Post-Processing Techniques for Reducing Errors in Weighing Lysimeter Evapotranspiration (ET) Datasets

Abstract. In agriculture, evapotranspiration (ET) is a major consumptive use of irrigation water and precipitation on cropland. Global interest in sustainable management of limited freshwater supplies to meet increased food demand has resulted in increased reporting of ET measurement and modeling methods in the literature. Direct measurements of ET by large weighing lysimeters are commonly used to test other ET measurement and estimation methods. Numerous studies have emphasized the importance of proper lysimeter design and management for accurate ET measurement. However, equally important and noticeably absent from the literature are guidelines for data collection, processing, analysis, and quality assurance and quality control (QA/QC) measures. Omission and/or inadequate attention to these processes can have significant impacts on the accuracy of ET derived from large weighing lysimeters. Improper processing of rainfall, irrigation, snowfall, dew and frost accumulation, wind, and management events can also lead to substantial errors. In this study, we present data processing strategies and techniques for ET datasets measured using large weighing lysimeters. Measurements from two large lysimeters built and managed by the USDA-ARS Conservation and Production Laboratory (CPRL) in Bushland, Texas, were used for this purpose. Examples used in the study demonstrate that indiscriminate application of smoothing functions, misidentification, and misinterpretation of changes in lysimeter mass can lead to significant errors and erroneous conclusions. Comprehensive record keeping is paramount in documenting lysimeter status and operations. Errors associated with data processing are dataset specific, corresponding to the magnitude, duration, and frequency of occurrences of the aforementioned events. Understanding and prudent application of the presented techniques and QA/QC procedures can help to minimize errors in processing lysimeter datasets.