Meteorological Drought Forecasting Using Markov Chain Model

The objective of this study is to introduce an early warning system to forecast drought using palmer drought severity index (PDSI) and Markov chain model. Based on DEM, time series of monthly PDSI of all pixels within the Laohahe Catchment from 1960 to 2005 were calculated. It is found that continuity of drought in the study area was very strong, durations of which were almost more than 2 years. There is an increasing tendency in the frequency of drought occurring in the Laohahe Catchment, which may be the result of temperature increase. Based on PDSI, the drought states of 12 months in 2000 year were forecasted using Markov chain model with different steps. The results show that Markov chain model has some capability of forecasting drought, especially for the states of normal and slight drought. The prediction performance of Markov chain model is related with the forecasting steps, little steps always getting good performance. As a result, the Markov chain model is able to work for the early warning.