University in United States - Wyoming, http://weather.uwyo.edu/
has archive of aerological soundings from whole world,in HTML-TXT form every termin since year 1973:
Europe has quite a lot of stations with ballon soundings. TXT output has values of many things through pressure levels, usually 15 standard levels and levels with some significant changes - 15 standard levels 925 850 700 500 400 300 250 200 150 100 70 50 30 20 10 hPa
. Ballon usualy rises up to 5-7 hPa (triple point Solid-Liquid_Gaseous for water is at 0.01 °C and 6.11 hPa). Many indexes are aviable from whole soundings.
Data are at html-txt format and parameters for output are chosen by URL address, it is easy to use Bash script for downloading. It is possible to get whole month, one or more days (at the same month) at one request (first and last termin have to be valid), one or more termins of soundings at one request. I have choosen one sounding termin to one file for better working with data after. At Praha Libus 11520, there are up to 4 soundings per day, 00 06 12 18 UTC during years 1973-2018, but not always, it is 67204 possible termins in 16801 days, but not all termins are aviable always. There was only 00 UTC and 12 UTC sounding during first years and there is no more 18 UTC soundings since year 2013. Most of other aerological stations usually make soundings twice a day - 00 a 12 UTC, some stations only once a day. Every ballon run cost about 150-200 Eur and mostly - it is not possible to reuse device. Ozone sounding cost more than 1000 Eur per run and about half of ozone sondes is returned for reuse.
We already have data from all soundings at TXT-HTML format, through years 1973-2018. Another Bash scripts is going to be used, at Linux of course. Script makes parsing of all TXT-HTML data (cat, awk, grep, cut, sed, tr), where is temperature and other values at 15 standard levels through years 1973-2018. I choose two termins per day - 00 a 12 UTC for better homogenity and data are aviable through all years. Only temperature will be analised in 15 standard levels. Output of Bash script for TXT-HTML parsing is CSV file with data for whole period 1973-2018 and with termins 00 and 12 UTC, with temperature at 15 standard levels. Now - old version (until 2003) of MS Excell (.xls) will be used for graphs creating, computing averages, highest and lowest values. Another Bash script was used for formulas into Excell, but now, Windows need to be used, that is pity. Older Excell has quite similar data processing and low system requirement. It is possible to use Libre Office calculator (.ods) at Linux, but it takes much more CPU, files are with worse compression ratio and graphs are not so simple, PNG images have bigger size.Method of sounding has been changed through years
and data are not complete homogenous. Today soundings goes through GPS and internet, but earlier - ballon was watched and data was downloaded through radio connection. Quality of data changes through years. There are quite a lot evidently error values at first years, especially at 30-10 hPa. And parsing of html produces sometimes another false values, if format of data is not always the same, there can be some anomalies, missing data etc. It is not possible yet to detect false values by script. It was necessary to look through big data file and find error values, error values are visible after plotting graph.
But it is not easy to see, what is error value and what not. It is lot of data, and data in stratosphere (30-10 hPa especially, most often at 10 hPa) are missing often. There are error values, that has not been found and it can change linear trends too, if errors are big.How results look like ?
Graphs, trends, statistics was made for two periods. First period is shorter - 1999-2018 (similar period is with aviable data at Ogimet SYNOP - since 28.09.1999). Data are with good quality in this period and trends are quite clear. All 15 standard levels has rising linear trend through years 1999-2018.
Level 925 hPa is aviable since year 1992. Another 14 levels are aviable since year 1973 (at first years, high altitude levels are with lot of missing data). Linear trend 1973-2018 is increasing at lower levels: 925 hPa - 300 hPa and decreasing 250 hPa - 10 hPa.
And 925 hPa is aviable only since year 1992. Through years 1973-2018 lowest level from 14 standard levels 850 hPa has highest linear increasing trend. But it can be higher at 925 hPa. Decreasing trend at higher levels is not so significant, it can be caused too with error values and mistakes from first years. Differences between years are quite big compare to trends, trend do not need to be so much significant. Stratosphere has stable stracification, layers above tropopause are not so mixed. Levels at stratosphere are at higher altitude in Summer, geopotential of upper levels is increasing,when lower levels are warmer. It can be one of reason for decreasing linear trend through years 1973-2018 at upper levels - 250 hPa and higher. At layer, 250-10 hPa, there is about 24 % of mass and thermal capacity of atmosphere. But there is lot of variability and error values at first years in these levels. Three graphs was made - from average, maximum and minimum value for all termins through 15 levels (some termins has lot of levels with missing data). This shows quality of data too.
All data for downloading - CSV, XLS files, PNG graphs, including non-final version with removing errors:
Graphs PNG gallery:
There is not only temperature, but another values for 15 standard levels: Geopotential height, ThetaA, ThetaE, ThetaA virtual, Dew point, Relative Humidity, Mixing ratio, Wind speed, Wind direction and plenty of indices for entire sounding.
But finding and removing of error values is more complicated than with temperature case. However - creating data file through years and downloading data with entire soundings, it can be easily done for all stations, all soundings, all values and sounding indices, BUT without data checking.
Average temperature at 15 standard levels and linear trend through 1973-2018, termins 00 and 12 UTC.
Data source (Number of station Praha Libus is 11520):http://weather.uwyo.edu/upperair/europe.html
And average value from yearly averages at 15 standard levels, highest value from yearly averages at 15 standard levels, lowest value from yearly averages at 15 standard levels. It shows data quality too. Aviability of 925 hPa since 1992 is visible too.