Mohammad Mousaei Sanjerehei
Department of Plant Protection, Yazd Branch, Islamic Azad University, Yazd, Iran
Key words: Logistic regression, Life zones, Occurrence probability.
The occurrence probability of the life zones of Iran including Hyrcanian humid forests, Zagros semiarid and humid forests, humid grasslands, semiarid scrub-grasslands, arid desert scrubs and arid deserts was determined using binary logistic regression. The environmental predictors in this study were elevation, mean annual precipitation, temperature, maximum and minimum temperature, relative humidity, and reference evapotranspiration. The occurrence of the life zones with the probability of 0-1, 0.2-1, 0.4-1, 0.6-1 and 0.8-1 was compared to the reference data using kappa coefficient and Z-statistic. Mean annual precipitation was the significant variable in determination of the presence probability of 5 out of 6 life zones followed by elevation and mean annual maximum temperature which were significant for predicting the occurrence probability of 4 life zones. The highest agreement between the predicted map and the reference map was related to the Hyrcanian humid forests and arid deserts. Continuously distributed life zones with no gap had the most accurate predicted presence, while the life zones which were discontinuously and sparsely distributed, had the lowest accurate predicted presence. The best agreement between the predicted data and the reference data for all the life zones was found for the occurrence probability of 0.2-1, and this range can be used as an efficient range of probability for comparing the predicted data and the reference data in vegetation and life zone modeling based on logistic regression.
Get the original articles in Source: Volume 4, Number 6, June 2014 – JBES
Published By: International Network for Natural Sciences