Mismatch in Köppen-Geiger climate classifications

[AdamRJensen]

One user reported that

there are some differences in the climate information documented on the BSRN webpage (https://wiki.pangaea.de/wiki/BSRN) with those given in solarstation.org.
For example, for the stations and PAY LIN, the climate zone is DFb in BSRN, whereas it is Cfb on solarstation.org.

This is somewhat to be expected, as Köppen-Geiger climate zones aren't static in time, and surely vary somewhat depending on what dataset is used. While this isn't a huge deal, it is also not ideal to have users reporting different climate zones.

One solution would be to override the climate information of the BSRN stations with that found on their website. Thoughts on this?

[kandersolar]

I wonder how the BSRN's zones were determined. I suppose we could compare calculation methods; if their method is simpler or out of date, perhaps the best option overall would be for them to use ours!

[AdamRJensen]

I've sent an email to Amelie from the BSRN asking what data source they have used.

[AdamRJensen]

The BSRN classifications is based on the dataset provided by https://www.gloh2o.org/ and available here: https://figshare.com/articles/dataset/Present_and_future_K_ppen-Geiger_climate_classification_maps_at_1-km_resolution/6396959/2

I have the Python code for how they derived it using the image above.

[AdamRJensen]

Here's a list of the BSRN climate zones that don't agree with the climate zones from the kgcpy Python package.

Abbreviation	Latitude	Longitude	climate_bsrn	climate_kgcpy
BER	32.3008	-64.7660	Af	Ocean
BON	40.0667	-88.3667	Dfa	Cfa
BUD	47.4291	19.1822	Dfb	Cfb
COC	-12.1930	96.8350	Af	Ocean
ENA	39.0911	-28.0292	Csa	Ocean
INO	44.3439	26.0123	Dfa	Cfa
ISH	24.3367	124.1644	Af	Ocean
KWA	8.7200	167.7310	Af	Ocean
LMP	35.5180	12.6300	BSh	Ocean
LER	60.1389	-1.1847	Cfc	Cfb
LIN	52.2100	14.1220	Dfb	Cfb
LYU	22.0370	121.5583	Af	Ocean
NAU	-0.5210	166.9167	Af	Ocean
PAY	46.8123	6.9422	Dfb	Cfb
PSU	40.7200	-77.9333	Dfa	Cfb
RUN	-20.9014	55.4836	Am	Ocean
SBO	30.8597	34.7794	BWh	BSh
TIK	71.5862	128.9188	ET	Dsd
TNB	-74.6223	164.2283	Aw	EF
XIA	39.7540	116.9620	Dwa	BSk

[AdamRJensen]

kgcpy labels a lot of climates either Cfa or Cfb, whereas the BSRN climates are Dfa/DfbCfc - these seem rather similar and such deviations are probably to be expected when using different datasets.

Also, the kgcpy package list a lot of climates as "Ocean", which are all small islands in the Ocean (with the exception of SYO and GVN, see below).

• TNB - correctly labeled as EF by kgcpy (currently incorrectly labeled as Aw)
• TIK - station on the north coast of Russia (currently labeled as ET which is probably correct). kgcpy labels it as Dsd.
• GVN and SYO - station have no climate zone on the BSRN listing, whereas kgcpy lists it as Ocean. The correct climate zone is probably E, as it is located on the coast of Antarctica.

[AdamRJensen]

Code:

import pandas as pd
import kgcpy

bsrn = pd.read_html('https://wiki.pangaea.de/wiki/BSRN#Sortable_Table_of_Stations')[0]
bsrn.index = bsrn['Abbreviation']

bsrn['climate_bsrn'] = bsrn['Köppen-Geiger classification'].str.split(' ').str[0]

for index, row in bsrn.iterrows():
    bsrn.loc[index, 'climate_kgcpy'] = kgcpy.lookupCZ(row['Latitude'], row['Longitude'])

bsrn['is_first_letter_same'] = bsrn['climate_bsrn'].str[0] == bsrn['climate_kgcpy'].str[0]
print(bsrn['is_first_letter_same'].value_counts())

bsrn['is_climate_same'] = bsrn['climate_bsrn'] == bsrn['climate_kgcpy']
print(bsrn['is_climate_same'].value_counts())

bsrn.loc[~bsrn['is_climate_same'] & bsrn['climate_bsrn'].notna(), ['Latitude','Longitude','climate_bsrn','climate_kgcpy']]