Extremely low yearly mean for STELLDBTN 2013

[veenstrajelmer]

When computing yearly means for STELLDBTN and surrounding stations, I noticed the yearly mean of STELLDBTN for 2013 is unexpectely low. This was originally reported in M220207273 on 9-2-2022.

import ddlpy
import matplotlib.pyplot as plt
plt.close("all")

locations = ddlpy.locations()
bool_hoedanigheid = locations['Hoedanigheid.Code'].isin(['NAP'])
bool_grootheid = locations['Grootheid.Code'].isin(['WATHTE'])
bool_groepering = locations['Groepering.Code'].isin(['NVT'])

start_date = "2005-01-01"
end_date = "2019-12-31"

fig, ax = plt.subplots()

for stat in ["LICHTELGRE","HARVT10","BROUWHVSGT08","STELLDBTN","HOEKVHLD"]:
    bool_stations = locations.index.isin([stat])
    selected = locations.loc[bool_grootheid & bool_hoedanigheid & bool_groepering & bool_stations]

    measurements = ddlpy.measurements(selected.iloc[0], start_date, end_date)
    meas_yearmean = measurements[['Meetwaarde.Waarde_Numeriek']].groupby(measurements.index.year).mean()
    meas_yearmean.plot(y='Meetwaarde.Waarde_Numeriek', ax=ax, label=stat)
ax.legend()
ax.grid()

Gives (y-axis units are centimeters):

Is this to be expected or is something wrong with the measurement timeseries?

In a study for RMM kalibration/validation, a shift in the bias was observed (figure from feb 2022):

However, there is no sudden jump of 0.5 meter visible in the measurement timeseries from ddl. It is possible that this is a model artifact, however unlikely. It could be related to the low yearly mean.

[TvLoon-RWS]

This question will be forwarded to our measurement expert, after we check if there is no discrepancy between DL and Donar