Using statistical and machine learning tools to assess the feasibility of reuse secondary-treated wastewater for irrigation agriculture in Algerian drylands

This paper assesses the feasibility for secondary-treated wastewater reuse in irrigation agriculture of Algerian inland drylands, selecting the Arris activated-sludge conventional wastewater treatment plant (WWTP) to document practical extents. Over the period 2019–2023, 240 weekly paired samples of raw (RW) and treated (TW) wastewater were analysed for eleven physicochemical parameters, including temperature (T), pH, electrical conductivity (EC), dissolved oxygen (DO), chemical and 5-day biochemical oxygen demand (COD and BOD 5 ), suspended solids (SS), total nitrogen (TN), nitrate (NO 3 ), phosphate (PO 4 ), and ammonium (NH 4 ). Treatment removal efficiency was generally higher than 90% for all compounds. However, the treated effluent exhibited concentrations of certain dissolved inorganic nutrients, which frequently exceeded the limit of international regulations for TW reuse in irrigation agriculture, especially during the dry season. An ensemble of multivariate statistical and machine learning methods revealed fluctuations in organic load and nutrient concentrations in TW, seasonal environmental variations in T and oxygen availability that influence biological processes during the treatment, and suboptimal quality of TW to be reused in irrigation agriculture. Given the intensifying aridity and escalating demand for irrigation water in the region, and to comply with regulation limits, it is recommended that public health and the environment are safeguarded by the orderly blending of secondary TW with conventional water sources and/or new tertiary nutrient-removal treatments. The methodology introduced also seeks to cover common limitations to confidently reuse TW for irrigation agriculture in other drylands with lower data density. • Treated wastewater reuse in irrigation agriculture of inland drylands was assessed. • Statistical and machine learning methods were applied to forecast wastewater issues. • Dissolved inorganic nutrients exceeded regulation limits for wastewater reuse. • Mix treated wastewater with conventional sources or new tertiary treatments are suggested. • Our trained methods can assess reuse feasibility in other drylands with suboptimal data.