3rd UF Water Institute Symposium Abstract

   
Submitter's Name Yogesh Khare
Session Name Posters - Water and Climate
Category Climate and Water
Poster Number 68
 
Author(s) Yogesh Khare,  Graduate Research Assistant (Presenting Author)
  Christopher Martinez,  Assistant Professor
   
  Global Sensitivity Analysis of a Drought Index Model – ‘Agricultural Reference Index for Drought (ARID)’
   
  Water stress (deficit) experienced by plants is one of the crucial factors that determine the loss in crop yield. Agricultural Reference Index for Drought – ARID is a generic plant water stress index which estimates the level of water stress as the ratio of reduction in potential evapotranspiration to the potential evapotranspiration, on daily time step, considering perennial turfgrass as the reference crop (Woli et al., 2010). Objective of this study was to perform global sensitivity analysis (GSA) of ARID to determine the most important parameters which shall be obtained carefully while applying this model. Root zone depth (Z), maximum uptake factor (MUF), wilting point (WP), field capacity (FC), drainage coefficient (DC) and run-off curve number (CN) are the model parameters; while inputs consist of rainfall and potential evapotranspiration to completing Plant-Soil-Atmosphere continuum. GSA was performed for five locations in south east USA and four soil types namely sandy loam, silty loam, sand clay loam and silty clay. Latin Hypercube Sampling technique was used to generate samples from the marginal probability distributions of the respective parameters. Correlation and regression based sensitivity indices were calculated amongst which the partial correlation coefficient (PCC) was considered for ranking the parameters. Scatterplots of average aridity index as well as PCCs indicate that root zone depth (Z) is the most important parameters for ARID for all soil types. FC and WP were ranked second and third in case of three soils while in silty clay the ranking was interchanged between the two. R2 values of regression model on raw values and ranked values were very high (> 0.92) indicating model is linear while scatterplots indicate the model is monotonic.