Advancing EO-Based Rapid Estimation of Flood and Drought Impacts on Rice in the Philippines
A published paper in the International Journal of Disaster Risk Reduction presents an Earth Observation (EO)–based workflow for rapidly estimating rice production losses and damages from floods and droughts, from pixel scale up to municipal and provincial summaries. The study is co-authored by Arnan Araza (IMPACT R&D) and Elmer Alosnos (Philippine Rice Research Institute), combining satellite-derived hazard information with crop-stage sensitivity and uncertainty analysis to support more timely and verifiable post-disaster estimates.
Key takeaways
The EO-based framework produced event-level loss estimates with uncertainty bounds:
Typhoon Odette (Rai) flood case (Iloilo): PHP 270.9 ± 0.8 million
2024 drought case (Iloilo): PHP 390.0 ± 0.3 million
Estimated hazard exposure differed by event:
Flood affected 9.3% of rice area (case study coverage).
Severe agricultural drought conditions were observed over 37.6% of rice area (under the study’s severity classification).
EO-derived losses were lower than consolidated provincial reports in both cases:
Flood: PHP 270.9M (EO) vs PHP 692.3M (provincial report)
Drought: PHP 390.0M (EO) vs PHP 706.6M (provincial report)
A modeled 1:100-year flood scenario yielded much larger potential losses:
PHP 1,694.2 ± 1.8 million (≈ PHP 1.694B), reflecting the much broader inundation extent.
The analysis identified multi-hazard hotspot municipalities with notable overlap of flood- and drought-exposed rice areas, supporting targeted resilience planning.
Background
In the Philippines, floods and droughts are recurring threats to rice production. Post-disaster loss and damage (L&D) estimates are critical for mobilizing assistance (e.g., calamity support, recovery programming, and insurance processes). However, administrative reporting can be delayed and may vary in precision across locations. EO-based methods offer a path toward faster, spatially explicit, and more comparable estimation—especially when uncertainty is reported transparently.
Approach and case studies
The study integrates four components into a single operational workflow:
Hazard mapping
Flood extent and flood depth mapping (including depth uncertainty)
Agricultural drought severity mapping using EO-based vegetation indicators
Exposure and crop timing
Spatial rice area information and rice phenology (growth stage) to reflect stage-specific vulnerability
Damage translation
Stage-specific damage curves to convert hazard exposure into yield loss estimates
Uncertainty propagation
Monte Carlo simulation to generate loss estimates with uncertainty bounds (rather than single deterministic numbers)
The workflow is demonstrated in Iloilo Province using flood impacts associated with Typhoon Odette (Rai), and the 2024 drought event.
Key Findings
1) Flood event (Typhoon Odette, Iloilo)
The EO-based analysis estimated:
9.3% of rice area affected by flooding, and
PHP 270.9 ± 0.8 million in losses.
The paper highlights the role of crop stage timing in shaping losses: a large share of exposed rice during the flood was at earlier stages (notably seedling), which can reduce loss magnitude relative to exposure occurring during more sensitive reproductive stages.
2) Drought event (2024 drought, Iloilo)
For the drought case, the EO-based approach estimated:
37.6% of rice area under severe drought conditions (per the study’s classification), and
PHP 390.0 ± 0.3 million in losses.
As with flooding, impacts are strongly shaped by the interaction of severity and phenology—i.e., when the drought occurs relative to crop development.
Comparison with provincial loss reports
A notable contribution of the paper is its side-by-side comparison between EO-derived estimates and consolidated provincial reporting:
Flood: EO PHP 270.9M vs provincial report PHP 692.3M
Drought: EO PHP 390.0M vs provincial report PHP 706.6M
At the municipal level, agreement between EO-based estimates and reported losses was limited, suggesting that EO can play an important role in verification, triangulation, and improving consistency in L&D accounting—especially when decisions depend on where impacts occurred and at what magnitude.
1:100-year flood scenario and forward-looking risk
Beyond observed events, the study models potential losses under a 1:100-year flood scenario, estimating:
PHP 1,694.2 ± 1.8 million (≈ PHP 1.694B)
This scenario illustrates how extreme events could drive substantially higher losses than those observed in a single typhoon—largely because a 1:100 flood inundates a much larger area.
Multi-hazard hotspots for targeting resilience interventions
The paper also maps municipalities where rice areas face overlapping exposure to both severe drought and extreme flooding. These hotspot outputs can support:
prioritization of climate-resilient rice interventions,
pre-positioning of response resources, and
planning for risk reduction in areas repeatedly affected by multiple hazards.
Implications for DRRM and agricultural decision-making
This study supports a practical direction for disaster and agriculture stakeholders:
Faster situational awareness: EO enables rapid, spatially explicit estimates after major events.
More transparent estimates: Reporting uncertainty helps decision-makers interpret confidence and avoid false precision.
Targeted response: Municipal-level mapping can guide where assessments and assistance should be prioritized.
Toward impact-based services: By connecting hazards → crop exposure → yield effects, the workflow contributes to more actionable early warning and recovery planning.
