CoSMoS (Coastal Storm Modeling System) Southern California v3.0 Phase 2 flood-hazard depth and duration projections: average conditions in San Diego County

Projected Hazard: Maximum depth of flooding surface (in cm) in the region landward of the present day shoreline that is inundated for the storm condition and sea-level rise (SLR) scenario indicated.
Model Summary: The Coastal Storm Modeling System (CoSMoS) makes detailed predictions (meter-scale) over large geographic scales (100s of kilometers) of storm-induced coastal flooding and erosion for both current and future sea-level rise (SLR) scenarios. CoSMoS v3.0 for Southern California shows projections for future climate scenarios (sea-level rise and storms) to provide emergency responders and coastal planners with critical storm-hazards information that can be used to increase public safety, mitigate physical damages, and more effectively manage and allocate resources within complex coastal settings.
Phase 2 data for Southern California include flood-hazard information for the coast from the border of Mexico to Pt. Conception. Several changes from Phase 1 projections are reflected in many areas; please read the model summary and inspect output carefully. Data are complete for the information presented.
Details: Model background: The CoSMoS model comprises three tiers. Tier I consists of one Delft3D hydrodynamics FLOW grid for computation of tides, water level variations, flows, and currents and one SWAN grid for computation of wave generation and propagation across the continental shelf. The FLOW and SWAN models are two-way coupled so that tidal currents are accounted for in wave propagation and growth and conversely, so that orbital velocities generated by waves impart changes on tidal currents. The Tier I SWAN and FLOW models consist of identical structured curvilinear grids that extend from far offshore to the shore and range in resolution from 0.5 km in the offshore to 0.2 km in the nearshore. Spatially varying astronomic tidal amplitudes and phases and steric rises in water levels due to large-scale effects (for example, a prolonged rise in sea level) are applied along all open boundaries of the Tier I FLOW grid. Winds (split into eastward and northward components) and sea-level pressure (SLP) fields from CaRD10 (Dr. Dan Cayan, Scripps Institute of Oceanography, San Diego, California, written commun., 2014) that vary in both space and time are applied to all grid cells at each model time-step. Deep-water wave conditions, applied at the open boundaries of the Tier I SWAN model runs, were projected for the 21st century Representative Concentration Pathway (RCP) 4.5 climate scenario (2011-2100) using the WaveWatch III numerical wave model (Tolman and others, 2002) and 3-hourly winds from the GFDL-ESM2M Global Climate Model (GCM).
Tier II provides higher resolution near the shore and in areas that require greater resolution of physical processes (such as bays, harbors, and estuaries). A single nested outer grid and multiple two-way coupled domain decomposition (DD) structured grids allow for local grid refinement and higher resolution where needed. Tier II was segmented into 11 sections along the Southern California Bight, to reduce computation time and complete runs within computational limitations.
Water-level and Neumann time-series, extracted from Tier I simulations, are applied to the shore-parallel and lateral open boundaries of each Tier II sub-model outer grid respectively. Several of the sub-models proved to be unstable with lateral Neumann boundaries; for those cases one or both of the lateral boundaries were converted to water-level time-series or left unassigned. The open-boundary time-series are extracted from completed Tier I simulations so that there is no communication from Tier II to Tier I. Because this one-way nesting could produce erroneous results near the boundaries of Tier II and because data near any model boundary are always suspect, Tier II sub-model extents were designed to overlap in the along-coast direction. In the landward direction, Tier II DD grids extend to the 10-m topographic contour; exceptions exist where channels (such as the Los Angeles River) or other low-lying regions extend very far inland. Space- and time-varying wind and SLP fields, identical to those used in Tier I simulations, are applied to all Tier II DD grids to allow for wind-setup and local inverse barometer effects (IBE, rise or depression of water levels in response to atmospheric pressure gradients).
A total of 42 time-series fluvial discharges are included in the Tier II FLOW domains in an effort to simulate exacerbated flooding caused by backflow at the confluence of high river seaward flows and elevated coastal surge levels migrating inland. Time-varying fluvial discharges are applied either at the closed boundaries or distributed as point sources within the relevant model domains.
Wave computations are accomplished with the SWAN model using two grids for each Tier II sub-model: one larger grid covering the same area as the outer FLOW grid and a second finer resolution two-way coupled nearshore nested grid. The nearshore grid extends from approximately 800-1,000 m water depth up to 8-10 m elevations onshore. The landward extension is included to allow for wave computations of the higher SLR scenarios. Time- and space-varying 2D wave spectra extracted from previously completed Tier I simulations are applied approximately every kilometer along the open boundaries of the outer Tier II sub-model SWAN grids. The same space- and time-varying wind fields used in Tier I simulations are also applied to both Tier II SWAN grids to allow for computation of local wave generation.
