Temperature variability assessment at vineyard scale: control of data accuracy and data processing protocol
Abstract
Context and purpose of the study – Climatic variability studies at fine scale have been developed in recent years with the reduction of material cost and the development of competitive miniaturized sensors. This work is forming part the LIFE-ADVICLIM project, of which one of the objectives is to model spatial temperature variability at vineyard scale. In the Bordeaux pilot site, a large network of data loggers has been set up to record temperature close to the vine canopy. The reduced distance between plant foliage and measurement equipment raises specific issues and leads to an increased rate of outliers compared to data retrieved from classical weather stations. Some of these were detected during data analysis, but others could not be easily identified. The present study aims to address the issue of data quality control and provide recommendations for data processing in climatic studies at fine scale.
Material and methods – Temperature variability at vineyard scale was assessed from a network of 90 temperature stations set up in Saint-Emilion, Pomerol, and their satellite appellations. In order to test the accuracy of the measurement, 2 temperature sensors T1 and T2 (Tinytag talk 2, Gemini UK) have been connected to each temperature station and programmed to record hourly minimum and maximum temperature. The accuracy given by the constructor for this material is 0.4°C. The difference between the 2 sensors for each temperature station was analyzed during the 2017 campaign and compared. A classical meteorological station installed in Saint-Emilion (Meteo France) provided the information on climatic condition in the pilot site. A temperature station was also set up next to this meteorological station to assess both the impact of canopy and the type of material on temperature. Raw temperature data and bioclimatic indices like Winkler index were analyzed.
Results – Differences exceeding material accuracy have been detected over the whole network for several locations and dates. Average of differences is higher for maximum temperature than minimum when the whole year is taken into account. Differences can change Winkler index up to 106 degree.days for the same temperature station. Seasonal effect was observed for minimum and maximum temperature with higher differences between T1 and T2 during the winter.
Significant difference on maximum temperature was observed between data from the classical meteorological station and temperature recorded by the neighboring data logger installed in the canopy. Temperature recorded by temperature station is 1 to 4 °C warmer because the solar shield is less ventilated. A seasonal effect was observed, with higher difference recorded during the summer, which induced significant differences between calculated degree days. To eliminate confusion between degree days recorded by these 2 systems, a “Canopy Winkler Index” was created for the Winkler Index constructed with the temperature station, located inside the canopy.
Careful data processing is needed to obtain accurate temperatures from miniaturized temperature station located inside the canopy. Installation of 2 sensors for each temperature station is recommended to control and detect outliers. An automatic data processing system is under development to detect and replace outliers.
DOI:
Issue: GiESCO 2019
Type: Poster
Authors
1 EGFV, Bordeaux Sciences Agro, INRA, Univ. Bordeaux, F-33882 Villenave d’Ornon,
² LETG-Rennes, CNRS-UMR 6554, Université Rennes-2, Place Recteur H. Le Moal, 35043 Rennes cedex, France
Contact the author
Keywords
Fine scale, Temperature variability, Temperature stations, Data accuracy, Data processing , Vineyards