Remote-sensing data are used extensively to monitor water quality parameters such as clarity, temperature, and chlorophyll-a (chl-a) content. This is generally achieved by collecting in situ data coincident with satellite data collections and then creating empirical water quality models using approaches such as multi-linear regression or step-wise linear regression. These approaches, which require modelers to select model parameters, may not be well suited for optically complex waters, where interference from suspended solids, dissolved organic matter, or other constituents may act as “confusers”. For these waters, it may be useful to include non-standard terms, which might not be considered when using traditional methods. Recent machine-learning work has demonstrated an ability to explore large feature spaces and generate accurate empirical models that do not require parameter selection. However, these methods, because of the large number of included terms involved, result in models that are not explainable and cannot be analyzed. We explore the use of Least Absolute Shrinkage and Select Operator (LASSO), or L1, regularization to fit linear regression models and produce parsimonious models with limited terms to enable interpretation and explainability. We demonstrate this approach with a case study in which chl-a models are developed for Utah Lake, Utah, USA., an optically complex freshwater body, and compare the resulting model terms to model terms from the literature. We discuss trade-offs between interpretability and model performance while using L1 regularization as a tool. The resulting model terms are both similar to and distinct from those in the literature, thereby suggesting that this approach is useful for the development of models for optically complex water bodies where standard model terms may not be optimal. We investigate the effect of non-coincident data, that is, the length of time between satellite image collection and in situ sampling, on model performance. We find that, for Utah Lake (for which there are extensive data available), three days is the limit, but 12 h provides the best trade-off. This value is site-dependent, and researchers should use site-specific numbers. To document and explain our approach, we provide Colab notebooks for compiling near-coincident data pairs of remote-sensing and in situ data using Google Earth Engine (GEE) and a second notebook implementing L1 model creation using scikitlearn. The second notebook includes data-engineering routines with which to generate band ratios, logs, and other combinations. The notebooks can be easily modified to adapt them to other locations, sensors, or parameters.

We analyzed chlorophyll-a (chl-a) concentrations in shallow, turbid Utah Lake using Landsat data from 1984 to 2021. Utah Lake is ~40 km by 21 km, has a surface area of ~390 km², an average depth of ~3 m, and loses ~50% of inflow to evaporation. This limits spatial mixing, allowing us to evaluate impacts on smaller lake regions. We evaluated long-term trends at the pixel level and for areas related to boundary conditions. We created 17 study areas based on differences in shoreline development and nutrient inflows. We expected impacted areas to exhibit increasing chl-a trends, as population growth and development in the Utah Lake watershed have been significant. We used the non-parametric Mann–Kendall test to evaluate trends. The majority of the lake exhibited decreasing trends, with a few pixels in Provo and Goshen Bays exhibiting slight increasing or no trends. We estimated trend magnitudes using Sen’s slope and fitted linear regression models. Trend magnitudes in all pixels (and regions), both decreasing and increasing, were small; with the largest decreasing and increasing trends being about −0.05 and −0.005 µg/L/year, and about 0.1 and 0.005 µg/L/year for the Sen’s slope and linear regression slope, respectively. Over the ~40 year-period, this would result in average decreases of 2 to 0.2 µg/L or increases of 4 and 0.2 µg/L. All the areas exhibited decreasing trends, but the monthly trends in some areas exhibited no trends rather than decreasing trends. Monthly trends for some areas showed some indications that algal blooms are occurring earlier, though evidence is inconclusive. We found essentially no change in algal concentrations in Utah Lake at either the pixel level or for the analysis regions since the 1980′s; despite significant population expansion; increased nutrient inflows; and land-use changes. This result matches prior research and supports the hypothesis that algal growth in Utah Lake is not limited by direct nutrient inflows but limited by other factors.

During the 2021 (n = 15) and 2022 (n = 13) summers, we measured the total and dissolved (<0.45 μm) concentration of 25 elements in Utah Lake using Inductively Coupled Plasma–Optical Emission Spectrometry (ICP-OES) with detection limits in the order of a few parts-per-billion (ppb). This resulted in 1400 measurements, which is a unique dataset in terms of sensitivity and temporal resolution. Regulated elements are not commonly measured at the ppb-level; thus, these data provide insight into both the behavior and existence of these elements in an aquatic environment and have implications for both the management and regulation of the lake. Utah regulates twelve of these elements. While ICP-OES has ppb-level sensitivity, it is not the approved regulatory analysis method for these elements. All regulations are for dissolved concentrations, except aluminum (Al) and phosphorus (P), which are for total recovery. We found total Al above the allowable concentration, but dissolved concentrations were well below allowable concentrations. We attribute high total concentrations to suspended clays. This suggests that regulatory methods should be reviewed for lakes with a high suspended-solid content. Dissolved copper (Cu) concentrations were below regulatory levels in 2021, but some samples were above regulatory levels in 2022. This could be related to the use of Cu-based algaecide treatments, or from other sources. Lead (Pb) data were inconclusive; dissolved Pb concentrations were well below the acute (1 h average) limit, but the chronic concentration limit (4 h average) was below the ICP-OES minimal detection limit. Arsenic (As) concentrations exhibited a seasonal trend that we attribute to groundwater inflows—they were below regulatory levels for aquatic environments but around the levels for drinking water. This ppb-level study with high temporal resolution provides insight into regulated elements in Utah Lake previously not available due to the high sensitivity of the method and measurements of both total and dissolved concentrations.

During 2021 and 2022 summers, we measured the total and dissolved (< 0.45 mm) concentration of 25 elements in Utah Lake using Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES). Utah regulates twelve of these elements. ICP-OES sensitivity is at the ppb-level but is not the approved regulatory method. All regulations are for dissolved concentrations; except aluminum (Al) and phosphorus (P) which are for total recovery. We found total Al concentrations above allowable, but dissolved concentrations were well below allowable concentrations. We attribute high total concentrations to suspended clays. Dissolved copper (Cu) concentrations were below regulatory levels in 2021, but some samples were above regulatory levels in 2022. This could be related to the use of Cu-based algaecide treatments, or from other sources. Lead (Pb) data were inconclusive; dissolved Pb concentrations were well below the acute (1-hour average) limit, but the chronic concentration limit (4-hour average) is below the ICP-OES minimal detection limit. Arsenic (As) concentrations exhibit a seasonal trend that we attribute to groundwater inflows. This ppb-level study provides insight into regulated elements in Utah Lake previously not available due to the high sensitivity of the method and measurements of both total and dissolved concentrations.

Estimates of ebullitive methane (CH ₄ ) emissions from reservoirs are highly uncertain, often based on spatial and temporal upscaling from a limited number of sampling sites and periods. While some of these estimates may adjust for seasonal variation in temperature, they rarely account for drawdown and the resulting changes in site depth and the areal extent of the reservoir. In Douglas Reservoir (Tennessee, USA), we conducted monthly sampling of 20 sites over 1 year to capture temporal and spatial variation in CH ₄ ebullition. We then compared various approaches for obtaining annualized, whole‐reservoir ebullition estimates. CH ₄ ebullition fluxes were zero‐inflated and highly skewed, with fluxes rising appreciably above zero in July and peaking in autumn. Estimates that did not account for drawdown overestimated annualized rates by 20%, though with high uncertainty. Seasonal variability and hotspots strongly influence annualized ebullition estimates and are important to account for in inventories and global upscaling.