{p_{a,\text{ambient}} - p_a}{\rho_f g} \approx \frac{\Delta p_a}{9810} \quad \text{[m per Pa]}\]A central pressure of 970 hPa in a storm with ambient pressure 1013 hPa gives:
\[\eta_{\text{pressure}} = \frac{4300 \text{ Pa}}{9810 \text{ N m}^{-3}} = 0.44 \text{ m}\]Wind setup: Wind exerts a shear stress $\tau_w = \rho_a C_D U_{10}^2$ on the sea surface, where $\rho_a \approx 1.25$ kg m⁻³ is air density, $C_D \approx 1.2 \times 10^{-3}$ is the wind drag coefficient, and $U_{10}$ [m s⁻¹] is wind speed at 10 m height. This stress drives water toward the coast; the steady-state momentum balance per unit width gives:
\[\frac{d\eta}{dx} = \frac{\tau_w}{\rho_f g h} = \frac{\rho_a C_D U_{10}^2}{\rho_f g h}\]The total wind setup over a fetch $F$ [m] on a shelf of uniform depth $h$:
\[\Delta\eta_{\text{wind}} = \frac{\rho_a C_D U_{10}^2 F}{\rho_f g h}\]For Hurricane Sandy parameters: $U_{10} = 45$ m s⁻¹, $F = 500$ km (continental shelf), $h = 30$ m:
\[\Delta\eta_{\text{wind}} = \frac{1.25 \times 1.2 \times 10^{-3} \times (45)^2 \times 5 \times 10^5}{1025 \times 9.81 \times 30}\] \[= \frac{1.25 \times 1.2 \times 10^{-3} \times 2025 \times 5 \times 10^5}{301,275}\] \[= \frac{1.519 \times 10^6}{3.013 \times 10^5} = 5.04 \text{ m}\]This is the order of Sandy’s observed 2.7 m surge (the discrepancy reflects the simplification of uniform depth and steady wind; a full numerical model gives better agreement). The calculation reveals that a shallow, wide continental shelf amplifies surge dramatically — exactly the geometry of the US East Coast and Gulf of Mexico.
Surge decay: After the storm passes, the surge decays exponentially as the pressure gradient drives return flow:
\[\eta_{\text{surge}}(t) = \eta_{\text{max}} \exp(-t/\tau)\]where $\tau = F/(g h)^{0.5}$ is a characteristic decay time — the time for a gravity wave to traverse the shelf. For $F = 500$ km and $h = 30$ m: $\tau = 5\times10^5 / \sqrt{9.81 \times 30} = 5\times10^5 / 17.15 = 2.9 \times 10^4$ s $\approx 8$ hr.
3.3 The Bruun Rule for Shoreline Retreat
As sea level rises, a beach adjusts its profile to maintain its equilibrium form relative to the new water level. The Bruun Rule (1962) derives the shoreline retreat $R$ [m] from a simple geometric argument.
Consider a beach profile extending from the shoreline to the closure depth $D_c$ [m] (the depth below which sediment does not move). The active profile has a horizontal extent $L^*$ [m] from shoreline to closure contour.
When sea level rises by $\Delta S$ [m], the entire profile must shift upward and landward to maintain equilibrium. The volume of sand needed to raise the offshore profile by $\Delta S$ over distance $L^*$ must come from landward erosion of the beach:
\[R \times (B + D_c) = L^* \times \Delta S\]where $B$ [m] is the berm height (elevation of the beach crest above still water level). Solving for retreat:
\[R = \frac{L^*}{B + D_c} \Delta S\]This is the Bruun Rule. The ratio $L^* / (B + D_c)$ is the inverse of the mean slope of the active profile — a gentle profile retreats more for the same sea level rise.
For a typical low-energy beach: $L^* = 500$ m, $B = 1.5$ m, $D_c = 6$ m. For $\Delta S = 0.5$ m (plausible 21st century rise):
\[R = \frac{500}{1.5 + 6} \times 0.5 = \frac{500}{7.5} \times 0.5 = 66.7 \times 0.5 = 33 \text{ m}\]A 33 m retreat for 0.5 m of sea level rise. The Bruun ratio $R/\Delta S = L^*/(B+D_c) \approx 50$–$200$ for most beaches — typically cited as “50 to 200 m of retreat per metre of sea level rise.”
Important caveat: The Bruun Rule assumes that all eroded material is redistributed offshore within the active profile — no longshore transport, no net losses offshore. It is a geometric identity, not a physical model of the erosion process. On coasts with significant longshore transport gradients, the Bruun Rule underestimates actual retreat.
