D’Ercole et al. (2025) Using daily vegetation and precipitation products to study drought events in the Horn of Africa

Identification

• Journal: International Journal of Applied Earth Observation and Geoinformation
• Year: 2025
• Date: 2025-10-23
• Authors: Riccardo D’Ercole, Daniele Casella, Paolo Sanò
• DOI: 10.1016/j.jag.2025.104837

Research Groups

• National Research Council of Italy, Institute of Atmospheric Sciences and Climate (CNR-ISAC), Rome, Italy
• University of Naples Federico II, Naples, Italy

Short Summary

This study assesses the capability of high-frequency daily Earth observations (vegetation and precipitation) to detect and monitor meteorological and agricultural drought events in the Horn of Africa, revealing the benefits of daily resolution for capturing short-term wet-dry spells and identifying optimal precipitation products for the region.

Objective

• Assess the capability of high-frequency daily Earth observations (vegetation and precipitation) in detecting meteorological and agricultural drought events in Ethiopia, Somalia, Kenya, and Djibouti.
• Explore the benefits of daily temporal resolution data for monitoring drought dynamics, including short-term wet-dry spells.
• Evaluate the suitability of various daily precipitation products for describing rainfall phenomena in data-sparse regions.
• Investigate the impact of different accumulation periods (30, 60, 90, 180 days) for the Standard Precipitation Index (SPI) on drought detection.
• Analyze the heterogeneous response of different soil types to water stress during drought events.

Study Configuration

• Spatial Scale: Horn of Africa (Ethiopia, Somalia, Kenya, Djibouti).
• Temporal Scale: Daily resolution over the period 2005–2020.

Methodology and Data

• Models used:
  • Normalized Difference Vegetation Index (NDVI) derived from geostationary satellite data.
  • Whittaker filter for NDVI time series reconstruction.
  • Bidirectional Reflectance Distribution Function (BDRF) method for atmospheric correction.
  • SEVIRI cloud mask product for cloud removal.
  • Vegetation Condition Index (VCI).
  • Standard Precipitation Index (SPI) calculated using a gamma distribution at 30, 60, 90, and 180-day accumulations.
• Data sources:
  • Vegetation: Meteosat Second Generation-Spinning Enhanced Visible and Infrared Imager (MSG-SEVIRI) radiometer (daily, 12:12 UTC image).
  • Precipitation:
    • Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) v2 (daily, 0.05° spatial resolution, from 1981).
    • Global Precipitation Climatology Centre (GPCC) (daily, 1° × 1° spatial resolution, from 1891) - used as reference.
    • Tropical Applications of Meteorology using Satellite Data and Ground-based Observations (TAMSAT) (daily, 4 km spatial resolution, from 1983).
    • ECMWF Reanalysis v5 (ERA5) (hourly converted to daily, 0.25° spatial resolution, from 1940).
    • Global Precipitation Mission, Integrated Multi-satellite Retrievals for GPM (GPM-IMERG) (daily, fuses TRMM and GPM data).
    • Multi-Source Weighted-Ensemble Precipitation (MSWEP) (daily, 0.1° resolution, from 1979).
  • Land Cover: ESA Land Cover 100 m product (Buchhorn et al., 2020), resampled to SEVIRI spatial resolution.

Main Results

• MSWEP demonstrated the highest accuracy (80.34%) and Probability of Detection (POD) (57.72%) with the lowest RMSE (0.324 mm/day) for detecting daily rain events (≥ 1 mm/day) compared to the GPCC reference. ERA5 was the second-best performer.
• CHIRPS and TAMSAT products showed the lowest POD (37.5% and 41.69%, respectively) and a pronounced dry bias, frequently missing actual rainfall events at daily resolution.
• The Standard Precipitation Index (SPI) calculated with a 90-day accumulation (SPI-90) exhibited the strongest average correlation with NDVI (approximately 33%), making it the most suitable for monitoring vegetation response to drought.
• Longer SPI accumulation periods (e.g., SPI-180) provided greater robustness to wet–dry fluctuations but introduced a delay in drought detection, while shorter periods (e.g., SPI-60) could anticipate drought onset by approximately 15 days but were more sensitive to rapid transitions.
• Analysis of the 2009–2010 drought revealed heterogeneous impacts across soil types: cropland and forest soils showed greater resilience to short-term water stress, whereas shrublands, herbaceous vegetation, and sparse vegetation were more severely affected, particularly during their green-up phases.
• The use of daily temporal resolution data enabled the identification of sub-monthly precipitation events and rapid dry–wet alternations that are often obscured in coarser, monthly datasets, providing earlier indicators of vegetation stress.

Contributions

• Demonstrates the significant value of daily temporal resolution Earth observation data (both vegetation and precipitation) for drought monitoring in the Horn of Africa, particularly for detecting rapid dry-wet alternations often missed by monthly products.
• Develops and utilizes a novel, cloud-reduced daily NDVI time series from geostationary MSG-SEVIRI satellite imagery, enhancing the accuracy and timeliness of vegetation health assessment.
• Provides a comprehensive qualitative assessment and inter-comparison of several global and regional daily precipitation products (MSWEP, ERA5, IMERG, CHIRPS, TAMSAT, GPCC) for drought monitoring in the region, identifying MSWEP and ERA5 as superior for daily rainfall detection and highlighting limitations of CHIRPS and TAMSAT at this resolution.
• Investigates the optimal accumulation periods for the SPI in relation to vegetation response, showing SPI-90 as most correlated with NDVI anomalies and discussing the trade-offs between timeliness and robustness for different SPI accumulation windows.
• Analyzes the heterogeneous impacts of drought events across different soil types, revealing varying resilience and vulnerability, which is crucial for targeted humanitarian and agricultural interventions.

Funding

• EUMETSAT, Germany Satellite Application Facility for Operational Hydrology and Water management (H SAF) Fourth Continuous Development and Operations Phase (CDOP-4).
• Italian National PhD Program in Artificial Intelligence ‘‘AI & agrifood and environment’’ pillar.

Citation

@article{DErcole2025Using,
  author = {D’Ercole, Riccardo and Casella, Daniele and Sanò, Paolo},
  title = {Using daily vegetation and precipitation products to study drought events in the Horn of Africa},
  journal = {International Journal of Applied Earth Observation and Geoinformation},
  year = {2025},
  doi = {10.1016/j.jag.2025.104837},
  url = {https://doi.org/10.1016/j.jag.2025.104837}
}