ABSTRACT:

We present the first comprehensive, open-access groundwater database for this region, developed by applying a novel workflow that integrates Optical Character Recognition (OCR) with Large Language Models (LLMs). This methodology systematically extracted and unified data from 3,675 previously unstructured 'gray literature' sources, including government reports and academic theses, spanning the period 1966–2024. The resulting database consolidates 4,775 multi-parameter data points—encompassing variables such as groundwater level, well elevation, water level (m a.s.l.), hydraulic conductivity, aquifer thickness, transmissivity, storage coefficient, specific yield, methodology used, and lithology, among others—and is augmented with 931 depth-to-water records from Peru's National Water Resources Information System (SNIRH). A rigorous three-phase quality assurance protocol, comprising 100% manual verification, statistical outlier screening, and spatial validation, was applied to ensure database accuracy. Designed to overcome data fragmentation, it enables trend analysis, groundwater modeling, and sustainable yield assessments. Publicly available via HydroShare with code on GitHub.

------------------------------------------

📌 Note on Dataset Versions:

- The file 'arequipa_final_gw_database.cvs' includes all the above data PLUS additional records sourced from the public database at [snirh.ana.gob.pe](https://snirh.ana.gob.pe), specifically for the following districts:
Apurímac, Arequipa, Ayacucho, Cusco, Ica, Moquegua, Puno, and Tacna.

- The file 'arequipa_final_gw_database_per_basin.xlsx' includes all of the aforementioned data, along with additional records sourced from the public database at snirh.ana.gob.pe. To facilitate easier classification and analysis, the workbook is organized so that each sheet corresponds to a specific basin.

➤ This extended version is intended for broader analysis and cross-validation with official national hydrological records.

ABSTRACT:

We present the first comprehensive, open-access groundwater database for this region, developed by applying a novel workflow that integrates Optical Character Recognition (OCR) with Large Language Models (LLMs). This methodology systematically extracted and unified data from 3,675 previously unstructured 'gray literature' sources, including government reports and academic theses, spanning the period 1966–2024. The resulting database consolidates 4,775 multi-parameter data points—encompassing variables such as groundwater level, well elevation, water level (m a.s.l.), hydraulic conductivity, aquifer thickness, transmissivity, storage coefficient, specific yield, methodology used, and lithology, among others—and is augmented with 931 depth-to-water records from Peru's National Water Resources Information System (SNIRH). A rigorous three-phase quality assurance protocol, comprising 100% manual verification, statistical outlier screening, and spatial validation, was applied to ensure database accuracy. Designed to overcome data fragmentation, it enables trend analysis, groundwater modeling, and sustainable yield assessments. Publicly available via HydroShare with code on GitHub.

------------------------------------------

📌 Note on Dataset Versions:

- The file 'arequipa_final_gw_database.cvs' includes all the above data PLUS additional records sourced from the public database at [snirh.ana.gob.pe](https://snirh.ana.gob.pe), specifically for the following districts:
Apurímac, Arequipa, Ayacucho, Cusco, Ica, Moquegua, Puno, and Tacna.

- The file 'arequipa_final_gw_database_per_basin.xlsx' includes all of the aforementioned data, along with additional records sourced from the public database at snirh.ana.gob.pe. To facilitate easier classification and analysis, the workbook is organized so that each sheet corresponds to a specific basin.

➤ This extended version is intended for broader analysis and cross-validation with official national hydrological records.