Evaluation of efficiency of different estimation methods for missing climatological data |
| |
Authors: | Mahsa Hasanpour Kashani Yagob Dinpashoh |
| |
Institution: | (1) Department of Water Engineering, Faculty of Agriculture, University of Tabriz, Tabriz, Iran |
| |
Abstract: | Reliable estimation of missing data is an important task for meteorologists, hydrologists and environment protection workers
all over the world. In recent years, artificial intelligence techniques have gained enormous interest of many researchers
in estimating of missing values. In the current study, we evaluated 11 artificial intelligence and classical techniques to
determine the most suitable model for estimating of climatological data in three different climate conditions of Iran. In
this case, 5 years (2001–2005) of observed data at target and neighborhood stations were used to estimate missing data of
monthly minimum temperature, maximum temperature, mean air temperature, relative humidity, wind speed and precipitation variables.
The comparison includes both visual and parametric approaches using such statistic as mean absolute errors, coefficient of
efficiency and skill score. In general, it was found that although the artificial intelligence techniques are more complex
and time-consuming models in identifying their best structures for optimum estimation, but they outperform the classical methods
in estimating missing data in three distinct climate conditions. Moreover, the in-filling done by artificial neural network
rivals that by genetic programming and sometimes becomes more satisfactory, especially for precipitation data. The results
also indicated that multiple regression analysis method is the suitable method among the classical methods. The results of
this research proved the high importance of choosing the best and most precise method in estimating different climatological
data in Iran and other arid and semi-arid regions. |
| |
Keywords: | |
本文献已被 SpringerLink 等数据库收录! |
|