Tier III for the entire Southern California Bight consists of 4,802 cross-shore transects (CST) spaced approximately 100 m apart in the along-shore direction. The profiles extend from the -15 m isobath to at least 10 m above NAVD88. The CSTs are truncated for cases where a lagoon or other waterway exists on the landward end of the profile. Time-varying water levels and wave parameters (significant wave heights, Hs; peak periods, Tp; and peak incident wave directions, Dp), extracted from Tier II grid cells that coincide with the seaward end of the CSTs, are applied at the open boundary of each CST. The XBeach model is run in a hydrostatic (no vertical pressure gradients) mode including event-based morphodynamic change. Wave propagation, two-way wave-current interaction, water-level variations, and wave runup are computed at each transect.
XBeach simulations are included in the CoSMoS model to account for infragravity waves that can significantly extend the reach of wave runup (Roelvink and others, 2009) compared to short-wave incident waves. The U.S. west coast is particularly susceptible to infragravity waves at the shore due to breaking of long-period swell waves (Tp > 15).
Resulting water levels (WLs) from both Delft3D (high interest bays and marshes) and open-coast XBeach (CSTs) were spatially combined and interpolated to a 10 m grid. These WL elevations are differenced from the originating 2 m digital elevation model (DEM) to determine final flooding extent and depth of flooding.
Events: The model system is run for pre-determined scenarios of interest such as the 1-yr or 100-yr storm event in combination with sea-level rise. Storms are first identified from time-series of total water level proxies (TWLpx) at the shore. TWLpx are computed for the majority of the 21st century (2010-2100), assuming a linear super-position of the major processes that contribute to the overall total water level. TWLpx time-series are then evaluated for extreme events, which define the boundary conditions for subsequent modeling with CoSMoS. Multiple 100-yr events are determined (varying Hs, Tp, Dp) and used for multiple model runs to better account for regional and directional flooding affects. Model results are combined and compiled into scenario-specific composites of flood projection.
Digital Elevation Model (DEM): Our seamless, topobathymetric digital elevation model (DEM) was based largely upon the Coastal California TopoBathy Merge Project DEM, with some modifications performed by the USGS Earth Resources Observation and Science (EROS) Center to incorporate the most recent, high-resolution topographic and bathymetric datasets available. Topography is derived from bare-earth light detection and ranging (lidar) data collected in 2009-2011 for the CA Coastal Conservancy Lidar Project and bathymetry from 2009-2010 bathymetric lidar as well as acoustic multi- and single-beam data collected primarily between 2001 and 2013. The DEM was constructed to define the shape of nearshore, beach, and cliff surfaces as accurately as possible, utilizing dozens of bathymetric and topographic data sets. These data were used to populate the majority of the Tier I and II grids. To describe and include impacts from long-term shoreline evolution, including cumulative storm activity, seasonal trends, ENSO, and SLR, the DEM was modified for each SLR scenario. Long-term shoreline (Vitousek and Barnard, 2015) and cliff (Limber and others, 2015) erosion projections were efficiently combined along the cross-shore transects to evolve the shore-normal profiles. Elevation changes from the profiles were spatially-merged for a cohesive, 3D depiction of coastal evolution used to modify the DEM. These data are used to generate initial profiles of the 4,802 CSTs used for Phase 2 Tier III XBeach modeling and determining final projected flood depths in each SLR scenario. All data are referenced to NAD83 horizontal datum and NAVD88 vertical datum. Data for Tiers II and III are projected in UTM, zone 11.
Outputs include: Projected flood-hazard depth and duration for the storm and sea-level rise indicated. Data correspond to the areas vulnerable to coastal flooding due to storm surge, sea-level anomalies, tide elevation, and wave run-up during the same storm and sea-level rise simulation.
References Cited: Howell, S., Smith-Konter, B., Frazer, N., Tong, X., and Sandwell, D., 2016, The vertical fingerprint of earthquake cycle loading in southern California: Nature Geoscience, v. 9, p. 611-614, doi:10.1038/ngeo2741.
Limber, P., Barnard, P.L. and Hapke., C., 2015, Towards projecting the retreat of California’s coastal cliffs during the 21st Century: in, Wang, P., Rosati, J.D., and Cheng, J., (eds.), The Proceedings of the Coastal Sediments: 2015, World Scientific, 14 p., doi:10.1142/9789814689977_0245
Roelvink, J.A., Reniers, A., van Dongeren, A.R., van Thiel de Vries, J., McCall, R., and Lescinski, J., 2009, Modeling storm impacts on beaches, dunes and barrier islands: Coastal Engineering, v. 56, p. 1,133–1,152, doi:10.1016/j.coastaleng.2009.08.006.
Tolman, H.L., Balasubramaniyan, B., Burroughs, L.D., Chalikov, D.V., Chao, Y.Y., Chen H.S., Gerald, V.M., 2002, Development and implementation of wind generated ocean surface wave models at NCEP: Weather and Forecasting, v. 17, p. 311-333.
Vitousek, S. and Barnard, P.L., 2015, A non-linear, implicit one-line model to predict long-term shoreline change: in, Wang, P., Rosati, J.D., and Cheng, J., (eds.), The Proceedings of the Coastal Sediments: 2015, World Scientific, 14 p., doi:10.1142/9789814689977_0215.

