C-2 
extrapolation from measured areas into unmeasured areas, often around numerous 
bends and twists in a tidal river. Furthermore, they create the potential for interpo¬ 
lating from one tidal tributary to another, which may be inappropriate since tidal 
tributaries are often hydrodynamically independent. Most spatial interpolation algo¬ 
rithms operate in two dimensions in a relatively simple spatial domain. Thus, 
specific refinements need to be made for the algorithms used in Chesapeake Bay 
criteria assessment. 
The Chesapeake Bay dissolved oxygen criteria depend on designated-use areas— 
specific volumetric areas with both vertical and horizontal dimensions (U.S. EPA 
2003a, 2003b). Dissolved oxygen levels are naturally lower in bottom waters. There¬ 
fore, the designated-use areas were defined as vertically stratified layers to allow 
establishment of criteria levels that support the ecological communities residing in 
the lower depths of the Bay. Any spatial interpolation supporting dissolved oxygen 
criteria assessment must allow interpolation throughout the designated-use volumes 
in three dimensions. The IDW algorithm developed and used by the Chesapeake Bay 
Program was designed in this way and has been used consistently to provide 
baywide maps of dissolved oxygen concentrations (see Appendix D). Kriging, 
however, has not been used for three-dimensional interpolation in the Chesapeake 
Bay to date; in fact, only limited research has taken place to develop the capability 
of three-dimensional kriging for any purpose (STAC 2006). Thus, more research 
may be required for the use of kriging in the assessment of dissolved oxygen criteria. 
The complexity of the Chesapeake Bay shoreline presents several obstacles for 
spatial interpolation in Bay tidal waters, mostly related to interpolating across land 
area. Most spatial interpolation algorithms assume a relatively simple spatial domain 
(e.g., rectangular) and interpolation takes place without regard to direction. In 
contrast, the Chesapeake Bay (for example, see Figure III-1 in Chapter 3) displays 
tidal flow patterns that make some locations independent or virtually independent. 
For Bay water quality criteria assessment, therefore, the influence between some 
locations must be limited when interpolating spatially. The current Chesapeake Bay 
Program interpolator provides limits by using data regions in which the data used to 
estimate values in given locations are limited to certain areas (see Appendix D for 
additional details). Similar or alternative methods may be required to apply kriging 
broadly. 
As described above, the Chesapeake Bay Program collects two types of data for 
criteria assessment; these two data types supply information at different spatial 
scales. The fixed-station Chesapeake Bay Water Quality Monitoring Program 
collects data consistently for the entire Bay as well as its tidal tributaries and embay- 
ments. The Chesapeake Bay Shallow-water Monitoring Program offers much more 
detailed information within Bay tidal tributaries and across all shallow-water habi¬ 
tats. Given the different spatial scales of these two monitoring programs, it is 
unlikely that they can be used in the same interpolations. Thus, two separate inter¬ 
polation approaches—each designed for specific types of criteria attainment 
assessments—may prove necessary. 
Since the Chesapeake Bay water quality criteria and the CFD-based criteria assess¬ 
ment methodology were developed and published, interest has developed in creating 
appendix c 
Evaluation of Options for Spatial Interpolation 
