Abstract Accurate precipitation estimation is critical for climate modelling, hydrological analysis, and agricultural planning. However, ground-based precipitation measurements are often geographically sparse or temporally inconsistent, especially in underdeveloped nations. This study compares the performance of four widely used precipitation datasets, including APHRODITE, IMD gridded (gauge-based), ERA5 (reanalysis), and CHIRPS (satellite-based), with independent in situ rain gauge data from Tamil Nadu’s agriculturally critical Cauvery Delta region over 20 years (1986–2005). The datasets were evaluated on an annual, seasonal, and monthly basis using eight statistical performance indicators (Mean Bias Error, Correlation Coefficient, Percent Bias, Root Mean Square Error, Index of Agreement, Nash-Sutcliffe Efficiency, Kling-Gupta Efficiency, and Volumetric Efficiency) and ranked using the Compromise Programming Index. Results show that APHRODITE consistently outperforms others in replicating regional precipitation climatology across all temporal scales, despite a tendency to underestimate intensity. The IMD gridded and CHIRPS datasets demonstrated moderate reliability, with constant underestimation and minor overestimation, respectively, but ERA5 revealed the most significant variability and least dependability. These results are further confirmed with the Taylor diagram, which highlights the need for multi-timescale and multi-metric evaluation for region-specific applications. The work highlights APHRODITE’s potential to complement in situ observations in data-scarce areas, allowing for more precise planning for agricultural, water, and climate resilience in the Cauvery Delta and other monsoon-driven areas. Research highlights <UnorderedList Mark="Bullet"> <ItemContent> Comprehensive evaluation of four precipitation datasets (APHRODITE, IMD gridded, ERA5 and CHIRPS) against 13 rain-gauge stations in the Cauvery Delta over a twenty year period (1986–2005). </ItemContent> <ItemContent> Multi-metric and multi-timescale validation using eight continuous performance indicators and Compromise Programming Index (CPI) ranking for robust dataset comparison. </ItemContent> <ItemContent> APHRODITE showed highest accuracy across scales, followed by IMD gridded and CHIRPS with moderate performance, while ERA5 exhibited lowest agreement. </ItemContent> <ItemContent> APHRODITE is recommended as a reliable supplementary dataset for agricultural and hydrological applications, with IMD and CHIRPS suitable for operational use, and ERA5 requiring caution due to elevated uncertainty in monsoon-dependent regions. </ItemContent> </UnorderedList>

Evaluating the accuracy and applicability of precipitation datasets for sustainable agro-hydrological and climatic planning in Tamil Nadu’s Cauvery Delta

期刊论文
发表日期：

2025年11月28日

查看全文

Gunavathi Sundaram,
Selvakumar Radhakrishnan

摘要

Abstract

Accurate precipitation estimation is critical for climate modelling, hydrological analysis, and agricultural planning. However, ground-based precipitation measurements are often geographically sparse or temporally inconsistent, especially in underdeveloped nations. This study compares the performance of four widely used precipitation datasets, including APHRODITE, IMD gridded (gauge-based), ERA5 (reanalysis), and CHIRPS (satellite-based), with independent in situ rain gauge data from Tamil Nadu’s agriculturally critical Cauvery Delta region over 20 years (1986–2005). The datasets were evaluated on an annual, seasonal, and monthly basis using eight statistical performance indicators (Mean Bias Error, Correlation Coefficient, Percent Bias, Root Mean Square Error, Index of Agreement, Nash-Sutcliffe Efficiency, Kling-Gupta Efficiency, and Volumetric Efficiency) and ranked using the Compromise Programming Index. Results show that APHRODITE consistently outperforms others in replicating regional precipitation climatology across all temporal scales, despite a tendency to underestimate intensity. The IMD gridded and CHIRPS datasets demonstrated moderate reliability, with constant underestimation and minor overestimation, respectively, but ERA5 revealed the most significant variability and least dependability. These results are further confirmed with the Taylor diagram, which highlights the need for multi-timescale and multi-metric evaluation for region-specific applications. The work highlights APHRODITE’s potential to complement in situ observations in data-scarce areas, allowing for more precise planning for agricultural, water, and climate resilience in the Cauvery Delta and other monsoon-driven areas.

Research highlights

Comprehensive evaluation of four precipitation datasets (APHRODITE, IMD gridded, ERA5 and CHIRPS) against 13 rain-gauge stations in the Cauvery Delta over a twenty year period (1986–2005).

Multi-metric and multi-timescale validation using eight continuous performance indicators and Compromise Programming Index (CPI) ranking for robust dataset comparison.

APHRODITE showed highest accuracy across scales, followed by IMD gridded and CHIRPS with moderate performance, while ERA5 exhibited lowest agreement.

APHRODITE is recommended as a reliable supplementary dataset for agricultural and hydrological applications, with IMD and CHIRPS suitable for operational use, and ERA5 requiring caution due to elevated uncertainty in monsoon-dependent regions.