Death comes to us all. Life's but a walking shadow | I've been doing some state level yield models this past few months, fitting various weather parameters and crop condition numbers to historical yields. My measure of any given parameter's contribution is its effect on the correlation coefficient, R squared. I add a parameter, if it improves the R squared it stays, if not I take it out. Depending on the crop I can generally account for 90% of the year to year variation in yield but not much more. Granted there is error both in the parameters (ie precipitation, temp variation, crop condition scores) as well as yield (USDA final) but with both temperature and rainfall I'm using statewide monthly averages so theoretically effects on yield should "average out".
At some point you need to begin to consider the error in the reported yield. My question for grain handlers, people who buy & sell grain every day and who should have a pretty good feel for the grain supply in an area especially after a few months. could the USDA's yield & supply numbers be routinely off by 10%? |
A question for those involve in the grain trade.
