7. FAQ: Forecast Data

7.1. What do I see on the graphic for my selected site (legend) ?

The blue band represents the full weather uncertainty for your site. This means, that WEPROGs Ensemble forecast system (MSEPS) computes 75 different physical based forecasts. In order to make the weather uncertainty visible, those 75 ensemble forecasts are collected in 9 percentiles (P10,P20....,P90). Each percentile represents a blue colour area. The more values are included in a percentile, this area becomes more dark blue in the graph.

Yellow Line (f1) = optimal forecast Orange Line (f2) = Meridian forecast Red Line (f3) = weighted mean forecast

Special cast offshore: Wind speed or sign. waveheight Orangene Line (f2) ´= Wind gusts in [m/s] Red Line (f3) = wave direction on negative axis (North=0, East=-0.5, South=-1, West=-1.5 North=-2) Green Line (f4) = swell direction on negative axis (North=0, East=-0.5, South=-1, West=-1.5 North=-2)

Therefore you can see in the graph:

1. the total uncertainty
2. the most certain probability

Additionally you can mark in the legend the min, max and optimal forecast which are shown in different colors.

What is the difference between the “optimal” forecast and the 9 percentiles?

7.2. What is the “optimal” forecast and how are the 10 percentiles computed?

The optimal forecast is a best guess forecast that is generated out of WEPROG’s 75 ensemble forecasts. If historical measurement data is available these are used to train the optimal forecast and thereby localise the raw forecast data to the location and characteristics of the generating unit. This optimal forecast is statistically optimised and lies often near the percentile P50. The forecast curve is shown as the yellow line and defined as f1 in the legend.

There are 10 percentiles computed from WEPROG’s 75 weather ensemble members of the MSEPS (Multi-Scheme Ensemble Prediction System) and are coloured as bands. For wind power percentiles, there are 300 forecasts behind the percentiles, 4 x 75 wind power forecasts at different heights. A percentile (or a centile) is a measure indicating the value below which a given percentage of the forecasts fall. For example, the 10th percentile (P10) is the value below which 10 percent of the forecasts may be found. The 50th percentile is also known in statistics as the median or second quartile (Q2). In general, percentiles and quartiles are specific types of quantiles (see also definitions of these terms in the Glossary).

Note

If you can not see the yellow line, make sure that in the Forecast Data Interface the f1 is marked in the legend.

7.3. What is the difference between the f1-f4 and u1-u4 curves and how can I activate them?

f1 = optimal forecast

How is the definition of the Optimal Forecast ?

f2-f3 forecasts with different characteristics

WEPROG has set up another 3 different forecast curves based on different characteristics and weights out of the 75 ensemble members (for more details go in contact with WEPROG team)

u1-u4 are empty placeholder for uploaded files (for example: production data, another forecast, user defined forecast....)

Note

If the curves are not marked in the legend&option the curve can not be seen.

Note

The upload file needs a special data format and has to be uploaded into the data base.

How can I upload a file ?

7.4. How can I change the start time and end time of the forecasts that I want to look at?

There are three ways to change the time frame for the displayed data.

1. Using the slider controls
2. Using the date/time input fields
3. Using the skipping buttons

Time slider controls 1

In most cases, this is the most convenient way of changing the start and end time. Click and hold one of the arrows on top of the marked time frame, and slide it to the desired position. The current time will be displayed above the slider in the date/time input boxes. The forecast data image is updated autmatically as soon as the timeframe changes.

Date/time input fields 2

Using the date/time input fields you can precisely pick the start and end time of the forecast data. Clicking on one of the fields will open a calendar where you can manually choose a new start or end time. The forecast data updates automatically as soon as the time frame changes.

Skipping buttons 3

The skipping buttons are a convenient way to shift the time frame backwards and forewards without changing the length of the time frame. With the skipping buttons you can shift the selected period to beginning or end of the currently selected time frame.

7.5. Can I stay logged in over longer time and if so, how do I load new forecasts?

The forecast interface will automatically refresh the available locations and forecasts every ten minutes, as long as you have the forecast window open. With every refresh your login session will also be extended. Your login session expires only if you leave the forecast interface; for example when you remain idle for a long time on another WEPROG page, or leave the website completely.

If you change the forecast location, type, timeframe or timezone the forecast automatically refreshes. Sometimes it is necessary to do a manual refreshment which can be done with a click on the button load.

7.6. What does the restricted Data Mode mean?

Availability data and maximum export settings are ignored by default in the forecast interface. To activate them, mark Restricted as Data Mode in the forecast interface (legend&options/Data Mode).

Availability: If changes are made in the Availability it will effect the forecast graph and values if the restricted data mode is enabled. (Availability you can find in the main menue and is designed for reductions of the capacity in cases of O&M or malfunktions at site. The reduced values can be put into a table.

