Satellite-Derived Environmental Change Profiles for Three Malagasy Regions (2000–2024)

© 2025 — Public archive on Zenodo: https://doi.org/10.5281/zenodo.17838200 Satellite-Derived Environmental Change Profiles for Three Malagasy Regions (2000–2024) A 25-Year Satellite-Derived Assessment of Environmental Conditions Zo Rivomanana Rasoanaivo, University of Antananarivo—2025 0. Executive Summary This brief provides a satellite-derived assessment of environmental change across three regions of Madagascar—Ranobe (West), Andringitra (Center), and Maromizaha (East)—using long-term indicators of surface temperature (LST), rainfall (CHIRPS), vegetation greenness (NDVI), and surface moisture (NDWI). Results reveal a consistent warming signal across all sites, with a statistically significant LST increase at Andringitra and comparable but less certain trends at Maromizaha and Ranobe. Rainfall trajectories show weak and non-significant tendencies, though Ranobe displays an indicative drying pattern. Vegetation greenness remains broadly stable over the 2000–2024 period, while NDWI data limitations prevent robust interpretation of moisture trends. Cross-regional comparison highlights a clear environmental gradient: Ranobe exhibits multiindicator symptoms of increasing stress, Andringitra maintains stability, and Maromizaha remains productive with early signs of change. These findings, while constrained by varying data availability and confidence levels, underscore the value of multi-sensor monitoring for anticipating spatially differentiated environmental trajectories. 1. Background Understanding long-term environmental trajectories is essential for anticipating ecological stress, prioritizing restoration, and strengthening environmental monitoring. This brief synthesizes 25 years of satellite-derived eco-hydrological indicators — rainfall (CHIRPS), land surface temperature (LST), vegetation greenness (NDVI), and surface moisture (NDWI) — across three contrasting Malagasy landscapes: Table. Strategic Pressures & Ecological Profiles Across the Three Sites Site Ecological Profile Major Projects / Emerging Pressures Ranobe PK32 (West) Dry coastal mosaic ecosystem Base Toliara (ilmenite mining), RN9–RN10 economic corridor, solar mini-grids, expanding coastal tourism Andringitra (Center) Highland mixed grass–forest landscapes RN25 East corridor rehabilitation, REDD+ & conservation programs, mid-scale hydropower prestudies, climate-resilient agriculture initiatives Maromizaha (East) Evergreen humid forest corridor Tana–Tamatave highway (indirect influence area), CAZ corridor strengthening, hydrometeorological stations rollout, ecotourism & research hub These regions represent key ecological gradients from dry sub-arid systems to humid evergreen forests. © 2025 — Public archive on Zenodo: https://doi.org/10.5281/zenodo.17838200 2. Data and Methods For each site, monthly satellite measurements (2000–2024) were aggregated into annual indicators: • Rainfall: CHIRPS daily values averaged spatially → annual totals. • LST: MODIS land surface temperature → annual mean. • NDVI: vegetation greenness index → annual mean. • NDWI: surface moisture index → annual mean. The analysis focuses on directional trends, relative differences among sites, and signatures of environmental pressure. 3. Key Findings 3.1 Rainfall (CHIRPS) • Ranobe (West) shows a marked decline, with an estimated –64 mm per decade. • Maromizaha (East) exhibits moderate interannual variability and a weak downward trend. • Andringitra (Center) remains relatively stable over the 25-year period. Implication: Western drylands are becoming progressively drier, increasing exposure to vegetation stress and reducing natural regeneration. 3.2 Land Surface Temperature (LST) • All three landscapes show a consistent upward trend. • Warming is slightly stronger in the West, aligned with regional aridification. • The Center and East also exhibit persistent warming, though at a slower rate. Implication: Rising surface temperatures reinforce evapotranspiration, moisture deficits, and vegetation vulnerability. 3.3 Vegetation Greenness (NDVI) • Maromizaha (East) maintains high NDVI but shows mild long-term decline. • Andringitra (Center) is stable, consistent with mixed grassland–forest systems. • Ranobe (West) records significant decline, confirming reduced vegetative cover. Implication: The vegetation productivity gradient is widening, with the West undergoing structural degradation. 3.4 Surface Moisture (NDWI) • NDWI decreases across all sites, but the decline is most pronounced in Ranobe. • The East maintains higher baseline moisture but shows gradual drying, particularly in recent years. Implication: Soil and canopy moisture levels are declining broadly, increasing susceptibility to drought stress. 