Click on “Download PDF” for the PDF version or on the title for the HTML version.


If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Analysis of Models to Replace Missing Stage Data in an Everglades Marsh and Canal System

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org

Citation:  Transactions of the ASABE. 59(5): 1311-1317. (doi: 10.13031/trans.59.11743) @2016
Authors:   Kyle R. Douglas-Mankin, Donatto D. Surratt
Keywords:    Water balance, Water stage measurement, Wetlands.

Abstract. Complete, accurate water stage data are often essential for timely, informed water resource analysis, management, and planning. Periods of missing data in long-term water-monitoring programs are inevitable and must be addressed. This study addressed the concern that missing daily stage data have led to miscalculation of the stage-based nutrient criterion in a Florida Everglades marsh and canal system by assessing two models to replace the missing data: a new daily water-balance-based model (WB) and the well-documented gap-fill model (GF) that estimates daily stage based on statistical relationships to selected reference gages. Both models were assessed for data gaps of 1, 7, and 14 days by comparing daily stage estimates to measured stage for every possible 1-day, 7-day, and 14-day data gap over the January 1999 through April 2015 period of record. Both models resulted in smaller overall errors for back-filling marsh gage data than for canal gage data. The WB model was best for filling data gaps up to 14 days at the marsh gages in all seasons. The WB model also outperformed the GF model for the canal gage in all months except June and, to a lesser degree, May and August. Although each model provided a systematic method to replace missing stage data, thereby reducing the bias inherent in calculating the nutrient criterion for cases with missing data, the physical-process-based WB model outperformed the empirical-based GF model and minimized the need for manual error-screening methods to offset errors during periods of poor model performance.

(Download PDF)    (Export to EndNotes)