Upper Mississippi River Restoration Program

Long Term Resource Monitoring

Methods for Stratified Random Sampling

Water quality sampling procedures are described in detail in the LTRMP Procedures Manual (Soballe and Fischer 2004). Additional documentation is contained within the annual status reports (e.g., Pool 8) produced by each field station for the period 1993-1996. The following is a brief overview of the methods deployed by the Water Quality component of the LTRM element.

Sampling with a random stratified design began in 1993 in the six study reaches of the Upper Mississippi River. The strata from which sampling sites are selected is based on enduring geomorphic features (Wilcox 1993). The aquatic strata (Table 1) sampled are as follows: main channel (the navigation channel and its border), side channels (channels other than the main channel), contiguous backwaters (off-channel area with apparent surface water connection with the main channel), and impounded areas (large, mostly open-water off-channel areas located in the downstream portion of the navigation pools). All strata are not present in all of the sampled reaches, and additional strata are found in Pool 4 (Lake Pepin, a tributary delta lake formed by the Chippewa River delta) and Pool 26 (Swan Lake). The sampling sites are selected from a modification of a spatial database of aquatic areas (i.e., strata; Owens and Ruhser 1996). Strata are partitioned into 50-m² sampling grids for backwater and side channel areas and 200-m² sampling grids for main channel and impounded areas.

Annual sampling is conducted in each of the six study areas at a maximum 150 sites in each of four seasonal sampling episodes (Table 2). Each sampling episode is usually completed over a 14-day period. The LTRM design for stratified random sampling (SRS) requires that each day's sampling be centered on 1200 h, central standard time, and that the order of site visits within each sampling day be randomized to the extent feasible within operational constraints. Basic limnological measurements are made in the field with in situ instruments and in the laboratory using water samples obtained at a subset of the sample sites. A full listing of sampled parameters and details about analytical methods are contained in the LTRM Procedures Manual (Soballe and Fischer 2004). A description of the quality factor codes is contained in a "Primer on quality factor codes for LTRM Water Quality data".

Design Changes

There have been some changes in parameters measured, and data were not collected during some SRS episodes, during the life of the LTRM. A summary of design changes can be found in Table 3 (Table 4 contains acronyms). Please see this memo for background on differences in fixed site sampling frequency across LTRM study reaches.

Statistical Analyses

Statistical analyses of these data can range from simple design-based estimates to complex model estimates. The summary data presented on these Web pages are limited to within strata means derived from design-based methods. We report no reachwide statistics here; if doing so, then weights are needed to account for unequal probability sampling. A description of unequal probability sampling and details on recommended procedures for calculating study reach means, and the related data on strata size (population units) can be found on the LTRM statistics web page.