Comparison of the SPI and ESPI on predicting drought conditions and streamflow in Canada

Canada is experiencing a growing demand for water use. This is concern for agriculture as 80 percent of the agricultural land in Canada is within prairie landscapes, and in these landscapes water is one of the most limiting variables for crop production. While drought is a common occurrence in the prairie landscape, droughts can occur in almost any climate zone in Canada and the consequences can have a significant impact on agricultural production and water use in almost any region of Canada. While Canada has experiences few severe and prolonged droughts since conventional agricultural practices were introduced, recent studies have linked the occurrence of droughts to increasing temperatures in over past few decades. Future projections of climate conditions indicate that more severe droughts are likely to occur.

One drought index that is becoming a standard tool for assessing drought is the Standardized Precipitation Index (SPI), which is based solely on precipitation variability. However, the Standardized Precipitation and Evaporative Index (SPEI) takes into account the effect of temperature on drought. In Canada this could be an important element in understanding the effect of increasing temperatures on drought. This study compares the SPI and SPEI in different climate zones in Canada in order to determine which one better represents past drought conditions. The study also introduced artificial warming to determine whether one index was better at representing drought conditions under anticipated increased temperature scenarios. Both indices were also assessed in terms of how they represent monthly stream flows in selected prairie river systems.

There was no significant difference in how the indices represented past drought conditions however the SPEI was better at capturing the effect of increasing temperatures on droughts based on the artificial warming models. The SPEI was able to only partially represent streamflow conditions. The correlations between SPEI and streamflow varied with the season, showing stronger correlations during the summer and autumn months and weaker correlations during the winter and spring. Statistically significant correlations were observed for 6, 9 and 12 month temporal scales however the correlations weakened at temporal scales higher than 12 months.