Site Selection for a Weather Radar and Rain Gauge Network
A disaster-management authority wants better rainfall monitoring for a 12,000 km² basin with steep hills in the north and a flat, densely populated delta in the south. Existing records: 14 rain gauges, unevenly distributed - 10 in the delta, 4 in the hills. The hills generate flash floods with lead times under 3 hours; the delta floods slowly over days. A Doppler weather radar is proposed. Constraints: radar needs a clear horizon, cannot see into deep valleys, and has a blind cone near the tower. Budget allows either one radar or about 20 new automatic rain gauges. Satellite rainfall estimates are freely available but biased in mountainous terrain.
1. Compare radar and rain gauges for the two sub-regions, and recommend how to allocate the budget.
Read the worked answer
Hills (flash floods, <3 h lead time):
- Rain gauges measure point rainfall accurately but are too sparse (4 gauges) to capture convective cells and orographic gradients. They give no spatial picture and no advance warning of a cell forming upstream.
- Radar gives area-wide rainfall at high time resolution and can detect a developing storm 30–60 minutes ahead - critical for a 3-hour flash flood. But radar has a blind cone near the tower and beam blockage in deep valleys, and quantitative rainfall from radar needs gauge calibration.
Delta (slow floods, days of lead time):
- The 10 existing gauges already give a reasonable picture; the need is less urgent and lead time is long, so spatial detail matters less.
- Radar here would be useful but less decisive than in the hills.
Recommendation: place the single radar to cover the hills and the upper basin, accepting some valley blockage, and use the remaining budget for automatic gauges in the hills to calibrate radar and fill valley gaps. Use satellite estimates only as a supplementary layer, bias-corrected against gauges, not as the primary source. The delta's existing gauges can be upgraded to automatic transmission within the same network.
2. Explain why satellite rainfall estimates alone would be risky here, and how you would use them.
Read the worked answer
Why risky:
- Satellite algorithms infer rainfall from cloud-top temperature or microwave scattering. In steep terrain, orographic enhancement means actual ground rainfall can be much higher than the cloud-top signal suggests, producing systematic underestimation on windward slopes.
- Valley and ridge contrasts occur at scales finer than many satellite pixels, so a flash-flood cell in one valley may be averaged away.
- Warm-rain and shallow convective processes can be poorly detected.
- For a flash flood with under 3 hours of lead time, a biased or lagged estimate can trigger a false alarm or miss an event entirely.
How to use them:
- As a spatial context layer - showing where rain is falling between gauges - not as the trigger for warnings.
- Bias-corrected against the automatic gauges, with separate correction factors for hills and delta, recalculated each season.
- Blended with radar and gauges in a merged product, so warnings are triggered by observed gauge and radar data, with satellite filling gaps.
- Used for post-event analysis and for monitoring ungauged tributaries.
Judgement: satellite data are valuable for coverage but cannot substitute for ground truth in a mountainous flash-flood setting.