Search

Romic, M.; Mornar, V.; Reljic, M.; Slapnicar, A.; Bagic Babac, M., and Romic, D., 2024. Utilizing artificial intelligence (AI) to develop soil health index for salinized coastal agricultural land. In: Phillips, M.R.; Al-Naemi, S., and Duarte, C.M. (eds.), Coastlines under Global Change: Proceedings from the International Coastal Symposium (ICS) 2024 (Doha, Qatar). Journal of Coastal Research, Special Issue No. 113, pp. 225-229. Charlotte (North Carolina), ISSN 0749-0208. The fertile soils found in delta regions globally are critical for food production, but the increase in sea level and the intrusion of saltwater into aquifers affect the soil health and agricultural sustainability. In the Eastern Adriatic coast in Croatia, particularly in its southern part, the shortage of water, severe droughts, and anomalous weather events in recent decades have become the main constraints on efficient agricultural productivity. The Neretva River delta has been converted to agricultural land through numerous and extensive reclamation projects resulting in the creation of polder-type agricultural land. Generally, the surface-water levels within polder-type land are artificially maintained and controlled by a network of drainage canals, pumping stations, and sluice-gate complexes, which are used to release water into the sea. The potential of AI to assist in developing informed management practices in promoting soil health in agriculture was tested using data of long-term soil and water quality monitoring. A sophisticated sensor system has been developed, providing continuous collection of data on soil and water salinity at hourly intervals, with more frequent updates occurring every ten minutes in select locations. This data is seamlessly transmitted to a centralized database, serving as a robust storage solution for comprehensive data analysis using AI. Through the collection and analysis of data, machine learning models were developed, implemented, and evaluated to identify the most effective model. In addition, ongoing monitoring allows the models to continuously learn and improve over time as new data becomes available. Optimal AI models should help farmers to adjust management practice, particularly irrigation, in accordance with the current soil and water quality condition. [ABSTRACT FROM AUTHOR]

The vine-growing region of Lumbarda, located in the southeastern part of the island of Korcula, Croatia, has deep sandy soils (paleodunes) associated with the Mediterranean climate and provides optimal conditions for cultivating the autochthonous vine variety Vitis vinifera L. cv. Grk. Unfortunately, recently growers have noted declining yield and quality of grapes resulting from inadequate vineyard management (particularly unsustainable management of soil organic matter) and the occurrence of viral diseases. To revitalize and maintain the vineyards in the Lumbarda region, a detailed study was carried out involving intensive soil survey, analysis of grapevine nutrient status, positive clonal selection, and establishment of new vineyards with virus-free plant material. Preliminary soil survey results provided insight into the extent of variability in major physical and chemical soil characteristics. Based on these results, eight markedly different vineyards were selected for detailed analysis of grapevine nutrients [boron (B), calcium (Ca), copper (Cu), iron (Fe), potassium (K), magnesium (Mg), manganese (Mn), phosphorus (P), and zinc (Zn)] at key development stages to detect potential imbalances that may hamper vine growth and affect its production. Leaf blades and petioles were sampled at midflowering and midveraison stages during the 2007–8 growing season from each of the selected vineyards. The greatest variability in nutrient content between the study sites was recorded for Cu, and considerable variability was also determined for B, Mn, Mg, and P. Differences in fertilization practices are the most common reasons for this variability. The results point to B deficiency in some of the vineyards and suggest potentially toxic levels of Cu in others. [ABSTRACT FROM PUBLISHER]

Abstract: The assessment of soil contamination and location of pollution sources represent a crucial issue in soil remediation. Topsoil samples were collected in the Zagreb area (Northwest Croatia) and the total contents of trace and major elements were determined. A multivariate geostatistical analysis was used to estimate soil chemical composition variability. Factorial Kriging Analysis (FKA) was used to investigate the scale-dependent correlation structure of some variables by modelling co-regionalization of ten chemical variables, co-kriging specific factors and mapping them. The FKA provided two regionalized factors at different spatial scales of variability: the first factor at shorter range for Zn, Pb, Cd, Cu and Ni indicated different sources of anthropogenic contamination, whereas Ca (mainly loading on the longer range factor) was related to the lithology and parent material composition. The methodology used has proved to be a useful tool to separate geological and anthropogenic causes of variation in soil heavy metal content and to identify common pollution sources. [Copyright &y& Elsevier]