Climate data can help predict some crop failures several months before harvest, according to a new study from an international team. Scientists found that in about one-third of global cropland, temperature and soil moisture have strong relationships to the yield of wheat and rice at harvest. For those two key crops, a computer model could predict crop failures three months in advance for about 20 percent of global cropland, according to the study