Trend Analysis Process Overview

This page displays the results of a statistical ten-year trend analysis developed to use selected water quality parameters from the Water Atlas. In addition to a summary visualization of the trend results, it also provides a compiled data set, and explanatory digital documents that will be valuable to natural resource managers as they attempt to characterize and react to water quality conditions and trends. The statistical analytical techniques employed by this tool were modified from the approach used by Janicki Environmental, Inc. in the 2013 Water Quality Data Analysis Report for the Coastal & Heartland National Estuary Partnership. Data used in this analysis must meet requirements for testing/correction of seasonality: i.e., there must be at least four (4) sample values in each calendar month over the course of a 10-year period.

Prior to 2024, trend analysis was performed with the Seasonal Kendall Tau test in the EnvStats R package; from 2024 onward, we have switched to the Seasonal Mann–Kendall (SMK) test implemented in the PymannKendall Python package^{[1, 2]}. The model employs techniques to account for seasonality, autocorrelation and duplicate sampling, in an effort to detect statistically significant trends in the data. Analysis was performed on a suite of water quality measures for all available monitoring sites. For each site/water quality measure, the analysis determined whether a statistically significant trend was detected, and if so, whether it was increasing or decreasing, weak (<10% change/year) or strong (>10%/year).

The amount of nutrients entering a water body has important effects on water quality. Plants and animals that live in lakes, rivers and estuaries use these nutrients, especially nitrogen and phosphorus, to grow and survive. However, when excessive amounts of nutrients enter the water, negative impacts can occur, such as algal blooms that block sunlight for submerged plants and trigger events that deplete the oxygen in the water and result in fish kills. The indicators shown here are those most valuable in assessing the health of our waterways relative to nutrient pollution.

1. Millard SP (2013). EnvStats: An R Package for Environmental Statistics. Springer, New York. ISBN 978-1-4614-8455-4, https://www.springer.com.
2. Hussain et al., (2019). pyMannKendall: a python package for non parametric Mann Kendall family of trend tests. Journal of Open Source Software, 4(39), 1556, https://doi.org/10.21105/joss.01556

Hotspot Analysis Process Overview

The last five years of water quality data collected at each monitoring location are used for this analysis. First, monitoring locations are grouped based on their type: freshwater lake, freshwater stream, tidal stream, or bay (as applicable). Then, frequency distribution analysis is performed to define a percentile distribution for each parameter, for each of the three groups. Finally, for each monitoring location, an annual median value is calculated for each parameter, and that median is used to place the monitoring location in the correct percentile range, depending on its group.

Monitoring stations are assigned to a group based on the type of water body monitored: freshwater lake, freshwater stream, tidal stream, or bay. For streams, the classification of the stream segment (freshwater or marine) is determined using the Florida Department of Environmental Protection's water body ID (WBID) definition.

As is the case for the Water Quality Trends analysis, the monitoring locations selected for hotspot analysis must meet minimum data sufficiency and seasonality standards (enough data points collected, spread throughout the year).