Entries by Markus Ungersböck

Dear Friends, may I rise your attention to the upcoming international Phenology Conference



Phenology 2018 Conference
One Planet, Two Hemispheres, Many Regions

Sept 23-27 2018, Melbourne, AUSTRALIA

More details can be found on the Conference Website

Recently our Italian partners published a detailed description of their IPHEN network in the International Journal of Biometeorology, the paper can be found on Springer online (Luigi Mariani et al. 2013)

In a recent edition of the German IT journal c’t (2013/17) was a very interesting paper about new visualization techniques based on R-project or a JavaScript library called D3.js.
You are right if you think reading is easier than doing so I spent the one or other evening for the creation of a working example with our PEP725 records. What I have done is to use the example of the streamgraph, adapt the JavaScript source and added some mouseover events. Together with some preprocessed datasets – that was the easy part – we have now the following new graph.

Due to the fact that nearly 90% of all records where contributed by the German colleagues I have to use a logarithmic scale to keep other countries visible. Well it’s only a qualitative representation of our content (so I skipped the y – axis) but nevertheless I think this kind of diagram is an excellent way showing the annual contribution of each national network.
So if you are curious try it (sorry Internet Explorer doesn’t work) Streamgraph of PEP725

For the EGU 2012 poster I’ve created a diagram with the number of reports for each day of the year. It wasn't a big surprise that most of our collected observations are from springtime with smaller amplitude from autumn.

I thought it would also be interesting to look at a statistic about the weekdays of the observations. If we think about ideal observations we can assume that phenological phases are equally distributed over all days of the week. So 14.3% (100% / 7 days) of observations will be on Sundays, 14.3% on Mondays and so on. Observers with a 5 day week will probably only report during their shift; volunteer observers (citizen scientists) maybe more on weekends.

However, that's all only assumptions. Let us take a closer look at the millions of records in the international PEP725 data base. Unfortunately we do not store the weekday of an observation,. Actually we don’t even have a simple date in our records. What we have is the year and the day of the year. So we have to calculate the missing weekdays from the existing information. Our database system (MySQL) provides some handy functions for this task, e.g. DAYOFWEEK and DATE_ADD. The final SQL statement looks like this:
SELECT DAYOFWEEK( DATE_ADD( CONCAT( CAST( (year -1) AS CHAR ) , '-12-31' ) , INTERVAL `day` DAY) ) AS Weekday, count( * ) FROM `pep_data` GROUP BY Weekday


The informative value Figure 2 is not very good,so I have calculated the relative percentage of each weekday and compared it to the ideal result of 14.3% (all observations are equally distributed over the week).

Figure 3 shows the mean reporting practice over the whole PEP725 data set - now let's take a look if there are differences between single countries. For this analysis I’ve selected the members with the most stations: the United Kingdom (UK), Germany (DE), Austria (AT), the Netherlands (NL) and Switzerland (CH).

Austria, Switzerland and Germany are very similar (Figure 4). Monday is always the busiest day of the week. Especially in Austria - it's the only day of the week with more records than expected. Switzerland looks like the generall PEP725 analysis with slightly higher amplitude. Beside the Sunday–Monday problem,Germanys observations are very well distributed over the week. Nearly the opposite case in the Netherlands and UK, probably due to the high numer of volunteer observers (NL Natuurkalender, UK natures's calendar) the weekends are over represented (Figure 5). It would be interesting to know the reason for the busy Monday in the UK

That brings us to one last question – the observation practice over time. The previous results were somehow expected, but it might be interesting if there is some kind of trend how we observe our nature. The PEP725 database reaches back to the year 1868, but there are only a few records available. From the year 1950 onwards we observe a fast development of phenological networks all over Europe with the peak in the 1980s. For that reason we will examine the early records all together and the later ones – those starting in the 1950s and onward.

For the period from 1868 until 1949 we can use about 14,000 observations. As we can see in Figure 6 the monitoring is very similar to the modern citizen scientist’s network as we have found in the Netherland’s records. It seems that volunteer observers are not an invention of recent years.

The occurrence of phenological phases strongly depend on local weather patterns, micro climate and adaption strategies of plants so it is very difficult to identify errors in the cloud of possible variation. The presented analysis of the relative number of observations per weekday shouldn’t be understood as a tool for quality control, but rather as a means for getting a feeling about different observer types and their behavior. If you have a large number of observations it is very interesting to see the influence of professionals and citizen scientists today and in the past.