一、 预估灾害
针对指定的风暴情景与海平面上升（Sea-Level Rise, SLR）情景，当前海岸线向陆一侧受风暴影响淹没区域内的洪水表面最大深度（单位：厘米）。

二、 模型概述
海岸风暴建模系统（Coastal Storm Modeling System, CoSMoS）可在数百公里的大地理尺度上，以米级分辨率详细预测当前及未来海平面上升情景下风暴引发的海岸洪水与侵蚀。南加州版本CoSMoS 3.0针对未来气候情景（海平面上升与风暴）开展预估，为应急响应人员与海岸规划师提供关键风暴灾害信息，助力提升公共安全、减轻实体损失，并在复杂海岸区域更高效地管理与调配资源。

南加州第二阶段数据涵盖从墨西哥边境至康塞普西翁角的海岸洪水灾害信息。相较于第一阶段预估结果，多个区域已作出若干调整，请仔细阅读模型说明并核查输出结果。所呈现的信息数据完整。

三、 模型详情
1. 模型背景
CoSMoS模型分为三个层级。第一层级包含一套用于计算潮汐、水位变化、水流与流场的Delft3D水动力FLOW网格，以及一套用于计算陆架海域波浪生成与传播的SWAN网格。FLOW与SWAN模型采用双向耦合模式：潮汐流会纳入波浪传播与成长的计算，反之，波浪产生的轨道流速也会对潮汐流产生影响。第一层级的SWAN与FLOW模型采用完全一致的结构化曲线网格，网格范围从远海延伸至岸线，分辨率从远海的0.5公里逐步提升至近岸的0.2公里。第一层级FLOW网格的所有开边界均施加了空间变化的天文潮汐振幅与相位，以及由大尺度效应（例如长期海平面上升）引发的水位热膨胀升幅。在每个模型时间步长内，所有网格单元均会应用时空变化的风场（分解为东向与北向分量）以及海平面气压（Sea-Level Pressure, SLP）场，数据来源于CaRD10（加利福尼亚州圣地亚哥斯克里普斯海洋研究所Dan Cayan博士，2014年书面通信）。第一层级SWAN模型开边界施加的深水波浪条件，是基于21世纪典型浓度路径（Representative Concentration Pathway, RCP）4.5气候情景（2011-2100年），通过WaveWatch III数值波浪模型（Tolman等，2002）以及来自GFDL-ESM2M全球气候模型（Global Climate Model, GCM）的3小时间隔风场推算得到。

