Precipitation interpolation, autocorrelation, and predicting spatiotemporal variation in runoff in data sparse regions: Application to Panama

•A SWAT model of Panama was built and tested without hydrological calibration.•Regional/seasonal differences in spatial patterning of rainfall were incorporated.•The modified model greatly improved prediction of hydrological variables.•Careful treatment of rainfall input is key to enabling an uncalibrated approach.•The approach is relevant to other areas of the Global South with data limitations.