Dynamical downscaling of CSIRO‐Mk3.6 seasonal forecasts over Iran with the regional climate model version 4

Abstract

Providing reliable seasonal forecasts of the climate system is essential to make decisions for resource management and planning, such that advances in seasonal forecasting by dynamical downscaling can bring significant socio‐economic benefits. A finer resolution Regional Climate Model version 4 (RegCM4) is nested within the large‐scale fields of the coarser CSIRO‐Mk3.6 general circulation model (GCM) to dynamically downscale outputs of CSIRO‐Mk3.6 over Iran in Southwest Asia. To this end, 120 6‐month‐long simulations were conducted. Starting from January 2007, each simulation run forwards 1 month compared to the previous run, altogether covering the period 2007–2016. Comparing model results with ERA‐Interim and CRU datasets indicated that RegCM4 is able to add some value to the driving CSIRO‐Mk3.6, particularly in low lands of central to eastern Iran where CSIRO‐Mk3.6 depicts significant cold biases. Nevertheless, relatively large biases in simulations of RegCM4 are found over some regions in Iran caused by propagation of errors from the driving GCM. It is also found that errors in prediction of different phases of the El Niño‐Southern Oscillation (ENSO) events in the driving CSIRO‐Mk3.6 significantly degrade the nested model simulations, such that associated with La Niña and El Niño, large warm and cold biases are found, respectively, over Iran. Improvements should be made in physics of the CSIRO‐Mk3.6 GCM to better simulate different phases of ENSO. Results also indicated that performance of RegCM4 in simulating 2‐m temperature is strongly related to the forecast lead time, such that model errors generally become larger when the forecast time increases.