An Alternative Approach to Measuring Drought in the Corn Belt

The impending increase in the frequency of weather extremes means that there may be an expected increase in drought frequency and severity, leading to more variability in agricultural production. Production variability in turn leads to price volatility, production management challenges and concerns over food security. Several drought measures or indices have been developed, but most are not linked directly to agricultural water demands, or if linked, are based on cumulative seasonal precipitation for relatively large geographic areas. Using cumulative seasonal precipitation assumes that each unit of water has the same effect on yields irrespective of when it occurs. This leads to inaccuracies, because the marginal effect of one inch of rain on yield is extremely high during certain periods of time and almost epsilon at others. In this paper drought is defined using a newly developed measure called the Net Crop Moisture Deficit (NCMD). NCMD has not been used in research previously. NCMD is the difference between, the amounts of water a crop would use with a full water supply, less an estimate of effective rainfall. NCMD is estimated at the county level using a simulation model that incorporates monthly crop water requirements, stored soil moisture and effective monthly precipitation. Preseason soil moisture conditions are incorporated based on inter-seasonal rainfall. Droughts which occurred during the 1980 to 2013 time period were identified based on observed dryland corn yield reductions relative to trend yields. These ground truth results were then compared to NCMD and seasonal precipitation measures of drought to evaluate which measure worked best. The results showed that basing drought classifications on NCMD resulted in improved predictions of yield relative to other drought measures. The positive predictive value, which gives the probability of correctly identifying drought years, was higher for NCMD compared to seasonal precipitation. The negative predictive value, which gives the probability of correctly identifying no drought years, was also higher for NCMD compared to seasonal precipitation. NCMD was also more precise in identifying the severity of drought. The average and marginal effect of NCMD on yield was the same for different annual events within a given county. This means that the NCMD regression coefficients are estimates of the marginal contribution of an inch of effective water to yield. It also means that for a given location the impact per inch doesn’t change when the magnitude of the shortage changes. For ex ante analysis regression equations were estimated on grain yields based on the cumulative NCMD value at the end of July. The forecasted yield deviations were compared to the observed yield deviations and to the forecasted values using the full season moisture shortfall. The results showed that there was no statistical difference in the yield prediction using July rather than end of season values. This effectively means that one can forecast the reduction in yield at the end of July.


Issue Date:
2016
Publication Type:
Conference Paper/ Presentation
PURL Identifier:
http://purl.umn.edu/235628
Total Pages:
29




 Record created 2017-04-01, last modified 2017-08-29

Fulltext:
Download fulltext
PDF

Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)