CoSMoS 3.2 Northern California sub-regional tier 2 FLOW-WAVE model input files

Model grids (.grd and .enc), elevations (.dep, m) and other Delf3D-specific setup files (.ddb, .url, .fil, .mdf., .mdw, .src, .dis, .bnd, .bct, .bcw, and meteo files) for tier 2 subregions (as described by O'Neill and others, 2018) are generated for Northern California. See Deltares Manuals (2018a and 2018b) for a description (and applicable units) of model files. Models are identified by a 2-letter designation (in runid files) and cover multiple areas of interest: 'de' covers Del Norte County, 'hu' covers Humboldt Bay, 'lc' covers Cape Mendocino, 'me' covers most of Mendocino County north of Point Arena, 'pa' covers southern Mendocino county from Point Arena to Bodega Bay, 'bt' covers Bodega Bay to Drakes Estero, and 'gn' covers south of Drakes Estero to the entrance to San Francisco Bay. Models are set up in the cartesian coordinate system (UTM 10) and elevations are referenced to NAVD88. Each model consists of one outer grid (identified by the 2-letter designation) and multiple two-way coupled domain decomposition (DD) grids (for example, hi000F, hi001F, and so on). To test the model grid/setup, water levels and wave heights were compared to observations of a storm system in January 2010 (O'Neill and others, 2018) using forcing data from the California Reanalysis Downscaling at 10 km (CaRD10) wind/pressure fields (Kanamitsu and Kanamaru, 2007). Modeled water levels and wave heights were compared to observations from tide gauges and wave buoys during the same 2-week time period preceding and during the January 2010 storm. For sub-regions covering areas with tide gages (de, hu, me, pa, bt, gn), modeled water level RMSE is 13 cm, 9 cm, 9 cm, 12 cm, 9 cm, and 12 cm respectively; modeled water levels in sub-regions without tide gauges (lc) are compared to validated Tier 1 outputs with RMSE of 7 cm. Similarly, sub-regions with buoy observations (hb, de, gn) have wave height RMSE 13 cm, 55 cm and 43 cm (corrected for timing offsets in forcing) respectively. Inspection of reanalysis wind fields (Kanamitsu and Kanamaru, 2007) indicate the January 2010 storm is not represented as accurately in the CaRD10 forcing dataset in the northernmost sections of the region as other areas of this Northern California study, affecting wave height comparisons and RMSE. Sub-regions without buoy observations (lc, me, pa, bt) are compared to validated Tier 1 output and match output with RMSE of 38 cm, 28 cm, 30 cm, and 40 cm (respectively). For an in-depth discussion of tier 2 validation and model setup see O'Neill and others (2018).

Peak discharges were identified for rivers of interest in the storm scenarios simulated in regional Tier 1 model, using methods described by O'Neill and others (2018). For Northern California, associated precipitation data from the same Global Climate Model were used to determine relationships to discharge (in lieu of using maximum SLP gradients). Relationships were identified using log-linear curves, rather than linear trends, for rivers with high correlation of extreme storm runoff estimates to neighboring rivers and creeks (O'Neill and others, 2018). Peak discharges for 'parent' rivers (Audobon Creek, Walker Creek, Big River, Noyo River, Ten Mile River, Elk River, and Little River) were scaled to associated 'child' rivers by watershed area (O'Neill and others, 2018) and included in model setup files (.dis, m3/s).

Setup files were adjusted for all storm scenarios simulated in regional Tier 1 model. Water level and wave data were extracted from each Tier 1 simulation to serve as boundary conditions for the sub-regional model. Storm events identified from future climate conditions (TWL proxies) that were simulated are: 3 May 2031 (no storm); 9 February 2020, 10 December 2071, and 19 October 2058 (1-year storm); 9 February 2074, 10 December 2052, and 15 December 2063 (20-year storm); and 8 December 2023, 27 November 2056, 5 March 2044, and 23 December 2083 (100-year storm). For each event, setup files (wind, sea-level pressure, water level, discharge and wave forcing files) are adjusted to use the extracted Tier 1 boundary conditions; discharge estimates were synced to the onset of storm conditions. For each storm event, simulations are run with a range of SLR: 0 m, 0.5 m, 1.0 m, 1.5 m, 2.0 m, 3.0 m, and 5.0 m. As done in Tier 1 and described in O'Neill and others (2018), all simulations are run over a representative spring tide in November 2010. All storm event conditions are synced to this representative tide cycle. See O'Neill and others (2018) on scenario setup and Erikson and others (2018) for information on storm selection.

Setup files supporting Mendocino County updates (following same process steps) were added to the dataset and metadata was updated accordingly.