Farzin Parchami-Araghi; seyed majid mirlatifi; Shoja Ghorbani Dashtaki; Majid Vazifehdoust; Adnan Sadeghi-Lari
Abstract
Introduction: In order to provide more realistic representation of processes governing the water and energy balances as well as water quality and plant physiological processes, weather data are needed at finer timescales than currently are available at most regions. In this study, a physically based ...
Read More
Introduction: In order to provide more realistic representation of processes governing the water and energy balances as well as water quality and plant physiological processes, weather data are needed at finer timescales than currently are available at most regions. In this study, a physically based framework was developed to disaggregate daily weather data needed for estimation of subdaily reference evapotranspiration, including air temperature, wind speed, dew point, actual vapour pressure, relative humidity, and solar radiation. In this paper, the results of performance comparison of the utilized disaggregation approaches are presented.
Materials and Methods: In developed framework, missing daily weather data are filled by implementation of a search-optimization algorithm. Meanwhile, disaggregation models can be calibrated using Unified Particle Swarm Optimization (UPSO) algorithm. Daily and subdaily solar radiation is estimated, using a general physically based model proposed by Yang et al. (YNG model). Long-term daily and three-hourly weather data obtained from Abadan (59 years) and Ahvaz (50 years) synoptic weather stations were used to evaluate the performance of the developed framework. In order to evaluate the accuracy of the different disaggregation models, the mean error (ME), mean absolute error (MAE), root mean square error (RMSE), and Pearson correlation coefficient (r), and model efficiency coefficient (EF) statistics were calculated. Different contributions to the overall mean square error was decomposed, using a regression-based method.
Results and Discussion: The results indicated that compared to the WAVE I, WAVE II, WCALC, ERBS, and ESRA models, the calibrated TM model had the best performance to disaggregate daily air temperature with a EF of 0.9775 to 0.9924. Compared to air temperature disaggregation models with an arbitrary value for the time of maximum and minimum air temperature, the models in which the above mentioned times are described as a function of sunrise and/or sunset had better performance in describing the diurnal variations of the air temperature. HUM III model (based on cosinusoidal disaggregation of daily actual vapour pressure) had the best performance to disaggregate daily dew point, actual vapour pressure, and relative humidity with an EF of 0.7266 to 0.8896. In addition, subdaily wind speeds were predicted with an EF of 0.3357 to 0.6300. The results showed high agreement between daily and sum-of-subdaily solar radiation (with an EF of 0.9801 to 0.9729). The use of the WAVE II and HUM II (based on linear disaggregation of relative humidity) models can be recommended for the regions with no subdaily weather data needed for calibration of the weather data disaggregation models. The results indicate the need for calibration of Green and Kozek model for disaggregation of the daily wind speed at different regions.
Farzin Parchami-Araghi; seyed majid mirlatifi; Shoja Ghorbani Dashtaki; Adnan Sadeghi-Lari
Abstract
Introduction: Subdaily estimates of reference evapotranspiration (ETo) are needed in many applications such as dynamic agro-hydrological modeling. The ASCE and FAO56 Penman–Monteith models (ASCE-PM and FAO56-PM, respectively) has received favorable acceptance and application over much of the world, ...
Read More
Introduction: Subdaily estimates of reference evapotranspiration (ETo) are needed in many applications such as dynamic agro-hydrological modeling. The ASCE and FAO56 Penman–Monteith models (ASCE-PM and FAO56-PM, respectively) has received favorable acceptance and application over much of the world, including the United States, for establishing a reference evapotranspiration (ETo) index as a function of weather parameters. In the past several years various studies have evaluated ASCE-PM and FAO56-PM models for calculating the commonest hourly or 15-min ETo either by comparing them with lysimetric measurements or by comparison with one another (2, 3, 5, 9, 10, 11, 16, 17, 19). In this study, sub-daily ET o estimates made by the ASCE-PM and FAO56-PM models at different timescales (1-360 min) were compared through conduction of a computational experiment, using a daily to sub-daily disaggregation framework developed by Parchami-Araghi et al. (14).
