Page 69 - IJPS-1-1

Raphael J. Nawrotzki, Fernando Riosmena, Lori M. Hunter, and Daniel M. Runfola

Figure 1. Location map of rural MMP municipalities and weather stations.

th
maximum temperature were above the 90 percentile of the 30-year reference period
(1961–1990). The 30-year period from 1961–1990 is known as “climate normal” and
recommended by the World Meteorological Organization (WMO) as reference period for
the study of climatological trends (Arguez and Vose, 2011). Precipitation during extremely
wet days was computed as the annual total precipitation from days when precipitation was
th
greater than the 99 percentile of the 30-year reference period (1961–1990). These climate
change indices have been formalized by the Expert Team on Climate Change Detection
and Indices (ETCCDI), sponsored by the World Meteorological Organization and the
United Nations, to increase the comparability of climate change studies across time and
space (Peterson and Manton, 2008).
Although the GHCN undergoes rigorous quality checks (Menne, Durre, Vose et al.,
2012), about 21% of the records were missing, largely due to instrumentation errors. As
recommended by Auffhammer et al. (2013), we imputed the missing data to generate a
balanced panel of complete weather station records. We employed Multiple Imputation
(MI) (Allison, 2002) using the R package Amelia (Honaker, King and Blackwell, 2011),
which was designed for the imputation of time-series data by explicitly accounting for
temporal trends. The complete time series of daily temperature and precipitation records
were then used as input to construct the two climate change indices for each weather sta-
tion for the years 1961–1999 using the R package climdex.pcic, maintained by the Pacific
Climate Impact Consortium (Bronaugh, 2014).
We then employed CoKriging as a geostatistical method of interpolation (Bolstad, 2012;
Hevesi, Istok and Flint, 1992) to generate a surface of climate change index values across
Mexico. CoKriging is a method frequently employed to interpolate climate measures and
indices (Aznar, Gloaguen, Tapsoba et al., 2013; Rogelis & Werner, 2013) and it allowed us

International Journal of Population Studies | 2015, Volume 1, Issue 1 63