Introduction
The most important factor in timely tillage operations is the estimation of the number of working days. Probability of working days (PWD) is the ratio of workable days to days available during the working season for the desired operation. The PWD is used in some relationships in agricultural mechanization and is one of the effective factors in calculating the optimal size of the required machines, the farm capacity of the existing machines as well as the cost of timeliness. Therefore, for each day or each time period, the possibility of performing the operation is checked separately.
There are usually two methods for determining the probability of a working day which are statistics from real conditions and estimation of the feasibility of operations in the past years.
Sometimes the weather conditions indirectly prevent the operation. Soil moisture content is one of the most important factors for most agricultural operations but it cannot be directly extracted from meteorological data. Therefore, it should be initially estimated by practical methods and then the feasibility of the operation determined by comparing the moisture obtained with suitable moisture conditions. A soil water balance model is needed to determine soil moisture content. In these models, by considering the humidity in a day or a period of time and also by calculating the inputs and outputs of water to the soil profile, they determine the amount of humidity for the next time step.
To calculate the possibility of working day, daily soil moisture prediction models are often used which are based on long-term meteorological data and soil characteristics of the region. Estimating the number of working days is one of the effective factors in calculating the optimal size of required machines, the farm capacity of existing machines, as well as the cost of timeliness. The purpose of this study is to validate the proposed model to determine the probability of suitable working days for tillage operation.

Methodology
Qazvin province, with an area of 15805 square kilometers, having 488 thousand hectares of agricultural land possesses more than three percent of Iran's total production. The total cultivated area of agricultural crops is roughly 259,088 hectares from which 56.57% is irrigated and the rest designated to rainfed.
Climatic conditions and soil moisture content are key factors in the production of agricultural products. Due to the difficulty and high costs of measuring meteorological parameters and soil moisture content in different conditions, models have been proposed for estimating them in the last few decades. Recent models require input data for implementation including meteorological variables (such as rainfall, air temperature and humidity, wind speed and radiation), soil physical properties, soil moisture content and plant residues. The difficulty of field measurement and their time-consuming nature has led to the expansion of the design and application of these models. To determine the probability of working days, a model is needed that can estimate the moisture regime of bare soil in different years and determine the possibility of working in a day by determining the soil moisture content in different layers and comparing it with the humidity limits. Then, by adding up the working days, the number and probability of working days would be determined. To determine the probability of suitable working days for tillage operations, a model was designed by the Delphi software. Also, this model could be used to estimate evaporation, run-off, soil moisture content in other regions providing that those variables were available.
After dividing the soil into several layers, this model estimates soil moisture content in different time steps. In this way, having the initial soil moisture content for whole soil profile and using meteorological data and soil properties, input (rain, deep infiltration) and output (evaporation, run-off), the moisture of each soil layer is computed and then added to the initial moisture. This cycle continues repeatedly and soil moisture content is determined in each time step and in each day. In this model, according to meteorological data and soil surface wetness, evaporation from the soil surface is calculated and it is subtracted from the wetness of the surface layers from which evaporation takes place. Moisture transfer between soil layers is estimated according to the total suction difference (sum of suction and gravity) between two layers and water conductivity in the soil (hydraulic conductivity). As a result, the input and output of each soil layer is specified in each stage, and according to the initial moisture of that layer at the beginning of the stage, the moisture at the end of the stage is determined and used as the initial moisture in the second stage of simulation. The wetness of each layer in each day is determined by averaging the time steps during the working hours of that day. By comparing soil moisture content with workability limits, the possibility of working on that day is determined. By adding up working days, the number and probability of working days is determined.
Conclusion
To check the validity of the model, it is necessary to compare its results with real measurements. For this purpose, the average standard deviation, the coefficient of variation, the average percentage of relative deviation, the regression of the sum of squares and the coefficient of determination were determined as the main indicators of model validation. The measured and estimated humidity are very close to each other in most places. The average value of the standard deviation in the cities of Buin Zahra and Abyek showed the underestimation of the model and in the city of Qazvin it showed the overestimation of the model. In general, the deviation from the observed values is very small. Since this value tends to zero, it indicates a good estimation of the model and very little deviation is observed in it.
Considering that the percentage of coefficient of variation in the studied cities is low, it shows the high accuracy of the model in estimating soil moisture content.
The efficiency value of the model, which shows the quality and how to fit the observed and estimated data, varies from -0.12 to -0.47. Due to the closeness of the model efficiency values to the number 1, there is a good fit between the measured and estimated moisture values which indicates the high accuracy of the model in estimating soil moisture content.
The results obtained from calculating the average percentage of relative deviation show the acceptable fit of the data obtained from the model and the model can be used to measure the amount of soil moisture content.
The regression value of the sum of squares squares (SSE) in the fitting process should be as low as possible, because it indicates the amount of random errors. The results obtained for the regression of the sum of squares squares show the high accuracy of the model. The coefficient of determination of the model in three cities ranges from 0.81 to 0.88, which indicates the probability of correlation between two categories of data in the future. About 81-88% of the variance is shared between the estimated value and the measured value. In the process of fitting the moisture curve and unsaturated hydraulic conductivity in the designed model, the variables are investigated separately. To ensure the validity of the output of the model, three available databases were used (Shapp et al., 2001; Rawls et al., 1982 and Marcel and Parrish, 1988). From the obtained results, the ratio of the calculated value to the observed one, it is realized that the percentage of remained moisture, the percentage of moisture in the saturated state, the inverse of the potential of air entering the soil, the experimental coefficient that determines the shape of the curve, the hydraulic conductivity of saturated soil and the calculated experimental parameter (tortuosity coefficient) are close to the value of the databases, which indicates the accuracy of the model in predicting the parameters. The results show that the mean value of the bias deviation (MBE) in the maximum temperature, maximum relative humidity and wind speed was negative, which indicates the underestimation of the model. In other cases, it is positive which indicates overestimation of the model. Considering that these values are close to zero, it confirms a good estimate of the model and very small deviation is observed. The model efficiency (ME) value, which indicates the quality and how to fit the observed and estimated data, varies between -0.2 and 0.9. Due to the closeness of the model efficiency values to the number 1, there is a good fit between the measured and estimated moisture values. This result indicates the high accuracy of the model in estimating meteorological data. The regression value of the sum of squares squares (SSE) is between 0.26 and 1.83, which indicates the low amount of random errors. In addition, its lower level indicates the high accuracy of the model in predicting meteorological data. According to the results, this model can be used with high accuracy in determining the probability of working days of tillage operations.