What can we learn from 100,000 freshwater forecasts? A synthesis from the NEON Ecological Forecasting Challenge: scores and targets

<p>This data publication contains the zipped scores parquet database and the targets used in the analysis for the NEON Challenge aquatics synthesis: scores.zip contains evaluated forecast scores, targets.zip contains in-situ water temperature and dissolved oxygen data generated for the Challenge, and Model_descriptions.csv contains the meta data for the submitted models used in the analysis.&nbsp;</p> <p><strong>scores</strong>: this is the primary dataset used in the manuscript. The database includes the scores (forecasts evaluated against observations).</p> <p><strong>targets</strong>: these are the data used to evaluate the forecast (observations) and were also available to forecast teams in near-real time.&nbsp; The R script included in this folder (aquatics_targets.R) is intended for reference only and cannot be run.&nbsp;</p> <p><strong>Model_descriptions.csv</strong>: table of summarised meta data on forecast model approaches submitted by forecast teams. This includes whether the forecasts include (1) or exclude (0) each source of uncertainty/initial conditions/model updating, which covariates are included, and the reported model class.&nbsp;</p>