Drought index evaluation for assessing future wheat production in Greece

The main objectives of this study were (1) to evaluate the SPI, and three variations of the Palmer Drought Severity Index (PDSI) (the original PDSI (Orig‐PDSI), a self‐calibrated version (SC‐PDSI) and a modified scheme employing Priestlay–Taylor's approach to compute potential evapotranspiration (PET) instead of Thornthwaite's method) and their respective moisture anomaly indices for assessing rainfed common wheat and durum wheat yield in two pilot crop regions in north and central Greece, and (2) to assess the vulnerability of wheat production to climate change, using the most appropriate drought index, with future scenarios provided by the Hadley Centre regional climate model HadRM3. The yield models that performed best at high‐drought risk years (the Orig‐PDSI index in the northern region and the SC‐PDSI in the southern region, explaining 82.5 to 84.7% and 92% of the measured yield variability, respectively) were also the most effective at predicting the observed wheat yields when soil moisture was not an important yield‐limiting factor. However, the strength of the relationship between the drought indices and the crop yields was much weaker. Improving the representation of PET in the PDSI algorithm did not improve the model's performance. The sensitivity of the two types of wheat to HadRM3 projections for the high‐drought risk years differed dramatically between the two pilot districts, with extremely decreased yields of 3.14 tn ha −1 expected in the southern district and much smaller changes expected in the northern district (−4.6 vs + 6.7% for durum wheat and common wheat, respectively). For the low‐drought risk years, the yield models in the northern region predicted lower yields by 30 to 60 kg ha −1 . A positive yield response by 30 kg ha −1 was found for the southern district. Copyright © 2006 Royal Meteorological Society