The researchers studied different methods to predict daily rainfall in a large forest area. They found that using elevation and slope did not improve the predictions. Simple kriging, ordinary kriging, and log-normal kriging gave similar results, with ordinary kriging being slightly better. Multi-Gaussian kriging was the least accurate due to data spikes. They also used a method called sequential Gaussian simulation to create multiple maps of rainfall. By combining different maps in a GIS system, they identified potential locations for future rain gauge placement.