Table 1 Summary of parameter sensitivities for the three moisture tracking models.

Surface flux forcing⁴⁸

Precipitation products (gauge vs. satellite) cause ~4% variation in precipitation recycling ratio; evaporation products (land model vs. remote sensing) cause ~3% variation

Temporal resolution³¹

Higher temporal resolution of forcing data improves accuracy

Resolution synergy³¹

Combined high spatial (0.25°) and temporal (1 h) resolution adjusts local vs. distant source attribution proportions

Tracking direction⁴⁸

Forward vs. backward tracking yields ~6% difference in precipitation recycling ratio

Vertical mixing²⁴

Including/excluding vertical wind speed: continental recycling differs by 25 percentage points, longitudinal transport by 4.8°

Release height²⁴

Vertical profile release vs. surface release: transport distance increases 0.2° latitude, 0.3° longitude

Vertical resolution²⁴

Degrading resolution causes up to 17 percentage points error in continental recycling ratio

Humidity threshold^31,49,50

Relative humidity threshold for precipitation particle selection: 80% → 70% increases identified precipitation amount by 32%, 80% → 90% decreases by 43%; regional optimization needed (global optima 80–85%)

Moisture uptake threshold^31,50

Specific humidity change for evaporation detection (0.05–0.2 g/kg per 6 h): less sensitive than humidity threshold

Tracking duration^31,33

Typically 10–20 days

Data resolution⁴⁹

A climate model (2.5° × 1.88°, 26 levels, 3 h) achieves 95–98% attribution of precipitation; ERA-Interim (1° × 1°, 61 levels, 6 h) achieves 95%