Assessing Uncertainties in Estimating Potential Vegetation Water Demand
Abstract
Sustainable management of land and water resources require understanding of the magnitude and variation of evaporative water losses from land surface to the atmosphere, a process known as evapotranspiration (ET). Factors such as weather, vegetation characteristics, and land use control regional ET. Many methods are used to estimate ET, many of which are physically-based that employ several weather parameters. Here, we evaluate seven methods to estimate potential ET (PET). PET is the upper limit of plant water use constrained by weather conditions. We evaluated the Penman-Monteith (PM) method and three variations with suggested weather parameter approximations wind speed (PM1), humidity (PM2) and, solar radiation (PM3) Hargreaves-Samani (HM), and Priestly-Taylor (PT) methods. The PM formulation that uses all the required weather parameters is used as reference to compare uncertainties in all other PET estimation methods. We use the daily weather records available at the Sanborn Field Station, Columbia, Missouri for the period 2011 to 2021. Our preliminary results reveal that the annual and growing season PET varies 1052±60 mm, and 770±50 mm, respectively based on the reference PM method. However, the PT method overestimates annual PET by 11%, and the HS method underestimates annual PET by 4% compared to the reference PM method. Multiple goodness-of-fit measures including Skill Score (SS), percent bias (a measure of the average tendency of the estimated values away from the reference values) and Coefficient of Efficiency (a normalized measure describing the magnitude of variance for estimated values relative to that of the reference values) are used to evaluate uncertainties at daily, monthly, seasonal, and annual time scales. The SS is further decomposed into three components: i) correlation between the estimated and reference values, ii) bias in the estimated and iii) bias in the reference values to explain the systematic and random errors in PET estimation methods. We will present additional evaluations based on weather data collected at several Missouri MESONET stations that spans multiple climatic conditions. The results of this research will be valuable for their insight into estimating crop water demand more accurately in a changing climate to effectively manage water use in the agriculture sector.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFMGC42I0802F