Landslide Early Warning Needs Ground-Truth Accountability, Not Rainfall Thresholds Alone
Rainfall thresholds are essential to landslide and debris-flow warning, but they are not the whole warning system. A threshold exceedance can be too broad, too narrow, stale after land-cover change, poorly connected to a landslide inventory, or communicated without protective-action follow-through. This conceptual synthesis combines USGS landslide programme, preparedness, post-fire debris-flow assessment, and inventory sources; NASA Landslide Reporter, Global Landslide Catalog, and disasters remote-sensing resources; NWS debris-flow safety guidance; WMO and UNDRR early-warning doctrine; Hong Kong landslip warning practice; and British Geological Survey landslide science. The contribution is a ground-truth accountability ledger with stages for trigger threshold, susceptibility context, ground-truth report, protective action, and post-event learning. The conclusion is that landslide warnings should report threshold performance and observed outcomes, not only that rainfall exceeded a number.
Introduction
Landslide early warning can appear rigorous because it has rainfall thresholds. But thresholds are only one part of a warning chain: susceptibility, burn severity, inventories, field reporting, communication reach, protective action, and post-event update all determine whether a warning is useful.
USGS sources show the technical side: landslide hazards, preparedness, post-fire debris-flow assessment, and inventories [[cite:usgsProgram,usgsPreparedness,postFire,inventories]]. WMO and UNDRR sources show the system side: risk knowledge, monitoring, dissemination, and response capability [[cite:wmoMhews,undrr]].
Method
The study is a conceptual synthesis of geoscience warning sources and early-warning governance sources. Graph search found no close duplicate landslide warning accountability item.
The analysis weights sources by whether they connect trigger information to observed events, protective action, and model learning.
Results
The first result is that threshold evidence must be attached to place context. Post-fire debris-flow assessment uses burn severity, terrain, and rainfall information, while inventories preserve where landslides have actually occurred [[cite:postFire,inventories]].
The second result is that ground truth and last-mile action cannot be assumed. NASA Landslide Reporter and catalog resources support event records, and WMO/UNDRR warning doctrine requires dissemination and response capability, not only monitoring [[cite:nasaReporter,nasaCatalog,wmoMhews,undrr]].
Discussion
A rainfall threshold is a hypothesis about hazardous conditions. It becomes an accountable warning only when the system also records susceptibility context, warning geography, field or remote confirmation, protective action, and post-event calibration.
The ledger also clarifies false alarms. A warning with no observed landslide may be a useful precaution, a threshold too sensitive, a reporting failure, or a missed event. Without ground-truth records, the system cannot distinguish those cases.
Limitations
This paper does not estimate thresholds for a particular basin or landslide type.
Some warning systems are intentionally conservative; accountability should not punish precaution, but it should preserve enough data to learn.
Conclusion
Landslide early warning needs ground-truth accountability, not rainfall thresholds alone. The reporting denominator should be trigger, susceptibility, observed outcome, protective action, and post-event threshold repair.