Read the full paper: Araza, A., & Alosnos, E. (2025). Advancing Earth Observation-based methods for rapid mapping and estimation of flood and drought impacts on rice production in the Philippines. International Journal of Disaster Risk Reduction. https://doi.org/10.1016/j.ijdrr.2025.105979
Land-use conversion from agricultural production areas to built-up areas in the Philippines for decades 2000–2020: Spatial analysis and policy implication
A newly published study in Land Use Policy examines where Philippine croplands have been converted to built-up areas over nearly two decades, and what this trend implies for agricultural land protection and development planning. The paper is authored by Arnan B. Araza (first author) and Ma. Christina Corales (co-author)—both part of IMPACT R&D—together with their co-authors.
Key takeaways
Using high-resolution Earth Observation (EO) land-cover data, they mapped cropland converted to built-up areas from 2003 to 2019 across all Philippine provinces.
The map validation using expert-labeled reference data produced 81% user accuracy, suggesting the conversion map is reasonably reliable for national and provincial planning use.
Cropland loss and built-up expansion are strongly linked (r = 0.89), indicating that where built-up areas grow, cropland tends to shrink.
About 10,397 hectares of cropland were mapped as converted to built-up areas—and 31% (3,228 ha) of these conversions occurred inside NPAAAD zones, which are intended to protect prime agricultural lands.
Most NPAAAD conversions were concentrated in agro-industrial and alluvial lands—areas that are both high-value for agriculture and attractive for development.
The findings reinforce the urgency of passing a National Land Use Act (NaLUA) to institutionalize stronger, evidence-based land use planning.
Why this study matters
The Philippines faces a difficult balancing act: enabling economic growth and housing/infrastructure development, while safeguarding agricultural land for food security and long-term sustainability. National land cover maps exist, but they can be updated infrequently and may be inconsistent across years. Earth Observation data offers a way to track land changes more consistently over time and across regions—supporting more timely decisions.
Data and analytical approach
They combined two satellite-derived EO datasets to detect where croplands changed into built-up areas:
A cropland dataset (capturing annual and perennial herbaceous crops; excluding tree crops like coconut and mango), and
A built-up dataset (capturing land surfaces associated with settlements and infrastructure).
To ensure reliability, they validated detected conversion areas using expert-labeled reference checks based on time-series visual inspection (e.g., historical basemaps). They also overlaid conversion results with official government GIS layers—especially NPAAAD (Network of Protected Areas for Agriculture and Agro-Industrial Development) and SAFDZ—to assess whether conversions were occurring in areas that should be safeguarded for agriculture.
Key Findings
1) Built-up growth and cropland loss move together
At the provincial level, cropland decrease and built-up expansion were highly correlated (r = 0.89)—a strong signal that urban expansion is closely associated with the loss of cropland across provinces.
2) Conversion hotspots are not evenly distributed
The conversion pattern is highly uneven across the country. Hotspots were identified in parts of Central and Southern Mindanao and Central Luzon, with additional conversion pressure visible in select provinces in the Visayas. These patterns often align with peri-urban growth, infrastructure corridors, and emerging development zones.
3) A significant share of conversion occurred in protected prime agricultural zones
One of the most policy-relevant findings: 31% of mapped cropland conversion to built-up areas (3,228 ha out of 10,397 ha) occurred within NPAAAD zones—areas meant to be “non-negotiable” for conversion under existing policy intent. This points to potential gaps in governance, monitoring, and enforcement.
4) High-value agricultural lands are especially vulnerable
Nearly 80% of NPAAAD conversions were concentrated in agro-industrial and alluvial lands—prime agricultural areas that are also road-accessible and therefore attractive for urban and industrial projects. This is particularly concerning because these areas are central to sustained agricultural productivity.
What this means for policy and planning
Our results suggest that protecting agricultural land is not only a technical mapping issue—it’s a governance and planning challenge.
What can be strengthened immediately:
Align local land use plans (CLUPs) and zoning ordinances with NPAAAD/SAFDZ guidelines, so local planning reflects nationally designated protected agricultural zones.
Improve monitoring and enforcement using EO-based change detection and digital GIS overlays to identify likely non-compliant conversions early.
Conduct policy compliance audits in conversion hotspots to check for possible irregular zoning decisions, speculative conversion, or implementation gaps.
Support more transparent decision-making by translating geospatial results into story maps and dashboards that are accessible to local stakeholders.
Why NaLUA matters here
The study underscores the urgency of passing a National Land Use Act (NaLUA)—a long-discussed reform that would provide a coherent national framework for land allocation, protection, and use. Without it, land conversion decisions can remain fragmented, inconsistent, and reactive.
Why this matters to IMPACT R&D
At IMPACT R&D, our work focuses on turning research into action—especially where science can support fairer, more resilient development outcomes. This paper contributes national-scale evidence that can help:
LGUs update land use plans using more consistent spatial information,
Policymakers target hotspots where agricultural land protection is breaking down, and
Civil society and communities participate in discussions grounded in data.
Read the full paper: Araza, A.B., Gagarin, W., Corales, M.C., Osorio, C.P., Mendoza, M.D., Ancog, R. (2026). Land-use conversion from agricultural production areas to built-up areas in the Philippines for decades 2000–2020: Spatial analysis and policy implications. Land Use Policy, 162, 107874. https://doi.org/10.1016/j.landusepol.2025.107874