regression modeling

For an accurate irrigation schedule, the daily soil water depletion should be estimated during the crop growth period. Soil water depletion is dependent on daily evapotranspiration. In this research, daily evapotranspiration of S.C 704 maize was measured in mini-lysimeters. Estimation of daily evapotranspiration was done by continuous measurement of soil moisture. Leaves stomatal resistance was measured daily, by AP4 Porometer device. Soil water allowable depletion was determined in four growth stages of initial (C1), development (C2), mid (C3), and late (C4), based on the leaves stomatal resistance response. At each growth stage, when leaves stomatal resistance increased relative to the control crops, readily available water was ending and the time was right for new irrigation. The main variables included growth stage effect on crop evapotranspiration and water depletion coefficient, which was investigated in a completely randomized basic design, with three replications. Regression functions (models) were used for simulation of allowable soil water depletion coefficient (P) based on the daily evapotranspiration (ETc). The models were calibrated by daily data at initial and development stages, and were evaluated by daily data in mid and late stages. The FAO-56 linear model was compared with the models introduced in this research. The results showed that maize ETc (S.C 704) in initial, development, mid, and late stages was in the range of 1.5-4.5, 3.9 -7.1, 1.4 -7.5, and 0.2 -2.1 mm.d-1, respectively. The allowable soil water depletion in the mentioned stages was calculated as 0.45, 0.66, 0.61 and 0.7, respectively. Different sensitivity in crop growth stages caused readily available water limit not to be constant during growth period. The ETc increase caused a decrease in P, and decrease in ETc increased P. Linear, exponential, logarithmic, polynomial, power, and FAO-56 linear functions were investigated. Polynomial function with statistical indices of RMSE=0.00035, NRMSE=0.054, ME=0.0008, CRM=-0/000005, R2=0.999 and EF=0.999, was the optimal model in estimation of P coefficient. The reason for weak performance of FAO-56 model was the constant limit for readily available water and mean ETc rate in the growing season. Therefore, the FAO-56 model was modified. The research result was to estimate the soil water allowable depletion coefficient (by using ETc), without daily measurement of soil moisture. This method will be useful in irrigation scheduling, especially those with short intervals.

For irrigation planning, parameters such as actual crop water needs (transpiration) and water losses (evaporation) are considered. In this research, for management of deficit irrigation, the amounts of maize evapotranspiration components were simulated under water stress conditions. Water stress was applied by reducing the soil water, relative to the readily available water. Four treatments were defined as depletion of the available soil water by 40% (I0), 55% (I1), 70% (I2), and 85% (I3). The amounts of maize evapotranspiration and its components (transpiration and evaporation rates separately) were measured in a mini-lysimeter. The seasonal total values of evapotranspiration and components of transpiration and evaporation were equal to 443, 319 and 124 mm (I0), 401, 282 and 119 mm (I1), 303, 211 and 92 mm (I2), and 201, 127 and 74 mm (I3), respectively. Soil water deficiency reduced the evapotranspiration and its components relative to the normal conditions (treatment I0). Reduction of evaporation losses was favorable point in this deficit irrigation method (long irrigation interval). Transpiration and evaporation values were simulated based on the evapotranspiration data (in I0), evapotranspiration stress coefficient (Ks), and crop growth stage sensitivity (Kpi). For this purpose, we used the linear, exponential, logarithmic, polynomial, and power functions as the regression models. By using the actual data, unknown coefficients in the functions were estimated by SPSS software and regression models were generated. Statistical analyses showed that the linear function (R2= 0.91) and polynomial function (R2= 0.874) were the optimal models for estimation of transpiration and evaporation components (under water stress conditions), respectively. The actual water requirement of crop and evaporation losses can be estimated more accurately by separate estimation of evapotranspiration components. This would provide a suitable criterion for irrigation planning and calculation of water use efficiency.

Drought occurs due to lack of humidity and deficiency in precipitation amount. Therefore, it is important to recognize and investigate the effective factors on precipitation deficiency and consequently drought occurrence. Enso is a phenomenon of climatic teleconnection that can affect weather in different regions of the world. Enso is one of the important large-scale phenomena affecting the temporal and spatial distribution of rainfall and consequently the drought that occurs in the tropical Pacific. The objective of this research was to investigate the effect of Enso on meteorological drought in the midwest of Iran.

First, precipitation data of synoptic stations in 14 locations were obtained in the west of Iran. Then, SPI values was calculated at different time scales of 1, 3, 6, 9, 12, 15, 18, 24, and 48 month using the spi-sl-6.exe software package. In the next step, using the data of Enso including Nino1+2, Nino3, Nino3.4, and Nino4, the Spearman correlation test was simultaneously performed in the Minitab16 program. Then, the asynchronous relationship was investigated with the cross-correlation function between Enso and SPI in time series of months 1 to 48 in SPSS16. Finally, linear regression modeling was performed. Multivariate regression test with simultaneous and delayed states of 1-6, 12, 24, and 48 months was used based on the stepwise method.

Results are presented in three sections: simultaneous, asynchronous relations, and regression equations. Results showed a high correlation between seasons of autumn and spring with indices of Nino4, Nino3.4, Nino3, and Nino1+2. In addition, the asynchronous relationship was better than the simultaneous mode. Better significance in the asynchronous mode is due to the effect of Enso on weather and drought in Iran. Considering the p-value <0.05, it was shown that the highest significance was observed between Enso indices and drought index in Bushehr station. In addition, in a number of stations, including Arak, Urmia, Shiraz and Hamedan, no significant relationship was observed in their simultaneous state. The regression analysis showed that Nino4, Nino3.4, Nino3 and Nino1+2 indices had the highest effect in different time series of SPI with the highest positive and negative coefficients of atmospheric oceanic index in all stations. The results showed that high correlation between Enso and SPI can affect predicting climatic conditions, drought, floods, planning, and crisis management in watershed scale.

Enso can be used as predictors of long-term climatic factors such as precipitation and temperature. Enso phenomenon in Nino1 + 2, Nino3, Nino3.4 and Nino4 regions can be used as predictors of climatic factors such as rainfall for long-term forecast of precipitation and finally drought and wet season in different parts due to the universal Enso phenomenon. It seems that due to the mechanism of the Enso phenomenon and its relationship with natural disasters, further studies can be effective in predicting climatic conditions, drought and floods in Iran which ultimately can be used in predicting hydro-climatic events.