Tarkegn et al. (2026) Comprehensive evaluations of gridded precipitation datasets across diverse climate zones of Brazos River Basin, Texas, USA

Identification

• Journal: Remote Sensing Applications Society and Environment
• Year: 2026
• Date: 2026-01-01
• Authors: Temesgen Gashaw Tarkegn, Ram L. Ray, Gebrekidan Worku Tefera
• DOI: 10.1016/j.rsase.2026.101947

Research Groups

• College of Agriculture, Food and Natural Resources, Prairie View A&M University, Prairie View, TX, USA

Short Summary

This study comprehensively evaluated four gridded precipitation datasets (Daymet, PRISM, IMERG, CHIRPS) against ground observations across diverse climate zones of the Brazos River Basin, Texas, USA, from 1998 to 2020, finding PRISM to be the most consistently accurate for various precipitation characteristics and temporal scales.

Objective

• To comprehensively evaluate the performance of four gridded precipitation datasets (Daymet, PRISM, IMERG, CHIRPS) against ground-based observations (GHCN) across diverse climate zones of the Brazos River Basin, Texas, USA, for precipitation occurrence, intensities, totals, and extremes across multiple temporal scales (daily to annual).

Study Configuration

• Spatial Scale: Brazos River Basin, Texas, USA, covering approximately 117,519 square kilometers, located between 29°20′ and 34°40′N latitude, and −104°10′ and −94°20′W longitude. The study focused on six of the nine climate zones within the basin: East Texas (ET), Edwards Plateau (EP), High Plains (HP), Low Rolling Plains (LRP), North Central (NC), and South Central (SC).
• Temporal Scale: 23 years, from 1998 to 2020, with evaluations performed at daily, monthly, seasonal (winter and summer), and annual scales.

Methodology and Data

• Models used:
  • Daymet (version 4 R1)
  • PRISM
  • IMERG (version 07, Final run)
  • CHIRPS (version 2.0)
• Data sources:
  • Reference Data: Global Historical Climate Network (GHCN) version 3 daily precipitation records from 19 selected ground-based meteorological stations. Missing data were imputed using the Multivariate Imputation by Chained Equations (MICE) method.
  • Gridded Precipitation Datasets:
    • Daymet: Gauge-based, 1 km spatial resolution.
    • PRISM: Gauge-based, 4 km spatial resolution.
    • IMERG: Satellite-based gauge-corrected, approximately 0.1° (~10 km) spatial resolution.
    • CHIRPS: Satellite-based gauge-corrected, approximately 0.05° (~5 km) spatial resolution.
  • Evaluation Methodology: A point-to-pixel approach was used to compare gridded data with station observations.
    • Precipitation Occurrence and Intensities: Assessed using categorical metrics (Probability of Detection [POD], False Alarm Ratio [FAR], Frequency Bias Index [FBI], Critical Success Index [CSI]) and probability density functions (PDFs) across eight WMO-defined intensity classes (threshold ≥1 mm for rainy event).
    • Precipitation Totals: Evaluated using continuous statistics (correlation coefficient [R], root mean squared error [RMSE], percent bias [PBIAS], and Kling-Gupta Efficiency [KGE]) at daily, monthly, seasonal, and annual scales.
    • Precipitation Extremes: Assessed using 10 extreme precipitation indices defined by the Expert Team on Climate Change Detection and Indices (ETCCDI), calculated using the RClimDex package.
    • Overall Performance Synthesis: A Comprehensive Rating Index (CRI) was applied to rank the datasets.

Main Results

• Overall Performance: PRISM consistently demonstrated the highest reliability across most climate zones for detecting precipitation occurrence, simulating intensities, estimating totals (daily to annual scales), and capturing most precipitation extremes.
• Precipitation Detection and Intensity:
  • PRISM outperformed other datasets in detecting precipitation occurrence in EP, HP, LRP, and SC zones, and showed the most robust performance in capturing dominant precipitation intensity patterns across most climate zones.
  • Daymet performed better in ET and NC zones for detection.
  • Daymet and CHIRPS generally performed poorly in precipitation detection. CHIRPS and IMERG were less effective in estimating intensity.
  • Most datasets exhibited an overestimation bias (FBI) for precipitation occurrence.
• Precipitation Totals:
  • Daily: PRISM was most accurate in EP, HP, LRP, and SC. Daymet was weakest in most zones but best in ET. IMERG and CHIRPS showed leading performance in NC.
  • Monthly, Winter, and Annual: PRISM consistently outperformed other products across all climate zones. Daymet ranked second for monthly totals. IMERG generally showed the lowest accuracy at monthly, winter, and annual scales.
  • Summer: PRISM was best in ET, LRP, NC, and SC. Daymet was best in EP and HP. IMERG and CHIRPS equally exhibited the weakest performance.
  • Bias: The direction and magnitude of estimation bias varied significantly by dataset, climate zone, and temporal scale.
• Precipitation Extremes:
  • PRISM showed the strongest performance for most extreme precipitation indices (CDD, CWD, R10mm, R20mm, R95p, RX1day, RX5day, PRCPTOT, SDII) across the majority of climate zones.
  • CHIRPS was generally the weakest performer for most extreme precipitation indices.
  • IMERG and CHIRPS demonstrated relatively weak performance for R99p.
  • Dataset performance varied by the specific type of extreme precipitation index and climate zone.

Contributions

• Comprehensive evaluation of gridded precipitation datasets across distinct climate zones within the Brazos River Basin.
• Assessment of these products across a full range of temporal scales, from daily to annual.
• Detailed analysis of the datasets' capability to capture 10 different precipitation extremes using a comprehensive set of evaluation metrics.
• Evaluation of the updated IMERG version 07 product across diverse climate zones.
• Provides valuable insights for water resource management, climate studies, and agricultural management in data-scarce regions of the Brazos River Basin and offers a transferable methodological framework for other regions globally.

Funding

• NASA (Award no. 80NSSC22K1781)

Citation

@article{Tarkegn2026Comprehensive,
  author = {Tarkegn, Temesgen Gashaw and Ray, Ram L. and Tefera, Gebrekidan Worku},
  title = {Comprehensive evaluations of gridded precipitation datasets across diverse climate zones of Brazos River Basin, Texas, USA},
  journal = {Remote Sensing Applications Society and Environment},
  year = {2026},
  doi = {10.1016/j.rsase.2026.101947},
  url = {https://doi.org/10.1016/j.rsase.2026.101947}
}