Search

Search Results

• 1. [Article] Modeling the uncertainty of evapotranspiration estimates

  Under today's environmental conditions and with the increasing worldwide competition for water, errors in irrigation management decisions can be costly. Therefore, providing both accurate evapotranspiration ...

  Citation × Citation Citation Abstract Copy Title: Modeling the uncertainty of evapotranspiration estimates Author: El Khoury, Laureine Under today's environmental conditions and with the increasing worldwide competition for water, errors in irrigation management decisions can be costly. Therefore, providing both accurate evapotranspiration estimation on a daily basis and estimation of the magnitude of error associated with this estimate is vital for informed water management decisions that allow irrigation managers to account for the associated uncertainties and risks. The present research focuses on the hypothesis, the analytical approach, and the results of a study formulated to quantify the various components of uncertainty in daily and cumulative reference (ET₀) and crop (ET[subscript c]) evapotranspiration estimates with and without accounting for serial correlation. These analyses were carried out using nine years of daily (24-h sum) standardized Penman-Monteith values of ET₀ downloaded from three CIMIS weather stations in the Sacramento Valley, one year of grass lysimeter data at Five Points, California and measured K[subscript c] using surface renewal, eddy covariance and lysimeter data in our area of interest. Moreover, strategies were developed, based on this study, to reduce the uncertainty in ET estimates. Random and systematic errors associated with CIMIS weather station data and with the Penman-Monteith model itself include: spatial variability of the weather parameters between the farm and the weather station; variations in weather station instrument accuracy; and the intrinsic errors of the daily (24-h sum) standardized PM ET₀ model. These sources of error were modeled using Monte Carlo simulation, and the simulated errors were used to modify daily ET₀ data for a specific case study in the Sacramento Valley. Modeling of errors in the crop coefficient was not possible given the limited data available; however an estimate of mid-season bias of the crop coefficient for the specific circumstances of the case study was derived from surface renewal data. The mean and standard deviation of simulated ET₀ estimates (including the embedded simulated errors) were used to characterize the errors in irrigation scheduling that would result from random, systematic and total errors in ET₀ and ET[subscript c] for different irrigation scenarios. The three alternative scenarios considered were represented as three different allowable depletions, 50 mm, 100 mm and 150 mm, between irrigations. Key elements of this research explored the importance of serial correlation in random errors and compared the errors associated with weather station factors with the intrinsic error of the Penman-Monteith model. The research also showed that observed biases in the CIMIS crop coefficient (K[subscript c]) produced a substantially greater effect than errors associated with weather station factors or intrinsic errors in the PM model. The research also found that local measurements of wind speed could considerably increase the accuracy of ET₀ estimates. Title: Modeling the uncertainty of evapotranspiration estimates Author: El Khoury, Laureine Under today's environmental conditions and with the increasing worldwide competition for water, errors in irrigation management decisions can be costly. Therefore, providing both accurate evapotranspiration estimation on a daily basis and estimation of the magnitude of error associated with this estimate is vital for informed water management decisions that allow irrigation managers to account for the associated uncertainties and risks. The present research focuses on the hypothesis, the analytical approach, and the results of a study formulated to quantify the various components of uncertainty in daily and cumulative reference (ET₀) and crop (ET[subscript c]) evapotranspiration estimates with and without accounting for serial correlation. These analyses were carried out using nine years of daily (24-h sum) standardized Penman-Monteith values of ET₀ downloaded from three CIMIS weather stations in the Sacramento Valley, one year of grass lysimeter data at Five Points, California and measured K[subscript c] using surface renewal, eddy covariance and lysimeter data in our area of interest. Moreover, strategies were developed, based on this study, to reduce the uncertainty in ET estimates. Random and systematic errors associated with CIMIS weather station data and with the Penman-Monteith model itself include: spatial variability of the weather parameters between the farm and the weather station; variations in weather station instrument accuracy; and the intrinsic errors of the daily (24-h sum) standardized PM ET₀ model. These sources of error were modeled using Monte Carlo simulation, and the simulated errors were used to modify daily ET₀ data for a specific case study in the Sacramento Valley. Modeling of errors in the crop coefficient was not possible given the limited data available; however an estimate of mid-season bias of the crop coefficient for the specific circumstances of the case study was derived from surface renewal data. The mean and standard deviation of simulated ET₀ estimates (including the embedded simulated errors) were used to characterize the errors in irrigation scheduling that would result from random, systematic and total errors in ET₀ and ET[subscript c] for different irrigation scenarios. The three alternative scenarios considered were represented as three different allowable depletions, 50 mm, 100 mm and 150 mm, between irrigations. Key elements of this research explored the importance of serial correlation in random errors and compared the errors associated with weather station factors with the intrinsic error of the Penman-Monteith model. The research also showed that observed biases in the CIMIS crop coefficient (K[subscript c]) produced a substantially greater effect than errors associated with weather station factors or intrinsic errors in the PM model. The research also found that local measurements of wind speed could considerably increase the accuracy of ET₀ estimates. Close