3.5 Cross-Regional Interpretation Across the three landscapes, a clear pattern emerges: • The West is undergoing coupled drying, warming, and vegetation decline. • The Center remains stable but sensitive to climatic fluctuations. • The East shows resilience but signals early signs of moisture stress. These combined indicators highlight ecosystems undergoing differential environmental pressure, with the West as the most vulnerable. © 2025 — Public archive on Zenodo: https://doi.org/10.5281/zenodo.17838200 4. Discussion and Limitations 4.1. Statistical Significance Trend diagnostics highlight a consistent signal of surface warming, with LST exhibiting a statistically significant increase at Andringitra of approximately 0.30 °C per decade (95% CI: 0.04–0.55 °C, p ≈ 0.02, R² = 0.20). Maromizaha and Ranobe also show positive warming trends, though the evidence is weaker and not statistically significant (p ≈ 0.09–0.13). In contrast, CHIRPS rainfall trends remain weak and non-significant across all sites (confidence intervals crossing zero, p ≫ 0.05), although Ranobe suggests a moderate drying signal (−64 mm/decade, p ≈ 0.06) that should be interpreted as indicative rather than conclusive. NDVI trends do not show statistical significance (high p-values, near-zero R²), suggesting an overall stability of vegetation greenness from 2000 to 2024 despite interannual variability. Finally, NDWI diagnostics rely on only 2–3 years of data per site, which precludes any robust assessment of surface moisture trends and should be treated with caution. 4.2 Uncertainties and Data Quality Considerations Several sources of uncertainty influence the interpretation of the satellite-derived indicators used in this assessment. First, statistical confidence varies across variables, with LST trends showing clearer significance while rainfall, NDVI, and NDWI exhibit weak or non-significant trends; confidence intervals frequently cross zero, indicating high interannual variability relative to long-term directional change. Second, data availability is uneven across indicators: NDWI, in particular, is limited to only 2–3 years of observations per site, which prevents any robust inference on multi-decadal moisture trends. Third, sensor-related and methodological uncertainties—including differences in spatial resolution, cloud contamination in optical data, and the aggregation of monthly values into annual means—may smooth short-term variability or amplify anomalies. Rainfall retrievals from CHIRPS can also introduce biases in dry coastal regions, where station density is historically low. Finally, although time series appear internally consistent, site-level spatial footprints remain simplifications of highly heterogeneous landscapes, and localized disturbances (e.g., microclimates, land-use changes) may not be fully captured at the satellite scale. These uncertainties do not invalidate the observed tendencies but highlight the importance of cautious interpretation and the need for continued multi-sensor, long-term environmental monitoring. 5. Strategic Implications (Non-Prescriptive) (These points remain strictly scientific and non-normative.) • Long-term monitoring should prioritize western drylands, where all variables converge toward degradation. • Central highlands show stability and offer a reference for understanding resilience mechanisms. • Eastern humid forests remain productive but require continued observation to detect threshold changes. • Multi-indicator approaches (LST + NDVI + NDWI + Rainfall) are essential to detect early ecosystem stress signals. • Satellite-based analytics provide scalable, consistent baselines for environmental assessment. © 2025 — Public archive on Zenodo: https://doi.org/10.5281/zenodo.17838200 6. Conclusion Over 2000–2024, satellite eco-hydrological indicators reveal diverging environmental trajectories across Madagascar’s West–Center–East gradient. The dataset highlights drying trends, warming, and vegetation decline, particularly in the West, while central and eastern regions show varying levels of resilience. These results reinforce the value of systematic, long-term, multi-sensor monitoring for understanding Madagascar’s environmental dynamics and supporting science-based environmental planning. © 2025 — Public archive on Zenodo: https://doi.org/10.5281/zenodo.17838200 7. Time Series of LST, NDVI, NDWI and Rainfall (2000–2024) © 2025 — Public archive on Zenodo: https://doi.org/10.5281/zenodo.17838200 8. Site-Level Environmental Statistics (NDVI, NDWI, LST, CHIRPS) © 2025 — Public archive on Zenodo: https://doi.org/10.5281/zenodo.17838200 9. Significance and Interpretation of Trend Diagnostics