Developing drought vulnerability curves for the agricultural sector in the Po Basin

The agricultural sector is the largest user of water in Italy; more than 50% of the country’s total water volume is allocated to irrigation according to ISTAT. Twenty percent of the arable land is irrigated on average at national level. The highest percentage of irrigated land is in the North, with 53%, 42% and 37% of irrigated land in Lombardy, Veneto and Piemonte, respectively. The three regions are part of the Po basin, which is the largest Italian agricultural area and accounts for 35% of the country’s agricultural production. The basin was struck by multiple droughts over the past years [1]. [2] highlighted that the 2003-2008 events caused huge economic impacts on the agricultural sector, with the 2005-2007 period accounting for around 1.857B€ of losses. The estimation of drought impacts on agriculture is fundamental to foresee the effects that future droughts will have on the sector. A quantitative evaluation of crop yield reduction due to water scarcity can be obtained through vulnerability curves, which relates water availability with crop yield. In this study an investigation of the major droughts that hit the basin over the past years was carried on using the SPEI computed based on the E-OBS dataset. E-OBS provides daily data from 1950 to 2020 with a 10km grid resolution. It is often used to assess magnitude and frequency of extremes. It has a good coverage for the Po basin [3]. Its main advantage with respect to station data stays in the consistency of the data over time, thus avoiding the issue of incomplete timeseries. The events identified based on E-OBS were compared with the ones reported in [1] where the SPEI was obtained from station data. The datasets showed a good agreement. Then crop specific drought vulnerability curves tailored on the Po basin were developed. The Agricultural Production System sIMulator (APSIM) crop model was selected to simulate the crop growth. The model was initialized with meteorological data from E-OBS and soil type data from the ISRIC dataset. The model was calibrated on ISTAT yield data aggregated at province level for the 2006-2020 period. The main droughts were correlated with reduction in crop water availability and thus with a consequent yield lowering. Winter wheat and maize were selected for the analysis. Winter wheat is traditionally rainfed in the area, while maize is irrigated. Results showed that 2017 was a critical year for winter wheat production; irrigation was necessary to avoid high yield losses. Flowering and yield formation were the crop growth stages more sensitive to water deficit, thus playing a key role in determining the final crop yield. In the case of maize, springs 2012 and 2017 were the growing seasons affected by drought. Flowering was found to be the growth stage more affected by water deficit, followed by yield formation and ripening. Results agree with the vulnerability functions proposed by the FAO who underlined the importance of avoiding water stress during flowering to get high yield.