Materials and Methods: Daily and sub-daily weather data at different timescales (1-360 min) were generated via a daily-to-sub-daily weather data disaggregation framework developed by Parchami-Araghi et al. (14), using long-term (59 years) daily weather data obtained from Abadan synoptic weather station. Daily/sub-daily net long wave radiation (Rnl) was estimated through 6 different approaches, including using two different criteria for identifying the daytime/nighttime periods : 1) the standard criteria implemented in both ASCE-PM and FAO56-PM models and 2) criterion of actual time of sunset and sunrise in combination with 1) estimation of clear-sky radiation (Rso) based on the standard approach implemented in both ASCE-PM and FAO56-PM models (1st and 2nd Rnl estimation approaches, respectively), 2) integral of the Rso estimates derived via a physically based solar radiation model developed by Yang et al. (25), YNG model, for one-second time-steps (3rd and 4th Rnl estimation approaches, respectively), and 3) integral of the calculated Rnl based on Rso estimates derived via YNG model for one-second time-steps (5th and 6th Rnl estimation approaches, respectively). The capability of the two models for retrieving the daily ETo was evaluated, using root mean square error RMSE (mm), the mean error ME (mm), the mean absolute error ME (mm), Pearson correlation coefficient r (-), and Nash–Sutcliffe model efficiency coefficient EF (-). Different contributions to the overall error were decomposed using a regression-based method (7).
Results and Discussion: Results showed that during the summer days, 24h sum of sub-daily radiation and aerodynamic components of ETo and the estimated ETo derived from both models were in a better agreement with the respective daily values. The reason for this result can be attributed to the nighttime value of cloudiness function (f) and the longer nighttime during the cold seasons. Because the nighttime values for f are equal the f value at the end of the previous daylight period until the next daylight period. The difference between sub-daily ETo derived from the ASCE-PM and FAO56-PM models during the day and night was highly dependent on the wind speed. In case of both models, daily aerodynamic component of ETo (ETod,aero) were reproduced more efficiently, compared to radiation component (ETod,rad). Except in the case of 6th Rnl estimation approach, FAO56-PM model (with a mean model efficiency (MEF) of 0.9934 to 0.9972) had better performance in reproducing the daily values of ETo (ETod), compared to ASCE-PM model (with a MEF of 0.9910 to 0.9970). The agreement between 24h sum and daily values of aerodynamic component had a lower sensitivity to the adopted time-scale, compared to the radiation component. Compared to the FAO56-PM model the performance of the ASCE-PM model in reproducing the ETod,rad, ETod,aero and ETod had higher sensitivity to the approach utilized for calculation of Rnl and hence, to the uncertainty of net radiation. Results showed that a smaller time step does not necessarily leads to an improvement in agreement between 24h sum of subdaily and daily values of ETo. Deficiency of the standard daytime/nighttime identification criteria resulted in a higher daily averaged daytime (1.3831 to 1.6753 h) used in cloudiness function calculations, compared to the respective value used in calculations of the radiation and aerodynamic components. In order to estimate the sub-daily ETo under climatic condition of the studied region, the use of ASCE-PM model based on the 6th Rnl estimation approach, (ASCE-PM)6, with a MEF of 0.9970 is preferred, compared to other studied alternatives. Another advantage of the (ASCE-PM)6 and (FAO56-PM)6 models is their computational efficiency in case of their implementation in hydrological models.
F. Parchami Araghi; seyed majid mirlatifi; Sh. Ghorbani Dashtaki; M. Vazifehdoust; A. Sadeghi Lari
Abstract
Introduction: Subdaily estimates of reference evapotranspiration (ET o) are needed in many applications such as dynamic agro-hydrological modeling. However, in many regions, the lack of subdaily weather data availability has hampered the efforts to quantify the subdaily ET o. In the first presented paper, ...
Read More
Introduction: Subdaily estimates of reference evapotranspiration (ET o) are needed in many applications such as dynamic agro-hydrological modeling. However, in many regions, the lack of subdaily weather data availability has hampered the efforts to quantify the subdaily ET o. In the first presented paper, a physically based framework was developed to desegregate daily weather data needed for estimation of subdaily reference ET o, including air temperature, wind speed, dew point, actual vapour pressure, relative humidity, and solar radiation. The main purpose of this study was to estimate the subdaily ETo using disaggregated daily data derived from developed disaggregation framework in the first presented paper.
Materials and Methods: Subdaily ET o estimates were made, using ASCE and FAO-56 Penman–Monteith models (ASCE-PM and FAO56-PM, respectively) and subdaily weather data derived from the developed daily-to-subdaily weather data disaggregation framework. To this end, long-term daily weather data got from Abadan (59 years) and Ahvaz (50 years) synoptic weather stations were collected. Sensitivity analysis of Penman–Monteith model to the different meteorological variables (including, daily air temperature, wind speed at 2 m height, actual vapor pressure, and solar radiation) was carried out, using partial derivatives of Penman–Monteith equation. The capability of the two models for retrieving the daily ETo was evaluated, using root mean square error RMSE (mm), the mean error ME (mm), the mean absolute error ME (mm), Pearson correlation coefficient r (-), and Nash–Sutcliffe model efficiency coefficient EF (-). Different contributions to the overall error were decomposed using a regression-based method.
