LAGOS - Predicted and observed maximum depth values for lakes in a 17-state region of the U.S.

This dataset includes predicted and observed values of maximum depth for lakes in the
upper Midwest and northeast United States. All observed values came from LAGOS ver 1.040.0
(LAke multi-scaled GeOSpatial and temporal database), an integrated database of lake
ecosystems (Soranno et al. 2015). LAGOS contains a complete census of lakes great than or
equal to 4 ha with corresponding geospatial information for a 17-state region of the U.S.,
and a subset of the lakes has observational data on morphometry and chemistry. Approximately
40 different sources of data were compiled for this dataset and were mostly generated by
government agencies (state, federal, tribal) and universities. Here, observed maximum depth
values (n = 8164) were used to train and validate a predictive mixed effects model for lake
depth using terrestrial and lake morphology as predictors (Oliver et al., submitted).
Predicted values (n = 50 607) generated by the model had a root mean squared error of 7.1 m.
This research was supported by the NSF Macrosystem Biology awards 1065786, 1065818, and
1065649.