3.4 Barrier Island Rollover
Barrier islands — long, narrow islands of sand separated from the mainland by a lagoon — are particularly sensitive to sea level rise because they have a finite sand budget and can respond in three ways:
- Accretion: if sediment supply and aeolian/overwash deposition keep pace with sea level rise, the island maintains elevation
- Rollover: if overwash transport carries sand from the ocean-facing beach over the crest and into the lagoon, the island migrates landward while maintaining its form
- Drowning: if sea level rise exceeds the rollover capacity, the island becomes submerged
The critical parameter is the ratio of sea level rise rate $\dot{S}$ [m yr⁻¹] to the overwash transport rate $Q_w$ [m³ m⁻¹ yr⁻¹] relative to island width $W_I$ [m] and height above sea level $H_I$ [m]:
\[\text{Rollover capacity} = \frac{Q_w}{W_I \cdot \dot{S}} \geq 1 \quad \text{(island survives)}\]The overwash transport rate is difficult to measure directly; it depends on storm frequency, storm surge magnitude, and island topography. For Atlantic barrier islands, $Q_w$ ranges from 1–10 m³ m⁻¹ yr⁻¹ on storm-dominated coasts to 0.1–1 m³ m⁻¹ yr⁻¹ on calmer coasts.
For a 150 m wide barrier island with $Q_w = 4$ m³ m⁻¹ yr⁻¹ and $\dot{S} = 5$ mm yr⁻¹ = 0.005 m yr⁻¹:
\[\text{Rollover capacity} = \frac{4}{150 \times 0.005} = \frac{4}{0.75} = 5.3 \gg 1\]The island is rolling over successfully — overwash transport far exceeds the demand of sea level rise. But at $\dot{S} = 10$ mm yr⁻¹ (a possible late-21st century scenario):
\[\text{Rollover capacity} = \frac{4}{150 \times 0.010} = \frac{4}{1.5} = 2.7\]Still positive, but the margin has shrunk. And if development on the island (roads, buildings, dune stabilisation) reduces $Q_w$ by suppressing overwash, the island may cross the drowning threshold.
3.5 Coastal Vulnerability Index
The Coastal Vulnerability Index (CVI) of Thieler and Hammar-Klose (1999) synthesises multiple physical variables into a single dimensionless index for communicating relative coastal vulnerability. The variables and their scoring (1 = very low vulnerability, 5 = very high) are:
| Variable | Score 1 | Score 2 | Score 3 | Score 4 | Score 5 |
|---|---|---|---|---|---|
| Geomorphology | Rocky cliffs | Medium cliffs | Low cliffs | Beaches/marshes | Barrier islands |
| Coastal slope [%] | >1.20 | 0.90–1.20 | 0.55–0.90 | 0.20–0.55 | <0.20 |
| RSL change [mm yr⁻¹] | <1.8 | 1.8–2.5 | 2.5–3.0 | 3.0–3.4 | >3.4 |
| Shoreline erosion rate [m yr⁻¹] | >2.0 (accr.) | 1.0–2.0 | -1.0 to 1.0 | -1.0 to -2.0 | <-2.0 |
| Mean tidal range [m] | >6.0 | 4.0–6.0 | 2.0–4.0 | 1.0–2.0 | <1.0 |
| Mean wave height [m] | <0.55 | 0.55–0.85 | 0.85–1.05 | 1.05–1.25 | >1.25 |
The CVI is computed as:
\[\text{CVI} = \sqrt{\frac{a \cdot b \cdot c \cdot d \cdot e \cdot f}{n}}\]where $a$–$f$ are the six variable scores and $n = 6$ is the number of variables. The geometric mean (rather than arithmetic mean) ensures that a high score on any single variable significantly elevates the index — a coast with very high wave exposure but moderate scores on other variables still registers as highly vulnerable.
CVI values: <6 = very low vulnerability; 6–12 = low; 12–24 = moderate; 24–36 = high; >36 = very high.
4. Worked Example by Hand
Setting: Two coastal segments of the British Columbia coast, one rocky headland and one sandy barrier spit, to illustrate the CVI contrast.
Rocky headland — Cape Scott area:
- Geomorphology: rocky cliffs → 1
- Coastal slope: 2.1% → 1
- RSL change: +1.5 mm yr⁻¹ (isostatic uplift partially offsets eustasy) → 1
- Shoreline change: +0.3 m yr⁻¹ (slight accretion) → 3
- Mean tidal range: 3.2 m → 3
- Mean wave height: 2.1 m → 5
Very low vulnerability — despite high wave energy, the rocky coast and active uplift protect it.
