Researchers used soil samples from different locations to predict soil moisture parameters using artificial neural networks. They found that linear regression helped reduce prediction bias, and the accuracy of predicting saturated water capacity was better than predicting other moisture capacities.