HYDROLOGICAL MODELLING FOR THE PREVENTION AND THE MANAGEMENT OF WATER SHORTAGE IN AGRICULTURE

In recent decades, frequent and severe droughts have occurred in several countries of the world under nearly all climatic regimes. Since the middle 20th century, drought areas have globally increased, and, more specifically, in southern and central Europe. Drought risk is expected to increase in the near future as a result of the climate change, leading to a decline in precipitation and an increase in air temperatures, and consequently in evapotranspiration rates in several regions, including southern Europe and the Mediterranean region. Droughts can significantly affect the agricultural sector since they provoke losses in crop yields and livestock production, increased insect infestations, plant diseases and wind erosion. Moreover, low rainfall during the growing season may affect irrigated agriculture over subsequent years, as a result of low levels of water in reservoirs and groundwater aquifers. In Europe, the monitoring and assessment of drought is entrusted to the European Drought Observatory (EDO), that applies a multi-indicator approach, based on earth observations (EOs) and hydrological modelling data. EDO indicators are computed considering rainfed agriculture, predominant in middle and northern Europe, and are produced on a 5 km grid. In southern Europe, however, the implementation of drought-coping measures (irrigation) can partially or completely alleviate the impacts of potentially severe droughts. Therefore, for these conditions, specific water scarcity indicators explicitly considering irrigation among the water inputs to agro-ecosystems need to be developed and adopted to inform and support stakeholders and decision makers of irrigated regions. In this context, the main objective of the Ph.D. thesis is the presentation of the Transpirative Deficit Index (TDI), a newly developed indicator for the monitoring and the management of Water Scarcity and Drought phenomena based on the use of hydrological modelling, applied at a spatial scale of interest for end-users (250 m grid) and suited for the assessment of water scarcity and drought in Italy as well as in other southern European countries. In particular, TDI was developed as a new module integrated into the spatially distributed hydrological model IdrAgra, and in the Ph.D. research it was tested over the Irrigation District of Media Pianura Bergamasca (IDMPB), considering a simulation period of 22 years (1993-2014) and subdividing the territory by means of a grid with cells of 250 m×250 m. As a first step in the thesis, D TDI was described as an agricultural drought index focusing on overcoming the limitation of other approaches, not taking into account with sufficient detail land cover and soil properties. The D TDI is based on the calculation of the spatially distributed actual transpiration deficit, to determine the level of drought experienced by crops within the single model cells; thus, it can provide a much more accurate measure of agricultural drought at the irrigation district scale than the one that could be achieved through meteorological drought indices such as SPI or SPEI. The auto-correlation analysis of D-TDI showed to be positive with a persistence of 30 days for the two more widespread crops in the study area, maize and permanent grass. The analysis demonstrated also that soils characterized by a high available water content can more easily compensate dry spells. Finally, a positive significant correlation between D-TDI and SPI was observed for maize, with a persistence of 40 days, while no correlation was observed for permanent grass, probably related to cutting cycles, that could mask the relation between storage capacity and short-time variability of the meteorological conditions. Successively, a methodology to compute crop yield using moderate spatial and temporal resolution Earth Observation (Landsat) data was set. In particular, the developed procedure, based on the integration of the Available Photosynthetically Active Radiation over the growing season, showed that statistical inventories and satellite data can be integrated to produce annual spatially distributed estimates of cropland productivity, while site-specific observational field data can be used to validate the relationship between APAR and productivity for specific crops (i.e. maize in this Ph.D. research). A phenological parameter extraction algorithm was developed to derive key phenology stages for the maize crop. However, the results presented in the study showed two main weaknesses: (1) cloud cover and noise in the original Landsat dataset were not appropriately removed by the Whittaker algorithm, and (2) SOS (Start of Season) and EOS (End of Season) extracted from satellite data were underestimated for a discrete numbers of fields with respect to observed ground-truths, probably as a consequence of the method adopted for setting the thresholds. A crop specific light use efficiency (ε_b^*) was estimated as the ratio between the average maize yield over the study period taken from Regional Statistic Inventory (Regional Authority and ISTAT), and the average APAR value calculated for the maize pixels over the same spatial extension and time period. The εb* estimated value fell within the range of the coefficients calibrated with other satellite-based algorithms. Finally, TDI was applied as a water scarcity index (WS-TDI), thus including water availability for irrigation within the inputs of the IdrAgra model. The behaviour of D-TDI and WS-TDI was compared over the same area, analysing their spatialized trend in response to varying meteorological conditions, and in particular considering drought events and dry spells. The two indices proved to be suitable to monitor agricultural drought and water scarcity over a territory, and helped in identifying drought and/or water scarcity prone sub-districts, as a function of crop, soil type and water availability. Both D-TDI and WS-TDI could therefore be used as operational indicators to produce periodic maps that could help farmers and irrigation district managers in coping with agricultural drought and water scarcity and, eventually, in setting up proper adaptation measures. In particular, in case of availability of real time meteorological data and water discharges at the main surface water diversions, the indicators may be adopted by an authority responsible for the monitoring of the state of agriculture (ERSAF or ARPA in the Lombardy region) to promptly inform (through newsletters or a web site) stakeholders on the agricultural drought/dry spells and water scarcity/shortages phenomena evolution. Additionally, the indicators may be adopted in climate change studies, allowing to visualize the evolution of drought and water scarcity phenomena over the territory, as a consequence of changes in meteorological forcing and in the availability of water by irrigation sources. Finally, they could be used as useful tools to support planning decisions on water resources allocation or action plans to reduce water consumptions in specific portions of the territory (e.g. conversion of irrigation methods, introduction of different crop species, etc.), also in view of an adaptation to the climate change. WS TDI maps over a pilot study area were statistically compared with the maize yield maps derived from EO data (Landsat dataset): an ensemble correlation analysis proved a positive correlation between the two variables.