A new damage-probability approach for risk analysis of rain-fed agricultural systems under meteorological drought

Droughts are natural part of virtually all climates and cause losses of income around the globe. Traditional crisis management approach has been ineffective since passive responses are poorly planned and coordinated. On the contrary, risk management paradigm aims at reducing vulnerability to disasters through advocating preparedness and mitigation. Although the core of risk management is quantified risk analysis, few studies have been reported to lay out the risk analysis framework for droughts. In this paper, a new approach to develop drought Damage-Probability Curve (DPC) for risk analysis is proposed. Drought damage estimation is performed via a neural network model and, for the first time, a trivariate copula was incorporated into damage probability estimation. On the basis of DPC, robust drought risk analysis tools such as the expected value of damage (annual risk), exceedence probability curve and damage return period curve are developed. Regression, Standardized Regression Coefficient (SRC) and Correlation Coefficients (CC) techniques are applied to investigate input sensitivity analysis. Eventually, reliability analysis of rain-fed crop production is performed. The proposed risk analysis approach is evaluated on rain correlation coefficients-fed wheat production over the Qazvin region, Iran. Results show that the proposed approach may be an applicable to assess drought risk.