tinw_a_2006553_sm1164.pdf (8.38 MB)
Download file

Imperfect slope measurements drive overestimation in a geometric cone model of lake and reservoir depth

Download (8.38 MB)
journal contribution
posted on 14.01.2022, 12:20 by Jemma Stachelek, Patrick J. Hanly, Patricia A. Soranno

Lake and reservoir (waterbody) depth is a critical characteristic that influences many important ecological processes. Unfortunately, depth measurements are labor-intensive to gather and are only available for a small fraction of waterbodies globally. Therefore, scientists have tried to predict depth from characteristics easily obtained for all waterbodies, such as surface area or the slope of the surrounding land. One approach for predicting waterbody depth simulates basins using a geometric cone model where the nearshore land slope and distance to the center of the waterbody are assumed to be representative proxies for in-lake slope and distance to the deepest point respectively. We tested these assumptions using bathymetry data from ∼5000 lakes and reservoirs to examine whether differences in waterbody type or shape influenced depth prediction error. We found that nearshore land slope was not representative of in-lake slope, and using it for prediction increases error substantially relative to models using true in-lake slope for all waterbody types and shapes. Predictions were biased toward overprediction in concave waterbodies (i.e., bowl-shaped; up to 18% of the study population) and reservoir waterbodies (up to 30% of the study population). Despite this systematic overprediction, model errors were fewer (in absolute and relative terms, irrespective of any specific slope covariate) for concave than convex waterbodies, suggesting the geometric cone model is an adequate representation of depth for these waterbodies. But because convex waterbodies are far more common (>72% of our study population), minimizing overall depth prediction error remains a challenge.


This work was supported by the US National Science Foundation (NSF) Macrosystems Biology Program [grant number: EF-1638679, EF-1638554, EF-1638539, and EF-1638550]. JS was also supported by the NSF Harnessing the Data Revolution Program [grant number: OAC-1934633] and Los Alamos National Laboratory [grant number: LDRD-20210213ER]. PAS was also supported by the USDA National Institute of Food and Agriculture, Hatch project [grant number: 1013544].