2. 第二层级模型
第二层级在岸线附近以及需要更高物理过程分辨率的区域（例如海湾、港口与河口）采用更高的网格分辨率。采用一套嵌套外网格与多套双向耦合域分解（Domain Decomposition, DD）结构化网格，可按需实现局部网格加密与更高分辨率。为缩短计算时长并在计算资源限制内完成模拟，第二层级沿南加州湾划分为11个分区。
从第一层级模拟结果中提取的水位与诺依曼（Neumann）时间序列，分别应用于各第二层级子模型外网格的岸线平行开边界与侧向开边界。部分子模型在采用侧向诺依曼边界时出现不稳定情况，针对此类场景，需将一侧或两侧侧向边界转换为水位时间序列，或留空未赋值。开边界时间序列均来自已完成的第一层级模拟结果，因此第二层级与第一层级之间不存在双向数据交互。由于这种单向嵌套可能在第二层级边界附近产生误差，且任何模型边界附近的数据均存在不确定性，因此第二层级子模型的范围在沿岸方向上设置为相互重叠。在向陆方向，第二层级DD网格延伸至10米地形等高线；仅在河道（例如洛杉矶河）或其他低洼区域延伸至内陆极深处的场景下存在例外。所有第二层级DD网格均应用与第一层级模拟一致的时空变化风场与海平面气压场，以模拟风增水与局部气压反演效应（Inverse Barometer Effect, IBE，即水位随大气压力梯度变化而抬升或下降）。
第二层级FLOW域共纳入42套河道径流时间序列，用于模拟因高流量河道向海径流与向内陆迁移的海岸风暴增水峰值交汇引发的回流加剧洪水。时变河道径流可施加于闭合边界，或以点源形式分布于相关模型域内。

3. 第二层级波浪计算
波浪计算采用SWAN模型，每个第二层级子模型均配备两套网格：一套覆盖范围与外FLOW网格一致的大尺度网格，以及一套分辨率更高的双向耦合近岸嵌套网格。近岸网格的水深范围约为800-1000米，向陆延伸至陆上高程8-10米处，设置向陆延伸范围是为了支持更高海平面上升情景下的波浪计算。沿第二层级子模型外SWAN网格的开边界，每约1公里施加一套从已完成第一层级模拟结果中提取的时空变化二维波浪谱。两套第二层级SWAN网格均应用与第一层级模拟一致的时空变化风场，以模拟局地波浪生成。

4. 第三层级模型
覆盖整个南加州湾的第三层级包含4802条跨岸断面（Cross-Shore Transect, CST），断面沿沿岸方向的间距约为100米。这些断面从-15米等深线延伸至至少高于北美垂直基准面88版（NAVD88）10米的位置。若断面向陆端存在潟湖或其他水道，则会对跨岸断面进行截断处理。从与跨岸断面向海端重合的第二层级网格单元中提取的时变水位与波浪参数（有效波高Hs、峰值周期Tp、峰值入射波向Dp），被施加于每条跨岸断面的开边界。XBeach模型采用静水压力模式（无垂直压力梯度）运行，包含基于事件的地貌动力变化过程。每条断面均会计算波浪传播、双向波流相互作用、水位变化以及波浪爬高。
CoSMoS模型纳入XBeach模拟，用于考虑与入射短波相比可显著提升波浪爬高影响范围的亚重力波（Roelvink等，2009）。由于长周期涌浪（Tp>15秒）的破碎作用，美国西海岸沿岸区域对亚重力波尤为敏感。
从Delft3D模型（针对重点海湾与沼泽区域）与开放海岸XBeach模型（针对跨岸断面）得到的最终水位（Water Levels, WLs），会进行空间拼接并插值至10米分辨率网格。将这些水位高程与原始2米分辨率数字高程模型（Digital Elevation Model, DEM）进行差值计算，即可确定最终的洪水淹没范围与洪水深度。

