Vegetation dynamics and their relationships with precipitation in Africa for drought monitoring purposes

Drought affects more people than any other natural disaster and results in serious economic, social and environmental costs. The development of effective drought monitoring and early warning has been a significant challenge because of the unique characteristics of drought. In fact, considering the multifaceted nature of drought phenomena (i.e. hydrological, meteorological, and agricultural), a comprehensive and integrated approach is required to define effective Early Warning Systems (EWS), which are thus based on the monitoring of different drought-related parameters and complex drought indicators. In such a context, several studies have shown how temporary changes of vegetation indices and their anomalies are strongly correlated with precipitations, especially in arid and semi-arid environments. Besides, satellite-derived vegetation indicators and climatic data have been widely used to study and monitoring droughts and included in the main existing EWS developed by the international community (e.g. global systems, such as US-AID FEWSNET, JRC MARS FOODSEC, FAO GIEWS, or designed for a specific area of interest, as in the case of MESA South Africa Drought Monitoring, the US Drought Monitor, and the JRC European Drought Observatory). In this work, a study aimed at investigating spatial and temporal vegetation dynamics in the whole Africa and their relationships with climate factors, considering as a base data long-term time-series of vegetation-related phenological parameters is proposed. The outcomes of this study have been used in order to define proper drought monitoring procedures to be used by ITHACA (Information Technology for Humanitarian Assistance, Cooperation and Action) for early warning purposes. In fact, in recent years, through its partnership with the World Food Programme (WFP), ITHACA has focused its efforts to develop an automated drought EWS, based on the monitoring of relevant environmental variables that allow the early detection of vegetation stress patterns and agricultural drought phenomena on a global scale, finally providing near real-time alerts about vegetation conditions and productivity. In particular, the fortnightly monitoring of satellite-derived vegetation indexes during growing seasons allows the early detection of water stress conditions of vegetation, and the assessment of derived phenological parameters. These parameters, coupled with the evaluation of precipitation conditions, allow the near real-time assessment of the vegetation productivity which can be expected at the end of the considered growing season. The timely detection of critical conditions in vegetation health and productivity, during a vegetation growing season, leads to the identification of the agricultural areas where crop failures are likely to occur. Finally, the proposed system incorporates a simplified drought vulnerability model, able to show food security conditions starting from the hazard situation evaluated in near real-time. The system outputs and information related to identified alerted areas are updated fortnightly and disseminated using a proper web display application. The described study has been conducted using time-series of vegetation phenological parameters extracted from satellite-derived NDVI datasets (global 15-day NDVI time-series, available from 2000 to present, at a 5.6 km spatial resolution, derived from the MODIS MOD13C1 Terra CMG dataset), and precipitation time-series obtained from the Tropical Rainfall Measuring Mission TRMM mission (0.25° x 0.25° spatial resolution) Multisatellite Precipitation Analysis estimation, computed at daily intervals (TRMM 3B-42 daily data), for period of 1998-present. For the purpose of the proposed statistical analysis, ten phenological metrics (the time for the start and the end of the season, the length of the season, the season base level, the time for the mid of the season, the largest NDVI data value during the season, the seasonal amplitude, the rate of increase at the beginning of the season and the rate of decrease at the end of the season, and, finally, the seasonal integral) have been extracted from the yearly NDVI function that best fits the original yearly NDVI time-series and considered for each vegetation growing season in the examined time interval (2000-2014). These metrics are able to describe synthetically the trend of the season in both the time and the integrated NDVI/time domains and are related to the seasonal vegetation productivity. Different precipitation fortnightly time-series have been used for the study, obtained taking into consideration different cumulating intervals (1-3-6-9-12 months values). Specific routines have been implemented in order to investigate, on a pixel basis, and to explain the statistical relationship between the considered time-series of phenological parameters and precipitation data. Obtained results have been spatially analyzed and aggregated taking into consideration different vegetation types, and maps showing the areas where the observed vegetation phenological parameters are largely dependent on rainfall patterns have been produced. Moreover, the precipitation cumulative interval and the period, in the year, when precipitation influence on vegetation productivity has proved to be significant, have been identified and discussed, also in relation to the rainfall seasonality and crop calendar in the examined area. The monitoring of vegetation conditions based on the analysis of phenological metrics, as originally provided in the ITHACA drought EWS, proved to effectively support WFP activities in several cases (i.e. Niger and Chad 2009, Sahel 2012, Horn of Africa crisis 2011). The final aim of conducted statistical study, object of this thesis work, was to correctly define the operational use of precipitation data for drought detection, in support to the vegetation monitoring procedures. The outcomes of the carried out work supported the planning and definition of effective procedures for the integration, where it is meaningful, in the ITHACA vegetation conditions monitoring activities , based on the analysis of phenological parameters, with the near real-time evaluation of precipitation deficits explained, for multiple time scales, using the Standard Precipitation Index (SPI). Indeed, the studied relationships between rainfall and vegetation dynamics allowed to determine the areas where the spatial and the temporal variability in vegetation conditions are closely related to the climate, and the best rainfall cumulating interval to be used for SPI monitoring purposes as well. In these areas, the fortnightly near real-time monitoring of the precipitation permits to earlier identify drought warnings, by considering also climate conditions before the start of the vegetation growing season. Moreover, in the same areas, the near real-time SPI analysis during the vegetation growing season supports the monitoring of phenological parameters in a way to identify very critical events characterized by both vegetation productivity and rainfall anomalies.