Get an insight of the quality of your ground measurements

Last update: Jan. 2019

    4-graphs - Quizz

 

The 4 graphs used for the analysis

     (++): advantages
     (--): limits

"One-day graph"

     

"One-day graph": visualize the data one day after the other is essential to spot particular issues in the data.

  • y axis: radiation values (in W/m2, Wh/m2, µmol/m2/s1...)
  • x axis: the instant (time) in a day

When x (time) is close to zero, this is night time and all radiation values are equal to zero. Then sun is rising and radiation starts being received at the top of atmosphere (cyan), in cloud-free conditions (red), and in all-weather condition (blue is the estimation, and green the reference). Then, sunset, and night again.

  • (++): the information of time is available = you know when an event occurs in the data.
  • (++): compare several resources at the same time.
  • (--): one day of data is visible one at a time, which might become very boring when exploring thousands of days of data.
          

"2D-view"

 

"2D-view": very convenient to visualize a dataset over the whole period of time at a single glance. Each one-day graph (as explained above) represents a column of this image. Dark blue is night values, other colors are radiation values above 0 with maxima in dark red. Red lines correspond to sunrise and sunset.

  • y axis: instants (time) in a day
  • x axis: days

On the top of the image (time close to 0), this is the night, and radiation (colorbar on the right-hand side) is equal to zero. Then, when moving down on the image, one meets the red line corresponding to the moment when the sun is rising. Radiation reaches its maximum at noon True Solar Time if no cloud, to finally go down again until dusk (red line again).

The shape observed on this "2D-view" is typical for mid-latitude locations in Northern hemisphere. When the day are short (for day around "0", "380", "750", "1100", and "1480" on this graph), this is winter time. Longest days correspond to summer time.

  • (++): particularly suited to visualize holes, shadows and temporal shifts in the data
  • (++): the information of time is available = you know when an event occurs.
  • (--): only one resource can be visualized one at a time.
  • (--): difficult to observe isolated events that affect a few instants in a day.
     

"Shadow map"

 

"Shadow map": maybe one of the more powerful (but complex to explain :o) ) tool to explore shadows (or masking) on measurements. This graphs depicts the log10 of the ratio of the two cumulative radiation values time series (typically ground measurements/estimation).

  • y axis: sun elevation in degrees
  • x axis: sun azimuth in degrees

This graph is very intuitive in the sense that each bin (pixel) corresponds to the exact position of the sun at one instant. It is as if an observer was lied down on the earth surface watching the sun at each instant and putting the quantity of energy (or power) in each corresponding box. The value inside each box is less intuitive (log10 of the ratio of the two cumulative radiation values). Most of the time, the value corresponds to the log10 of the ground measurements divided by any measurement that you know in advance that they don't face any nearby shadowing effect due to for instance trees, buildings or other in-situ measuring devices.

  • (++): particularly suited to identify nearby shadowing effects
  • (++): can be used as well to confirm a time shift in the data
  • (--): very dependant on the time step of the data => the smaller, the better.
  • (--): as bins result from cumulative values, an artefact observed during one year could be hidden by the values of the next year. Same remark for spring and autumn values that fill the same bins (=pixels).
     

"2D-histogram"

 
"2D-histogram": this graph is the most common way to compare a dataset (an estimation) with another one (typically a reference). Each bins (or colored pixel,  with colobar on the right-hand side) is a cumulative count of instants for a given radiation level.
  • y axis: radiation values of the estimation
  • x axis: radiation values of the reference
When both datasets are perfectly coincident, all points are located along the y=x line. The further from this line, the less the estimation is capable of representing the reference. Night values are discarded before plotting.
  • (++): this graph is interesting to identify if a model shows different performances depending on the type of weather: clear-sky (cloud-free , intermediate , or cloudy/rainy  ).
    NB: Please they might be some confusion between low radiation values observed for clear-sky days but in winter time with low radiation values in summer time due to clouds. We recommend users to work with normalized variables, such as clearness index (Kt=radiation/Top of Atmosphere) 
  • (--): the information of time is lost in this picture
  • (--): as bins result from cumulative values, an artefact observed during a short period of time could be hidden or blurred by correct data having same level of radiation during another period of time.

Top of page

QUIZZ: Let's play a little game. Would you be able to find what's going on with these data?

 

Reminder: by default, we read the data in UT (Universal Time).

   

(Click on the button to get the solution)

     

 
     

(Click on the image to magnify)
 
     

(Click on the image to magnify)
 

 

Top of page