Hello,
I used the AMET tool for the statistical analysis. One of the plots I generated using “run_stacked_barplot.csh” from “script_analysis/aqExample” folder in AMETv1.4.
So, The stacked_barplot_data.csv ARDEQ_REGION_F_1_stacked_barplot_data.csv (169 Bytes)
shows observed and model data for the SO4 is 1.0446 and 1.2378 respectively (attached).
If I am not mistaken, stacked_barplot.csh script takes an average of observed and model data ignoring NULL (-999) values.
When I manually average the data set from sitex_output IMPROVE_ARDEQ_REGION_F.csv (468.2 KB) avoiding NULL values for the observed and modeled value are 1.18 and 1.38 respectively.
Would you please help me understand why different averages are shown?
I’ll need to look closer, but one thing to note immediately regarding the stacked bar plots is that it’s not simply removal of missing/NULL values for each individual species. Any time AMET runs into ANY missing species that makes up the bar plot it removes ALL the species values for that particular observation time and location, not just the species that’s missing.
So, for example, if sulfate is missing but all the other species have valid values, all the valid values are still removed. Therefore, the stacked bar plots are made up of only observing times and sites where ALL the species are valid. So, in order to reproduce the values in the stacked bar plot you will need to be sure to remove not just the missing/NULL values for the species you’re averaging, but also any other times when any of the other species are missing.
My apologies is that is what you’ve already done, but I’m going to assume it is not what you did to create the average values yourself.
Something else to note, there is an option to control whether the stacked bar plot the mean or median. You need to check to see if the “use_median” flag is set to ‘y’ in the all_scripts.input file (see below). If it is, then the value calculated is the median and not mean.