New Patterns for Estimating Daily Relative Humidity In Iranian Mountainous regions

The relative humidity (RH) is one of the important variables of atmosphere. Average weekly, monthly and annually relative humidity is usually required. These averages calculated based on averaging daily relative humidity (DRH). Thus increasing the accuracy in estimating DRH causes the accuracy of the above averages. Conventional methods for estimating the DRH is mean relative humidity for local standard hours of Iran (6:30, 12:30 and 18:30) by equation (1). This equation is a global standard procedure that using by Iran Meteorological Organization and the Ministry of Energy (e. g., [1], [2]). 6:30 12:30 18:30 RH = 0.33RH + 0.33RH + 0.33RH (1) There are some weakness points in relation (1) which causes large errors. Yao (1974) showed that the beta function has good fitting to the DRH. The behavior of DRH curve is skewed, so it is nonsymmetrical. Therefore, using the eq.1 increases errors. Corvallis (2008) proposed the relationship (2) as a default for eq.1 It has little bias for several months. Eq.2 shows that the DRH is dependent only to 3 and 15 Greenwich hours (morning and afternoon). It is observed that the effect of morning RH is approximately 2-times of afternoon RH. morning afternoon RH = 0.65RH + 0.35RH (2) The DRH curves are dependent on the climatic conditions and the months of years. Court and Waco (1956) Said that DRH which is obtaining by averaging the morning and afternoon DRH is not accurate and it is greater than the actual DRH. They also found that the DRH is greater than the average of daily minimum and maximum RH and they stated that it is depends to the months too. Day (1917) calculated the average monthly RH and concluded that it is dependent tothe month, season and geographical coordinates and has errors. The errors are negative in some places and regions. The previous studies show that the estimation of DRH by the standard three hours is not accurate and dependent to the month and climate. The Purpose of the present paper is to establish a new relationship for estimating DRH with the standard hours and also adding the temperatures and daily precipitation into these relationships. Moreover,the climates and months are also involved in relationships. We presented an equation for every month too. Iran was partitioned by around medoids clustering method (PAM) with 9 variables and it is separated into three clusters. The relationships that presented in this paper are suitable for the Mountainous cluster. The Spss.18 Software (step by step method) fitted patterns. This was done after data screening. Survey tables show the model acceptance and powerful fitting (for annual pattern with adjR 2=0.995). Transformation at several predictors causes the increasing the goodness of fit some monthly patterns. The table (1) shows annual pattern of DRH. The DRH in the months depends on the logarithm of a day before DRH (RHY). The patterns witch presented in this paper were compared and calibrated with the old traditional and Oregon patterns (equation 1 and 2). The mean square error criteria (MSE) were used for these purposed. The results showed that our patterns are more accurate. MSE of our patterns are the lowest. The intercept of our patterns are nonzero and so it has physical meaning. Because the zero-DRH is not possible in the mountain regions. Moreover the RH of 15-hours has the greatest impact on the estimating DRH. This has been inconsistent with traditional and Oregon's patterns. The all Patterns that presented in this paper are purposed for Iran's mountainousregions. These equations decrease the errors of DRH.