About this Scenario: Drought Early Warning System
Background
Drought is a slow-onset natural hazard that can severely affect water resources, agriculture, ecosystems, and human livelihoods. Madagascar is highly vulnerable to recurring drought events, particularly in the southern regions, which face significant challenges in food security and water availability.
Anticipatory action is key to mitigating the impacts of drought, allowing for timely interventions such as water resource management, agricultural planning, and humanitarian aid deployment. Earth Observation (EO) data provides critical insights to monitor drought dynamics and support early warning systems.
How to Use this Scenario
The Drought Early Warning System scenario provides an interactive map to visualize drought risk across Madagascar, based on multi-component anomaly detection (precipitation, soil moisture, and vegetation condition).
You can explore the scenario by:
- Drought Risk: select and view the integrated drought risk layer for the region.
- Area Analysis: click on any administrative area to view
temporal anomalies (z-scores) of the three drought components: rainfall,
soil moisture (SWI), and vegetation (NDVI). These z-scores indicate how
many standard deviations each measurement deviates from its historical
average:
- • Z-score = 0: Values are at the historical average
- • Z-score > 0: Values are above average (wetter conditions)
- • Z-score less than 0: Values are below average (drier conditions)
- • Z-scores beyond ±2 indicate extreme conditions
- Time Selection: use the date selector to choose a specific dekad (10-day period) and observe drought risk patterns for that interval. A dekad is a 10-day period, starting on the 1st, 11th, 21st, etc. of the month.
- Affected Crops: explore the affected crops by drought in a concrete dekad.
- Population Exposure: visualize population density in relation to hazard zones.
- Relative Wealth Index: visualize the Relative Wealth Index of the region, to identify areas with lower wealth and therefore higher vulnerability to drought.
Use the layer selection menu and dekad selector to explore how drought risk evolves over time and across different areas of Madagascar.
About the Scenario
This scenario demonstrates a Drought Early Warning System for Madagascar based on multi-component anomaly detection. The platform integrates three complementary EO-based variables to monitor drought conditions at the level of administrative units:
- ERA5-Land Precipitation – high-resolution precipitation data from the Copernicus Climate Data Store.
- Soil Water Index (SWI) V3.0 – soil moisture estimates derived from ASCAT radar observations.
- NDVI300 – 10-day composites of the Normalised Difference Vegetation Index based on Sentinel-3 data.
The methodology applies anomaly detection through temporal standardization (z-score calculation) of each variable, comparing current observations to historical climatologies. By integrating precipitation, soil moisture, and vegetation condition, the system enables robust classification of drought conditions at three alert levels:
- Watch: Early signal of potential drought onset based on precipitation anomaly.
- Warning: Confirmed soil moisture deficit following a precipitation anomaly.
- Alert: Vegetation stress consistent with prior hydro-meteorological anomalies.
This multi-indicator approach, adapted from the methodology of Sepulcre-Canto et al. (2012), provides an operational framework for timely drought monitoring and early warning.
For each dekad where drought risk has been detected, the crops affected by drought have been calculated. Cropland areas are identified using the ESA WorldCover 2021, and only those zones categorized as cropland are considered for this analysis. This approach enables the estimation of agricultural areas impacted by drought conditions.
In addition, population exposure is assessed using data from the Copernicus Human Settlement Layer, helping users understand the potential human impact of drought.
Last, the Relative Wealth Index predicts the relative standard of living of each country based on privacy-protected connectivity data, satellite imagery, and other novel data sources. It can be used to identify areas with lower wealth and therefore higher vulnerability to drought. It is assessed using the dataset from The Humanitarian Data Exchange.
Objective
The objective of this scenario is to demonstrate how a Drought Early Warning System can be built using EO data from the EU Space Programme. By integrating precipitation, soil moisture, and vegetation condition, the scenario offers a comprehensive view of drought risk across Madagascar.
The platform supports anticipatory humanitarian action by enabling stakeholders to identify evolving drought conditions, prioritize areas for intervention, and inform early response strategies.
Credits
Administrative boundaries are provided by GADM.
All of the displayed basemaps are provided by © MapTiler © OpenStreetMap contributors