UPDATE 04/10/2023 
This update to the original dataset includes NEON Discharge data from data release RELEASE-2023, downloaded on 04/06/2023. Methodology, format, and other relevant steps in the analysis remain the same as Rhea et al., 2023 but this evaluation dataset is updated to include only NEON Continuous Discharge data in the RELEASE-2023.

This data product is an evaluation of the NEON Continuous Discharge product (https://doi.org/10.48443/h2ze-2f12). The dataset here provides a categorical quality classification of every site-month that NEON provides continuous discharge data. A detailed description of this evaluation dataset can be found here:

Rhea, S., Gubbins, N., DelVecchia, A.G. et al. User-focused evaluation of National Ecological Observatory Network streamflow estimates. Sci Data 10, 89 (2023). https://doi.org/10.1038/s41597-023-01983-w


Below is a brief description of the dataset:
Our evaluation dataset classifies NEON site-months into three categories of quality or flags months for potential uncertainty. Two type of flags exist in this dataset: 1) “drift flag”, and 2) “regression flag”. A drift flag indicates a site-month had detectable directional drift of continuous pressure transducer data used to measure stage height. A regression flag indicates the relationship of continuously measured stage height to manually measured gauge height has a NSE less than 0.9. Detailed information on our methods can be found in the publication describing the dataset. 

If a site-month was not flagged with one of these qualifiers, data was assigned a Tier 1, 2, or 3 rating based on the rating curve used in that site-month. Tier 1 data corresponds to data that had a Nash–Sutcliffe coefficient greater than 0.9 and had no more than 15% of reported streamflow values greater than the maximum manually measured streamflow value used to construct the rating curve. Tier 2 data corresponds to data that had a Nash–Sutcliffe coefficient greater than 0.75 and less than 0.9 or data with more than 15% and less than 30% of reported streamflow values greater than the maximum manually measured gauging. Tier 3 data corresponds to data that had a Nash–Sutcliffe coefficient less than 0.75 or had more than 30% of its continuous streamflow values over the maximum manually measured gauging. 

Data of any tier could be used based on a users needs. We recommend using Tier 1 data in almost any ecological analysis. Tier 2 data suffers from either a weaker fit of the rating curve between gauge height and streamflow or more than 15% of streamflow values over the maximum gauging used to build the rating curve. This data could be used in analyses but should be used with caution at high flows if a large amount of data exceeds the maximum gauging. Tier 3 data could be used in analyses but we recommend close inspection of relevant components before using this data. 

In addition to the information in this evaluation dataset, we encourage users of NEON’s Continuous Discharge data to use the “dischargeFinalQF” and “dischargeFinalQFSciRvw” flags in the discharge dataset as well as all other relevant information provided by NEON.

All metrics used to assign these classification are in the dataset and can be leveraged by users of this dataset to make their own decisions on what data is suitable for their needs. In addition, our code repository includes a function (set_q_criteria.R) that allows users to set thresholds for flagging and data Tiers to make it easy to set custom criteria. See github repo here: https://github.com/spencerrhea/neon_discharge_eval