Random Field Generation

Generating the Environment Variables

The depth of the Neuse estuary was obtained from a bathymetric map (NOAA, Map 11552) at a resolution of 500 m $\times$ 500 m. These depths are interpolated to the nodes on the finest level triangulation as shown in Fig. A1. Depth, $D(x,y)$ (m), is the only environment variable that does not vary temporally. The variables temperature, bottom salinity and DO all vary spatially and temporally. The actual mechanisms causing variations in temperature, salinity and DO are not explicitly modeled. Instead, the modeling goal is to create environment variables which are realistic inputs for the model and, particularly in the case of DO, which can be altered. Temperature, salinity and DO have both deterministic, spatial and temporally varying random components. Representing these variables spatially is done using a discrete representation over the estuary (Appendix A.2). The random component (Appendix A.3.1) is created by pre-generating Gaussian random fields, $X(x, y, t)$, (Ripley 1987) with mean 0, variance 1 and a particular spatial and temporal correlation structure which are then transformed to create the temperature (Appendix A.3.2) and salinity (Appendix A.3.3) variables and the disturbance regime altering bottom water dissolved oxygen (DO) (Appendix A.3.4).

Subsections

• Random Field Generation
• Temperature
• Bottom Water Salinity
• Bottom Water Dissolved Oxygen