Results and Discussion: The results of the sensitivity analysis showed that the daily air temperature and the actual vapor pressure are the most significant meteorological variables, which affect the ETo estimates. In contrast, low sensitivity coefficients got for wind speed and the solar radiation. The similar patterns of ETo sensitivity coefficient to the air temperature ( ) and the air temperature (TA) showed that the extent of the seasonal variation of was mainly determined by the TA. Results showed a good agreement between daily and 24h sum of subdaily ETo derived from ASCE-PM (with an EF of 0.990 to 0.994) and FAO56-PM (with an EF of 0.992 to 0.995) models. The results showed a good generalization capability of the disaggregation models to estimate the subdaily ETo for the validation data set (Ahvaz). The 24h sum of subdaily ETo derived from both models underestimated and overestimated the daily ETo in calibration (Abadan) and validation (Ahvaz) data sets, respectively. In case of both models, the daily values of aerodynamic component of ETo were reproduced more efficiently, compared to radiation part. In case of the FAO56-PM model, the goodness of agreement between 24h sum of subdaily and daily values of aerodynamic part of the ETo showed a low sensitivity to variation of the time scale of weather data. With the increase of the time scale of the subdaily weather data, the ability of both models in retrieving the radiation component of the daily ETo was improved. Generally, there was no apparent relationship between the efficiency of the ASCE-PM and FAO56-PM models for retrieving the daily ETo and the time scale of weather data. Results showed that adoption of a smaller time step does not always leads to an improvement in the agreement between 24h sum of subdaily and daily values of ETo. For most of the studied subdaily time scales (1 to 360 min), the FAO56-PM model had better performance in retrieving the daily ETo, compared to the ASCE-PM model.
Conclusion: The results of this study showed that the developed disagregation framework was able to estimate the subdaily ET o. In this study, the promising results got in retrieving the daily ETo can be attributed mainly to the high sensitivity of ETo to the air temperature and actual vapor pressure (which were desegregated with a reasonable accuracy) and low sensitivity to the wind speed (which were desegregated with a low accuracy) and the solar radiation (which were disaggregated with a reasonable accuracy). The main reason for the absence of an apparent relationship apparent relationship between the efficiency of the ASCE-PM and FAO56-PM models for retrieving the daily ETo and the time scale of weather data can be attributed to adopted nighttime and daytime criteria in both models which is highly affected by time-scale of weather data and the estimated net long wave radiation.
F. Parchami-Araghi; S.M. Mirlatifi; Sh. Ghorbani Dashtaki; M.H. Mahdian
Abstract
Abstract
Infiltration process is one of the most important components of the hydrological cycle. On the other hand, the direct measurement of infiltration process is laborious, time consuming and expensive. In this study, the possibility of predicting cumulative infiltration in specific time intervals, ...
Read More
Abstract
Infiltration process is one of the most important components of the hydrological cycle. On the other hand, the direct measurement of infiltration process is laborious, time consuming and expensive. In this study, the possibility of predicting cumulative infiltration in specific time intervals, using readily available soil data and Artificial Neural Networks (ANNs) was investigated. For this purpose, 210 double ring infiltration data were collected from different regions of Iran. Basic soil properties of the two upper pedogenic layers including initial water content, bulk density, particle-size distributions, organic carbon, gravel content (>2mm size), CaCO3 percent and soil water contents at field capacity and permanent wilting point were determined on each soil sample. The feedforward multilayer perceptron was used for predicting the cumulative infiltration at times 5, 10, 15, 20, 30, 45, 60, 90, 120, 150, 180, 210, 240, 270 minutes after the start of the infiltration test and the time of basic infiltration rate. The developed ANNs were categorized into two groups; type 1 and type 2 ANNs. For developing type 1 ANNs, the basic soil properties of the first upper soil horizon were used as inputs, hierarchically. In developing the type 2 ANNs, the available soil properties of the two upper soil horizons were used as inputs, using principal component analysis technique. Results of Reliability test for developed ANNs indicated that type 1 ANNs with a RMSE of 1.136 to 9.312 cm had the best performance in estimating the cumulative infiltration. Also, type 1 ANNs with the mean RMSD of 6.307 cm had the best performance in estimating the cumulative infiltration curve.
Keywords: Artificial Neural Networks, Cumulative Infiltration, Infiltration Process, Multilayer Perceptron
