I chose sites along the coast with more or less complete records. I’ve added a cluster around San Quentin
All the data for the stations is used to calculate a mean for the year day,
d$yday<-yday(d$Fecha)
d %>% na.omit() %>% group_by(var,name,yday) %>% summarise(val=mean(Valor)) %>% na.omit()->dd
As there are missing data and some other anomalies this is only a rough picture and would need further work if we really did need a proper trend analysis. The pattern should more or less match accross stations when it comes to any effect of regional climate change. When it doesn’t its a data quality issue. This is usual with Mexican data and can be resolved through matching data accross stations and taking majority votes. Wilmott’s data set for the same area has been cleaned up at global resolution. See my app for that, but that is monthly, not daily.
The aggregated data is held in this HTML document and can be filtered and extracted for Excel from it.