STEM与日常科技·英语30篇（2）

4 / 30

洪水预报与流域模型

Watershed models divide land into small units called hydrologic response units, each with unique soil, slope, and vegetation data.
Rainfall forecasts feed into these models, which simulate how water flows across surfaces, soaks into ground, or runs into streams.
Sensors upstream measure real-time rainfall, river depth, and soil moisture, updating predictions every few hours.
Model outputs show not only flood timing but also peak water levels, flow speeds, and affected neighborhoods on digital maps.
Emergency teams use those maps to decide where to deploy sandbags, evacuate residents, or open floodgates early.
Urban areas pose special challenges because pavement prevents absorption, causing rapid runoff that overwhelms storm drains.
Scientists calibrate models using historical flood data — if past events match simulations, future forecasts gain credibility.
Machine learning now helps refine parameters like evaporation rates or infiltration speed under extreme weather conditions.
Public alerts via apps or radio often cite ‘1-in-50-year flood probability’, derived directly from model uncertainty analysis.
Better models won’t stop floods, but they give people critical extra hours to protect lives and property.