Validation of the processed IMERG V07 rainfall dataset used for HEC-RAS 2D rain-on-grid modeling in Guyana.
Marker color uses pooled 2020–2021 PBIAS at the nearest valid final NetCDF cell. Red = IMERG low by more than 10%; blue = IMERG high by more than 10%; green = within ±10%. Grey gages do not have usable 2020–2021 daily gage rainfall records in the supplied DSS.
This is the most important hydrologic signal: IMERG tends to over-report light rainfall while increasingly underestimating the gage heavy-rainfall tail.
The bias totals are relatively insensitive to the daily boundary, but correlation and daily RMSE are not. The best-aligned boundary is therefore used only for daily-skill diagnostics. Confirm the actual gauge observation/readout schedule before formal correction calibration.
Gage rainfall time series were extracted from the supplied HEC-DSS files and paired with station coordinates from the supplied KMZ. The records used in the primary validation were converted to matched daily gage values for comparison with daily IMERG accumulations. Duplicate Cane Grove Back records from two DSS basin parts were checked; their values were identical before de-duplication.
The comparison uses the final EPSG:32621 IMERG NetCDF — the same processed rainfall intended for HEC-RAS — sampled at the nearest valid rainfall-grid cell. As an additional spatial check, rainfall was also averaged over the nine IMERG cells surrounding each gauge (3×3 neighborhood). This average was used only to assess point-to-grid spatial representativeness and was not used to replace the nearest-cell comparison.
The dashboard reports PBIAS, Obs/IMERG ratio, MAE, RMSE, Pearson correlation, and 1 mm/day wet-day detection metrics. Annual totals are insensitive to daily timing alignment. Daily skill metrics use the empirically best accumulation boundary shown on the Timing tab.
Do not apply a universal multiplier to the HEC-RAS rainfall yet. First confirm gauge timing, then test an intensity-dependent correction focused on the heavy-rainfall tail with strict out-of-sample validation (for example, calibrate on 2020 and validate on 2021, then test the independent 2022–2023 events). If stable geographic structure persists, a spatially varying correction can then be evaluated.
The primary gage network used in this assessment is daily, whereas HEC-RAS is driven by 30-minute IMERG rainfall. The heavy-daily-rainfall underestimation is consistent with attenuation of rainfall extremes, but it does not by itself prove that every 30-minute IMERG peak is too low. Hourly or sub-hourly gauge data should be used to validate the temporal-intensity correction before modifying the production HEC-RAS forcing.
A gauge is a point measurement, while IMERG represents an areal satellite pixel and the final NetCDF has also been reprojected. Some disagreement is therefore representativeness error rather than pure satellite bias. This is especially important when interpreting individual gauges.
This interactive dashboard presents a comparison of IMERG satellite rainfall estimates and rain-gauge observations in Guyana. The archived research version and associated documentation can be made available through HydroShare.
Suggested citation: Habib, E., Morsy, M., and Williams, G. (2026). Guyana IMERG Rainfall Validation: Interactive Dashboard. University of Louisiana at Lafayette and Louisiana Watershed Flood Center. HydroShare DOI: pending assignment.
Version: 1.0 (web publication preparation, November 11, 2026). Archive: HydroShare resource link to be added after publication.
Interactive charts and maps require an internet connection to load Plotly, Leaflet, and map tiles. For reproducibility, consult the HydroShare archive when available.