四、 模拟事件
模型系统针对预设的目标情景开展模拟，例如1年一遇或100年一遇风暴事件与海平面上升的组合情景。首先通过岸线处的总水位代理（Total Water Level Proxies, TWLpx）时间序列识别风暴事件。总水位代理时间序列的计算覆盖21世纪大部分时段（2010-2100年），假设总水位由各主要过程线性叠加得到。随后对总水位代理时间序列进行极端事件筛选，这些极端事件将作为后续CoSMoS模拟的边界条件。筛选出多组100年一遇风暴事件（有效波高Hs、峰值周期Tp、入射波向Dp各不相同），并开展多组模型模拟，以更好地体现区域与方向对洪水淹没的影响。将多组模型结果进行整合，汇编为针对各情景的洪水预估合成结果。

五、 数字高程模型（DEM）
本研究的无缝地形水深数字高程模型（DEM）主要基于加州海岸地形水深合并项目DEM，由美国地质调查局地球资源观测与科学（Earth Resources Observation and Science, EROS）中心进行了部分修改，以纳入当前可获取的最新高分辨率地形与水深数据集。地形数据来源于2009-2011年加州海岸管理局激光雷达项目采集的裸地激光雷达（Light Detection and Ranging, lidar）数据，水深数据来源于2009-2010年的水深激光雷达数据，以及主要采集于2001-2013年的声学多波束与单波束测深数据。本DEM通过整合数十套地形与水深数据集，尽可能精准地刻画近岸、海滩与崖岸的地形形态。这些数据被用于填充第一与第二层级的大部分网格。
为体现长期岸线演变的影响，包括累积风暴活动、季节趋势、厄尔尼诺-南方涛动（ENSO）与海平面上升，本DEM会针对每个海平面上升情景进行修改。沿跨岸断面高效整合长期岸线（Vitousek与Barnard，2015）与崖岸（Limber等，2015）侵蚀预估结果，以更新垂直岸线的断面形态。将各断面的高程变化进行空间拼接，得到连贯的海岸演变三维可视化结果，用于修改DEM。这些数据被用于生成第二层级第三阶段XBeach模拟所需的4802条跨岸断面初始形态，并用于计算各海平面上升情景下的最终预估洪水深度。所有数据均采用NAD83水平基准面与NAVD88垂直基准面。第二与第三层级的数据采用通用横轴墨卡托（UTM）投影，11带。

六、 输出内容
输出内容包括：针对指定风暴与海平面上升情景的预估洪水灾害深度与持续时间。数据对应于同一风暴与海平面上升模拟中，因风暴增水、海平面异常、潮汐水位以及波浪爬高而易受海岸洪水影响的区域。

七、 参考文献
1. Howell, S.、Smith-Konter, B.、Frazer, N.、Tong, X.与Sandwell, D.，2016年，《南加州地震周期荷载的垂直指纹特征》，《自然·地球科学》，第9卷，第611-614页，doi:10.1038/ngeo2741
2. Limber, P.、Barnard, P.L.与Hapke, C.，2015年，《21世纪加州海岸崖岸后退趋势预估》，收录于Wang, P.、Rosati, J.D.与Cheng, J.主编的《2015年海岸沉积物会议论文集》，世界科技出版社，第14页，doi:10.1142/9789814689977_0245
3. Roelvink, J.A.、Reniers, A.、van Dongeren, A.R.、van Thiel de Vries, J.、McCall, R.与Lescinski, J.，2009年，《风暴对海滩、沙丘与障壁岛的影响模拟》，《海岸工程》，第56卷，第1133–1152页，doi:10.1016/j.coastaleng.2009.08.006
4. Tolman, H.L.、Balasubramaniyan, B.、Burroughs, L.D.、Chalikov, D.V.、Chao, Y.Y.、Chen, H.S.与Gerald, V.M.，2002年，《NCEP海面风成波浪模型的开发与实现》，《气象与预报》，第17卷，第311-333页
5. Vitousek, S.与Barnard, P.L.，2015年，《预测长期岸线变化的非线性隐式单线模型》，收录于Wang, P.、Rosati, J.D.与Cheng, J.主编的《2015年海岸沉积物会议论文集》，世界科技出版社，第14页，doi:10.1142/9789814689977_0215