Quality management for research weather data: USDA-ARS, Bushland, Texas

Authors: Evett, Steven - Steve; Marek, Gary; Copeland, Karen; Colaizzi, Paul

Submitted to: Agrosystems, Geosciences & Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/26/2018
Publication Date: 12/20/2018
Citation: Evett, S.R., Marek, G.W., Copeland, K.S., Colaizzi, P.D. 2018. Quality management for research weather data: USDA-ARS, Bushland, Texas. Agrosystems, Geosciences & Environment. 1:180036. https://doi.org/10.2134/age2018.09.0036.
DOI: https://doi.org/10.2134/age2018.09.0036

Interpretive Summary: There is a lack of accurate weather and crop water use data to support the development and testing of computer simulation models of crop growth, water and fertilizer use and yield. Such models are increasingly part of decision support systems used by farm advisors, agronomists, water districts and many others to calculate possible returns on investments in fertilizer, irrigation, and expensive seeds for new high-yielding crop varieties. Scientists at the USDA ARS Conservation & Production Research Laboratory, Bushland, TX, have conducted 30 years of studies using very accurate methods to measure crop water use (evapotranspiration, ET) as affected by these factors. Due to their accuracy and completeness, the Bushland ET data sets are highly sought after by those charged with building and verifying the accuracy of these simulation models. However, these models only perform as well as the weather data that are used as inputs to the models, so the Bushland team has also made a point of gathering accurate weather data in combination with the accurate ET data. This paper describes how the weather data are collected and controlled for quality to assure a long-term, accurate weather database useful in simulation model development and testing for accuracy. The weather and ET data will be placed on an Internet web site called the USDA NAL Ag Data Commons in order to be easily available to users, including model developers and testers, researchers, students, independent consultants and agricultural companies, and action agencies.

Technical Abstract: Weather mediates the interactions of soil, water, organic matter, plants and atmosphere in complex ways that impact the outcomes of agricultural and environmental research efforts. Accurate and representative weather data are thus important to most aspects of agricultural and environmental research. Quality management is an important aspect of weather data collection and processing that aims to ensure data accuracy and representativeness. Research using large weighing lysimeters at the USDA ARS Conservation & Production Research laboratory (CPRL), Bushland, Texas, has produced 30 years of data, including evapotranspiration, soil water content, plant growth and yield in response to different methods of irrigation application and management, dryland cropping, crop choice and agronomic management and weather. Simulation modeling of the soil-plant-atmosphere continuum was a research effort from the beginning of the lysimeter project, which necessitated recording of quality weather data. Weather data on solar irradiance, air temperature and relative humidity, wind speed and precipitation were gathered on a 15-minute basis at a research weather station adjacent to the weighing lysimeter fields since the inception of the project in 1987, as well as at other stations at the CPRL. The goal of this paper is to review and demonstrate CPRL weather data quality management procedures. Data quality management has involved both quality assurance planning and operations as well as quality control methods. Analysis of long-term data revealed the importance of having multiple sensors replicating the same weather measurement because sensor failure or degradation often creates gaps in data that must be filled from another data source. Data from multiple stations also revealed issues of datalogger time keeping and spatial variation of weather. And, comparisons of data on the same weather measurement from various sensors that used different operating principles revealed systematic errors in some sensors. Methods of quality assurance and control involved long-term planning for sensor calibration, maintenance and replacement, and daily and longer-term quality control efforts based on data stream analysis.