Barrier spit — Boundary Bay, Delta, BC:
- Geomorphology: barrier/beach → 5
- Coastal slope: 0.15% → 5
- RSL change: +5.0 mm yr⁻¹ (eustasy + deltaic subsidence) → 5
- Shoreline change: -1.8 m yr⁻¹ (chronic erosion) → 4
- Mean tidal range: 3.4 m → 3
- Mean wave height: 0.6 m → 2
Moderate–high vulnerability — the deltaic setting, low slope, and rapid RSL rise dominate despite modest wave energy.
Bruun Rule for Boundary Bay:
$L^* = 2000$ m (very gentle nearshore), $B = 0.8$ m (low-lying spit), $D_c = 5$ m. For $\Delta S = 0.5$ m by 2100:
\[R = \frac{2000}{0.8 + 5} \times 0.5 = \frac{2000}{5.8} \times 0.5 = 344.8 \times 0.5 = 172 \text{ m}\]A 172 m Bruun retreat plus the observed chronic erosion of 1.8 m yr⁻¹ × 75 yr = 135 m suggests approximately 300 m of combined shoreline retreat by 2100 in the absence of intervention — a result with direct implications for existing infrastructure within 300 m of the current shoreline.
5. Computational Implementation
function wind_setup(U10, F, h, rho_a=1.25, Cd=1.2e-3, rho_f=1025):
tau_w = rho_a * Cd * U10^2
d_eta = tau_w * F / (rho_f * g * h)
return d_eta
function bruun_retreat(delta_S, L_star, B, Dc):
return (L_star / (B + Dc)) * delta_S
function coastal_vulnerability_index(scores):
# scores is list of 6 values (1–5)
product = 1
for s in scores: product *= s
return sqrt(product / len(scores))
import numpy as np
g = 9.81
def wind_setup(U10, F, h, rho_a=1.25, Cd=1.2e-3, rho_f=1025):
return rho_a * Cd * U10**2 * F / (rho_f * g * h)
def bruun(delta_S, L_star, B, Dc):
return (L_star / (B + Dc)) * delta_S
def cvi(scores):
return np.sqrt(np.prod(scores) / len(scores))
# Storm surge
eta = wind_setup(U10=45, F=500e3, h=30)
print(f"Wind setup (Sandy-like): {eta:.2f} m")
# Bruun Rule — Boundary Bay
R = bruun(delta_S=0.5, L_star=2000, B=0.8, Dc=5)
print(f"Bruun retreat (0.5m SLR): {R:.0f} m")
# CVI examples
scores_rocky = [1, 1, 1, 3, 3, 5]
scores_spit = [5, 5, 5, 4, 3, 2]
print(f"\nCVI — Rocky headland: {cvi(scores_rocky):.2f} (very low)")
print(f"CVI — Barrier spit: {cvi(scores_spit):.2f} (moderate-high)")
# Surge decay time
F, h = 500e3, 30
tau = F / np.sqrt(g * h)
print(f"\nSurge decay timescale: {tau/3600:.1f} hours")
6. Visualization
The three profiles diverge dramatically: a steep beach retreats less than 17 m per metre of sea level rise; a low-gradient spit retreats nearly 345 m per metre. The Bruun ratio $R/\Delta S$ is set entirely by the geometry of the active profile — gentle slopes magnify sea level rise into much larger horizontal retreats.
7. Interpretation
The decomposition of total water level into components is not merely academic — it directly determines what kind of management response is appropriate. Storm surge is episodic and can be partially managed through early warning systems and emergency response. Tides are predictable and fixed. Sea level trend is slow but cumulative and effectively irreversible on planning timescales. Wave setup is controlled by offshore wave climate.
The Bruun Rule result deserves careful interpretation. The 172 m retreat calculated for Boundary Bay by 2100 assumes the beach profile can freely adjust — no seawalls, no fixed infrastructure. The moment a seawall is built to hold the shoreline in place, the Bruun retreat can no longer occur horizontally. Instead, the beach narrows and eventually disappears as the rising sea level submerges it against the fixed structure. This is coastal squeeze — a process that eliminates intertidal habitat and beach recreation on defended lowland coasts regardless of the management intent.
The CVI captures a first-order picture of vulnerability but masks the nonlinear interactions between variables. A coast with high wave energy, high RSL rise, and a sandy barrier is not just the sum of three risks — it is a system where storm waves can drive catastrophic overwash precisely when the barrier is most vulnerable. Probabilistic compound event analysis is ultimately needed for high-stakes decisions.
