here is the reproducible problem:
import pandas as pd
data=range(50)
df1=pd.DataFrame(data, index=pd.date_range('2013-8-1 8:00:00',freq='1S',periods=len(data))).tz_localize('Europe/Berlin')
t1=pd.date_range(df1.index[0],df1.index[-1],freq='10S')
res1=df1.groupby(df1.index).last().reindex(t1, method='ffill')
res1.index.tzinfo.zone # 'Europe/Berlin'
df2=pd.DataFrame(data, index=pd.date_range('2013-8-2 8:00:00',freq='1S',periods=len(data))).tz_localize('Europe/Berlin')
t2=pd.date_range(df2.index[0],df2.index[-1],freq='10S')
res2=df2.groupby(df2.index).last().reindex(t2, method='ffill')
res2.index.tzinfo.zone # 'Europe/Berlin'
tot=pd.DataFrame()
tot=tot.append(res1)
tot.index.tzinfo.zone #'Europe/Berlin'
tot=tot.append(res2)
tot.index.tzinfo.zone # BUG: UTC !!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# concat works instead
tot_list=[]
tot_list.append(res1)
tot_list.append(res2)
tot=pd.concat(tot_list)
tot.index.tzinfo.zone #'Europe/Berlin'
so we call tot=tot.append(res2) the second time the timezone switches to UTC altough all involved objects have Berlin timezone (tot, res1, res2).
So I suppose we can safely declare this as a bug. Any comments?
In any case concat produces a different output than subsequent append calls. This could be a test case to add...