Application of geostatistical methods for climatic classification (A case study, Ourmieh lake basin)

Authors

1 Agrometeorologyst

2 Assistant professor, Soil conservation and watershed management research center, P.O. Box 13445-1136

3 Institute of Geophysics, University of Tehran, P.O. Box 14155-6466, Tehran, Iran

Abstract

Spatial analysis of temperature and rainfall is necessary in many natural resources studies including water resource improvements, management, modeling and irrigation. In order to determine spatial variability of temperature and rainfall, classic statistical methods are usually applied. These methods may lead to imprecision results because; they do not include data arrangement and location perfectly.
In the present research, three methods were used to estimate regional rainfall and temperature using geostatistics and geographic information system. These methods consist of Thin Plate Smoothing Spline (TPSS), with the power of 2, 3, 4 and 5, with and without elevation as co variable; Kriging (ordinary kriging and co kriging) and Weighted Moving Average (WMA) with the power of 1, 2, 3, 4, and 5. These methods were applied to a thirty years data set of the Ourmieh lake basin in the north west of Iran. Mean Absolute Errors (MAE) and Mean Bias Errors (MBE) were used as comparison criteria.
It was found that the TPSS method with a power of 2 and without co variable (elevation) is an appropriate method to estimate annual rainfall. Kriging and WMA methods are the second and third best methods, respectively. The same results are applicable for monthly rainfall most months.
For temperature estimation, the TPSS method with a power of 2 and with elevation as co variable was the best method and co kriging, kriging and WMA were second, third and fourth best methods, respectively.
By comparing geostatistical results with those calculated by the gradient method, a higher precision was obtained using geostatistical methods for considered variable in the area studied. Finally, the climate of the area studied, was scaled as semi arid and humid using the Selianinov climatic classification method. Geographical distribution of rainfall and annual sum of active temperature with based value of 10 C were used for the method mentioned.

Keywords