8. What Could Go Wrong?
The Bruun Rule ignores sediment budget terms. It assumes the eroded sand is spread uniformly over the entire active profile offshore. If the eroded sand is lost via longshore transport, blown inland, or trapped in a lagoon, the actual retreat will exceed the Bruun prediction. Conversely, if additional sand is supplied by cliff erosion or river input, the retreat may be less. The Rule is a geometric constraint, not a complete sediment budget.
Storm surge is highly nonlinear. The linear wind setup formula assumes steady, spatially uniform wind over a flat shelf. In reality, surge is strongly influenced by the storm track, translation speed, bottom topography, and the resonance of the shelf with the storm’s spatial scale. A slow-moving storm allows more time for setup to develop; a storm whose track is perpendicular to the coast produces more surge than one moving parallel. Numerical models (ADCIRC, SLOSH) are required for operational surge forecasting.
CVI treats variables as independent. The geometric mean in the CVI formula assigns multiplicative independence to six variables that are physically correlated. A coast that scores 5 on both wave height and RSL change is not simply the product of those two risks — high waves and high sea level often co-occur during the same storm events, creating compound hazards that exceed any individual risk assessment.
Vertical land motion data are sparse. RSL change estimates depend critically on VLM measurements from GPS or tide gauge records. In data-sparse regions (most of the developing world’s coastlines), VLM is unknown and must be estimated from geological proxies or GIA models with substantial uncertainty. CVI scores for RSL change may be off by 1–2 categories in regions without instrument coverage.
9. Summary
Coastal water levels are the superposition of trend, tide, surge, and wave contributions operating on timescales from seconds to centuries. Storm surge is dominated by wind setup, whose magnitude scales linearly with wind stress and fetch and inversely with shelf depth — explaining why shallow-shelf coastlines (US Gulf Coast, Bay of Bengal) suffer the world’s most extreme surges.
The Bruun Rule translates sea level rise into shoreline retreat through a purely geometric argument: the profile must shift upward and landward to maintain its equilibrium form, and the sand must come from somewhere. On low-gradient coasts, even modest sea level rise produces tens to hundreds of metres of long-term retreat. Barrier island survival depends on whether overwash transport can keep pace with rising sea — a threshold that may be crossed as both sea level rise rates increase and storm-return intervals shorten.
The Coastal Vulnerability Index synthesises multiple physical variables into a comparative score, identifying which coastline segments are most at risk and where more detailed analysis is warranted.
Key equations:
\[\dot{\text{RSL}} = \dot{\text{SL}}_{\text{eustatic}} - \dot{h}_{\text{VLM}} \quad \text{[relative sea level rate]}\] \[\Delta\eta_{\text{wind}} = \frac{\rho_a C_D U_{10}^2 F}{\rho_f g h} \quad \text{[wind setup]}\] \[R = \frac{L^*}{B + D_c}\Delta S \quad \text{[Bruun Rule]}\] \[\text{CVI} = \sqrt{\frac{a \cdot b \cdot c \cdot d \cdot e \cdot f}{6}} \quad \text{[coastal vulnerability index]}\]Math Refresher
Linear superposition. The total water level equation $\eta_{\text{total}} = \eta_{\text{MSL}} + \eta_{\text{tide}} + \eta_{\text{surge}} + \ldots$ assumes the components add linearly without interaction. This is a useful approximation for small amplitudes; in reality, surge modifies the depth $h$ and therefore the tidal propagation speed, creating nonlinear tide-surge interaction. The superposition principle is exact for linear systems and a useful approximation for weakly nonlinear ones.
The geometric mean in CVI. The arithmetic mean $(a+b+\ldots)/n$ is appropriate when variables are independent and additive. The geometric mean $(a \times b \times \ldots)^{1/n}$ is appropriate when variables are multiplicative — when a very high score on one variable should significantly elevate the combined index regardless of others. The geometric mean is always less than or equal to the arithmetic mean (AM-GM inequality), and the two coincide only when all scores are equal.
Exponential decay. The surge decay formula $\eta(t) = \eta_{\max} e^{-t/\tau}$ has the same mathematical form as radioactive decay, first-order chemical reactions, and the RC circuit voltage discharge. In all these cases, the rate of decrease is proportional to the current value: $d\eta/dt = -\eta/\tau$. The timescale $\tau$ is the time required for the quantity to fall to $1/e \approx 37\%$ of its initial value — equivalently, after $2\tau$ it is at $14\%$, and after $3\tau$ at $5\%$.