The detection and prediction of agricultural drought in the Canadian prairies
This thesis examined four drought indices (Palmer Drought Severity Index (PDSI), Standardized Precipitation Index (SPI), NOAA Drought Index (NDI), and Moisture Anomaly Index (Z)). The comparison revealed that the Moisture Anomaly Index (Z) is the 'best' index for measuring agricultural drought on the Canadian prairies. Not surprisingly, the moisture conditions during the month of June were found to be the most important determinant of yield. A subsequent diagnostic analysis determined that the performance of the Z-index (as an indicator of agricultural drought) is influenced by the number and timing of precipitation events during the growing season. This drought index was used to divide the study region into five relatively homogeneous crop district clusters. Statistics on drought frequency, severity, and spatial extent were calculated using these regions. In Cluster 2 and Cluster 5, approximately one out every six growing seasons experiences moisture conditions that are not suitable for crop production. A spatial analysis showed that there are three preferred patterns of drought on the prairies and a temporal analysis revealed the presence of coherent drought periodicities (in particular quasi-2, 4, and 10-15 year oscillations). Finally, the relationship between global teleconnection indices and the occurrence of agricultural drought was explored through a composite analysis and a Principal Components Analysis. Both of these analyses suggested that North Pacific circulation and SST anomalies during winter, and eastern equatorial SST anomalies during spring may be related to growing season drought. In addition, Atlantic sector teleconnections also appear to be important.