Search

A new pre-processing methodology for gridded Satellite Precipitation Products (SPPs) is developed to improve the performance of Machine Learning (ML) algorithms for runoff prediction. The developed approach was applied to capture the rainfall patterns, and to select relevant input data. This approach was tested using the FeedForward Neural Network (FFNN) and the Extreme Learning Machine (ELM) given their flexibility and ability in hydrological modelling. The methodology was tested in a semiarid transboundary watershed located in North Africa (Algeria, Tunisia) with the Integrated MultisatellitE Retrievals for Global Precipitation Measurement (GPMIMERG) and the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN) products. The results demonstrate the effectiveness of the proposed approach using all employed SPPs. In terms of Nash-Sutcliffe efficiency, the suggested pre-processing technique improved the prediction ability of FFNN by 13%, and of ELM by 15%, which highlights how pre-processing techniques significantly enhance ML models with SPP data. [ABSTRACT FROM AUTHOR]

Chronological series of monthly and annual precipitation data recorded in Gabes Watershed, south-eastern Tunisia, were analyzed. The study is based on the standardized precipitation index (SPI) values, computed for 10 rainfall stations over the period 1987–2012, which corresponds to an observatory period of 25 hydrologic years (from September to August). The results obtained show a great variability in SPI values. The historical evolution of the SPI made it possible to define the periods of excess and deficit, corresponding to wet and dry periods respectively. The wet years were found to be 1989–1990, 1995–1996 and 2006–2007 while the dry years were 1987–1988, 1996–1997, 2000–2001, 2001–2002, 2007–2008, 2008–2009 and 2009–2010. This clearly shows alternating wet and dry periods, but with drought episodes taking prevalence over rainy fronts throughout the study period. Indeed, a high tendency towards a drop in precipitation and important sequences of drought were observed. Spatial variability of drought throughout Gabes Watershed was examined by geostatistical analysis of SPI, as drought and rainfall distribution vary with latitude, longitude, topography and proximity to the Mediterranean Sea. The results obtained showed that, compared to coastal and southern areas, drought was observed to be more important in the West and the North of Gabes Watershed. The SPI showed that moderate droughts are generally more frequent than severe or extreme droughts in most of the Watershed.

Floods are considered among the gravest natural disasters worldwide and have resulted in enormous human and material damage. The Manouba–Sijoumi basin (Northeast of Tunisia) is often flooded due to urban expansion, population growth and unplanned land use. This study aims to identify and to define the flood-prone areas of this basin for the 2003 and 2018 extreme events based on a Geographic Information System, a qualitative method (analytic network process-ANP) and a statistical model (frequency ratio-FR). The flood risk maps obtained by both models were validated using the receiver operating characteristic, the area under the curve (AUC) and inventory map. Areas of high and very high flood sensitivity are located mainly in urban settings, with an increase in risk between 2003 and 2018. The AUC values for both models were of the same significance (98%) for the year 2003 while those for the year 2018 were 94% and 98% for the ANP and FR models, respectively. This would imply that both models yielded reasonable results. However, the FR model showed an ability to reduce the uncertainty associated with expert judgements. The results indicate that the most influential factor on flooding in this area was land use/cover. Indeed, populations were largely settled in unsuitable sites for urbanization and in potentially flood-prone areas located mainly around the Sijoumi Sebkha, especially to the west and south of it. The findings of the study are of great value for policy makers and state authorities to achieve greater awareness and adopt strategies for environmental preparedness and management. [ABSTRACT FROM AUTHOR]

Climate variability has been identified as one of the major factors affecting water resources. The purpose of this study is to characterize climate variability in the region of Souk Ahras and to evaluate its impact on surface and groundwater resources. Various statistical methods, including Mann-Kendall test, Pettitt test, drought indices, and Fourier filter were used. Pettitt test revealed the occurrence of breaks over the period (1994-1996), showing significant increase in the annual rainfall and temperature data time series. The analysis of drought indices provided insights into the relationship between meteorological and hydrological droughts in the study area. This mainly revealed that rainfall deficit results in an immediate response to runoff regime and a delayed response to groundwater level. Moreover, it was found that both meteorological and hydrological drought conditions significantly decreased after the climatic shift in terms of frequency, intensity and duration. Such information is of great importance for managing water resources and preparing drought mitigation measures. [ABSTRACT FROM AUTHOR]