Application of a Logit Model for Water well site location in fractured-bedrock aquifers in northeastern Brazil

Abstract

Wells drilling process in fractured-bedrock aquifers is a difficult task. Extreme variations in lithology and structure features, as well productive water zones sited at preferential points make geological and geophysical investigations difficult. To contribute to the understanding of the process of groundwater zones occurrence in fractured-bedrock aquifers, this paper develops a drilling prospective model by using regression analysis, whose parameters were calibrated according to the photogeological and cartographic analysis of 113 drilling points in the states of Rio Grande do Norte, Paraíba, and Ceará. The effectiveness of the resulting model was assessed through a sample of 43 additional drillings, which were carried out aiming at the distribution of water to the communities in the semiarid region of Brazil. The obtained results indicate the model as an important tool in the drilling process, with direct implications on the logistics costs of water well site location and consequent attendance to the population that needs the water.

A Robust Stochastic Programming Model for the Well Location Problem: The Case of The Brazilian Northeast Region

“…The probability of occurrence of groundwater 𝑝 𝑗 was estimated using Equation ( 1) from the prospective model developed by [27] using the same study area considered in this paper. The model proposed by [27] is based on logistic regression and considers the following independent variables: the lithotype (𝑙𝑖𝑡 𝑗 ); the lineament density (𝑙𝑖𝑛 𝑗 ); the surface drainage (𝑑𝑟𝑒𝑛 𝑗 ); and the distance from the closest structural lineament (𝑑𝑖𝑠𝑡 𝑗 ). The independent variables chosen to compose this equation are the result of a vast literature review and statistical analysis carried out by [27].…”

“…The model proposed by [27] is based on logistic regression and considers the following independent variables: the lithotype (𝑙𝑖𝑡 𝑗 ); the lineament density (𝑙𝑖𝑛 𝑗 ); the surface drainage (𝑑𝑟𝑒𝑛 𝑗 ); and the distance from the closest structural lineament (𝑑𝑖𝑠𝑡 𝑗 ). The independent variables chosen to compose this equation are the result of a vast literature review and statistical analysis carried out by [27]. Moreover, regarding the choice of the cut-off value of 50% for the composition of the subsets 𝑉 2 and 𝑉 3 , it was based on the definition proposed by [27] for the dependent variable of their prospective model in order to characterize it as a dichotomic variable.…”

“…The third subset V 3 contains the vertices that cannot be wells; in other words, vertices whose probability of occurrence of groundwater is less than 50%. The probability of occurrence of groundwater p j was estimated using Equation ( 1) from the prospective model developed by [27] using the same study area considered in this paper. The model proposed by [27] is based on logistic regression and considers the following independent variables: the lithotype (lit j ); the lineament density (lin j ); the surface drainage (dren j ); and the distance from the closest structural lineament (dist j ).…”

“…The independent variables chosen to compose this equation are the result of a vast literature review and statistical analysis carried out by [27]. Moreover, regarding the choice of the cut-off value of 50% for the composition of the subsets V 2 and V 3 , it was based on the definition proposed by [27] for the dependent variable of their prospective model in order to characterize it as a dichotomic variable. The arc set A contains all arcs connecting the vertices in V, including self-arcs (i, i) ∀i ∈ V. p j = e −5.943 + 0.853 × lit j + 0.537 × lin j + 0.101 × dren j + 2.739 × dist j 1 + e −5.943 + 0.853 × lit j + 0.537 × lin j + 0.101 × dren j + 2.739 × dist j (1)…”

“…The aim is to address both objectives of increasing the success rates in obtaining productive wells in hard rock terrains and prioritizing the communities most affected by drought. For the first objective, we use the model developed by [27], based on logistic regression, to consider the probability of occurrence of groundwater at the candidate points for locating a well. For community prioritization, we adopt the drought composite index (DCI) proposed by [23], a combination of both hazard and vulnerability.…”