Processed high resolution rainfall, flow rate, sediment and nitrate concentration data

This dataset contains data on the flow rate and chemistry (sediment, nitrate) of water in surface runoff and throughflow, plus rainfall quantities, from four field scale catchments in South West England. The catchments are part of the North Wyke Farm Platform (NWFP), and they varied with regards the crops grown. Over the time period of this data collection (October 2016 to March 2020), crops included three different pasture types, and a winter wheat. This dataset takes the raw data available on the NWFP data portal and provides a cleaned-up version, rectifying issues where there were data flagged as potentially poor quality by the quality control process, or there were missing data. This dataset has subsequently been used to model the effect of climate change on the losses of suspended sediment and nitrate from agricultural land (https://doi.org/10.1016/j.jclepro.2022.130633).

The data collection took place on the North Wyke Farm Platform (NWFP), a UK National Capability in SW England. The NWFP is split into a number of self-contained farms (‘farmlets’) that are managed according to different operation philosophies or practices. Within each farmlet are a number of hydrologically isolated catchments, and the slope and soil type of the NWFP means that all water falling within a catchment flows out via a flume. The flume is instrumented for the determination of various parameters, including water chemistry, turbidity, and flow rate, at 15-minute intervals.

The data are from four catchments in three farmlets. Catchments 2 and 3 belong to the red farmlet, which, at the start of the time period encompassed by this dataset, was sown with high sugar ryegrass (Lolium perenne cv. AberMagic). Between August and October 2019, the land was sprayed with herbicide, ploughed and sown with winter wheat (Triticum aestivum, cv. Crusoe). Catchment 5 is on the green farmlet, which has fields of permanent pasture. Catchment 8 belongs to the blue farmlet, where fields are sown with a mixture of high sugar ryegrass (Lolium perenne cv. AberMagic) and white clover (Trifolium repens cv. AberHerald). Details about the NWFP design and development, including catchment sizes and a map, can be found in the report at https://doi.org/10.23637/rothamsted.98y1x.

All raw data were download from NWFP data portal along with attached quality control codes. The NWFP data portal hosts rainfall data obtained from tipping bucket rain gauges on each catchment, plus a Pluvio rain gauge situated fairly centrally to the NWFP catchments. Information about the rainfall gauges is given at https://doi.org/10.23637/rothamsted.98y4w.

The NWFP data portal (accessible via http://resources.rothamsted.ac.uk/farmplatform) provides open access to other data which users may find beneficial as supplementary data, including: 15-minute Met measurements; 15-minute soil moisture measurements; 30-minute GHG emissions; soils, crop and botanical field survey data; livestock and crop performance data; farm operational activities; and contextual information.

All raw data were download from NWFP data portal along with attached quality control codes. Information on the quality control codes can be found at https://doi.org/10.23637/rothamsted.98y4y.

For rainfall data, the spatial variation among different catchments was assumed to be insignificant. Data with ‘Acceptable’ quality tags were used for the estimation of the median value for whole site. For data labelled as ‘Outlier’, they were reviewed by considering consistency among the catchments and reference to the monitored data from the Pluvio rain gauge. A knowledge-based judgement was made to accept or reject the data points. For any 15-minute intervals which had no valid data available after the quality tag-based filtering, data from the Pluvio rain gauge were used. A continuous rainfall data series was thereby generated and daily total rainfall was calculated from these figures.

For flow data, the following stepwise procedures were followed:

Flow data with an ‘Acceptable’ tag were used without changes. Data judged to be ‘Suspicious’ or ‘Highly suspicious’ were removed.

The ‘Outlier’ data were assessed based on the corresponding rainfall intensity, consistency among catchments, and reported discharges in the previous years. Case specific judgements were made about the validity of the data.

After the filtering based on quality tags, the data gaps were infilled to generate continuous flow data time series. If the gap length was fewer than 4 data points, or there was no recorded rainfall for the period, linear interpolations were undertaken using the existing data. For data gaps with longer length (> 1 hour) and/or with significant rainfall (>10 mm), empirical relationships between catchments with similar land uses, before the data gap, were established to predict the expected flow rate.

Similar approaches were adopted for the filtering of nitrate and suspended sediment data. Turbidity data were converted to suspended sediment concentration using existing universal calibrations (equation 2 of https://doi.org/10.1016/j.still.2020.104746), where no catchment specific relationships were considered (i.e., the entire NWFP). No data infilling was undertaken.