Monitoring and Forecasting of Groundwater Drought Using Groundwater Resource Index (GRI) and First to Third- Order Markov Chain Models (Case study: Boroujen Plain)

Management of water resources, especially groundwater, is important in arid and semi-arid regions. One of the important issues in optimum water resources management is the prediction of drought conditions. Groundwater is considered as the main resources of water supply for agriculture, industry and drinking uses in Boroujen plain. Therefore, it is important to investigate the drought condition of groundwater resources in the planning and sustainable management of these resources. So far, various methods have been developed and used by researchers to predict different types of droughts. One of these methods is the prediction of wetness conditions by the Markov chain. In most of the previous studies in the field of drought prediction, the Markov chain of first and second orders have been used. In this study, groundwater droughts in Boroujen Plain during the years 1985 to 2015 are assessed and the wetness conditions of this plain are predicted using the third order Markov chain model. In order to assess the groundwater droughts in the Borujen Plain, the GRI index values were calculated on the time scales of 1, 3, 6 and 12 months. For this purpose, the data of groundwater level of 13 piezometric wells in Boroujen plain during a 31 year period (1985-2015) was used. In order to predict the GRI index values in Broujen plain for the next months, the first, second and third- order Markov chain models were used and the performance of these models was evaluated based on contingency table method. After forming the contingency table from the results of first, second and third- order Markov chain models, the values of CSI, POD and FAR statistics were calculated. The CSI values for Broujen plain in the time scales of 1, 3, 6 and 12 months for the first order Markov chain model were calculated equal to 0.58, 0.50, 1.0 and 1.0, respectively. The CSI values for the second order Markov chain model in the time scales of 1, 3 and 6 months were obtained equal to. 0.45, 0.33 and 1.0, and for third order Markov chain model equal to 0.40, 0.38 and 1.0, respectively, which indicate the medium skill of the developed method in the prediction of wetness conditions at 1 and 3 months time scales, and good skill at 6 months’ time scale. Also, the delineation maps of GRI index were drawn by selecting the most suitable interpolation method. The delineation map of GRI in the Broujen Plain shows that the middle parts of the plain often experienced severe droughts. Comparing the performance of different orders of Markov chain in predicting the wetness conditions of Boroujen plain based on CSI, POD and FAR statistics showed that first order Markov chain method presented more accurate results than other models in predicting GRI values in all time scales. Therefore, it can be used to predict the groundwater drought in Boroujen Plain. The results of the GRI survey for Boroujen plain showed that during the period under study, the drought spell of groundwater began in 2008. In general, the severe droughts that have occurred in recent years due to reduce atmospheric precipitation, along with the overexploitation of groundwater have caused the severe decline in groundwater levels, which leads to groundwater quality degradation and land subsidence in the Boroujen plain.