Maximum export: If there is a difference between the capacity and the maximum export it will also effect the forecast graph and value if the restricted data mode is enabled. (you can find the maximum export value in the forecast sites configure location)

So if there are curtailments stored or their are maxium export settings given to the site you are looking for, then you have to activat Restricted for consideration.

7.7. What does scale to ANN mean?

ANN is the artificial neural network curve which can be used for a optimized forecast influenced by personal settings and real production data (will be explained later how to do that).

The scaling to ANN means, that the uncertainty curves (i.e. percentiles) are scaled to show representative uncertainty to the ANN forecast instead of the raw MSEPS ensemble forecast output. The definition of uncertainty curves can be found by following this link: What is the definition of the percentiles ?

To analyse the ANN forecast in uncertainty space, the user therefore can enable the scaling by clicking on the scae to ANN button in the data mode part of the legend.

7.8. What does the base line mean and how can I activate it?

The base line is a data curve which will be used as the basis for all displayed curves. In other words

• ...All data values will be displayed relative to the selected base line
• ...The base line is used to define the zero-value
• ...The base line will be used as x-axis.

The baseline can be activated at the bottom of the forecast graph page.

There you can select which curve you would like to use as a baseline, even from other locations and forecasts. But as long as Automatically select current forecast is selected, the curve is always refering to the current forecast. If you would like to select a curve from another forecast, uncheck Automatically select current forecast.

Example for a baseline setting: If you want to see the uncertainty in relation to the optimal forecast (f1) you can set the baseline to f1 forecast and the uncertainty to the max and min is shown. This graph can show you now in which hours there is low or high uncertainty.

7.9. Can I download the data I see?

Yes. The forecast data is available as download.

You can get there by opening the forecast interface, clicking on the Table 1 tab, scrolling down, and clicking on the Download as CSV 2 button ( or Excel 5 , Excel 2007 and xml).

...

Note

The interface also offers a Non-Interactive Download Link, which can be used to download the data without logging into the website.

7.10. How does a non-interactive download of forecast data work ?

You find the link at the bottom of the table interface.

The Non-Interactive Download Link can be used to download data from the WEPROG website without a login procedure. The Non-Interactive Link uses a unique access key instead of a login procedure for authentification. This means that you do not have to visit the WEPROG website, but instead can make changes to the download link to select what data you would like to download.

The Link is located at the bottom of the production data table. First open the Table 1 tab, then scroll down to the bottom of the page where you will see the Download as CVS 2 button. Right under it will be a link called show link 3. After clicking on it, a box 4 appears that contains your Non-Interactive Download Link 5.

...

The Non-Interactive Link consists of a list of settings which can be changed to define what data should be downloaded. The general format is &attribute-name=value:

http://www.weprog.com/production_download?location=12345&starttime=2012-11-06+18.00&endtime=2012-11-12+18.00&grouping=hourly&datamode=unrestricted&timezone=UTC+0&variable=1&accesskey=GmuslIxxxxxxxxxxxxxx&output=csv

location

the ID of the location for which forecast data should be loaded lalala

starttime

The beginning of the data time frame. Format: YYYY-MM-DD+HH.MM

endtime

The end of the data time frame. Format: YYYY-MM-DD+HH.MM

grouping

How data should be grouped. Available options are

• all
• monthly
• weekly
• daily
• hourly

Depending on the forecast type, the grouped data will either be summed, or averaged over time.

datamode

The data mode. Can be either restricted or unrestricted.

timezone

The time zone in which the data will be generated. Can be a value between UTC-12 and UTC+12.

accesskey

This key is used for authentification. Do not change.

output

The data output format. Cannot be changed.

7.11. What are the most used features in ELFI?

The most frequent used feature in ELFI is the graphics in the “Forecast Data” part. Here, all graphical time series of the future and past development of the variable at hand include an uncertainty band of the chosen variable in terms of minimum and maximum, as well as percentile bands of P10 to P90 that provide a helpful overview of the possible weather development or the possible output of the gnerating units.

For the further study of a weather situation or evaluation of the weather development for e.g. the next days wind power production, the following features are also helpful:

• performance of the forecast with supplied measurement data
• tables of aggregated or averaged summaries of forecast performance inclusive percentiles
• animated weather maps of the large scale weather
• baseline feature: absolute forecast uncertainty (Delta |min-max|)
• total wind uncertainty
• total solar uncertainty
• total wind+solar uncertainty

The picture below shows such a total uncertainty in relation to the optimal forecast, because the optimal forecast has been set as baseline.

7.12. Which forecast horizon does the historic data show ?

Historic data is by default saved in 6-hourly forecast horizon, because every new forecast overwrites older values.

All verification and analysis of historical data hence takes place with 6-hour forecasts.

For convenience, there are a number of forecast variables, where we save “day-ahead” forecasts, i.e. the forecasts are only updated once per day with the 00 UTC run.

7.13. Can you explain the individual curves (f1-f4, u1-u4) in the legend ?

Yellow Line (f1) = optimal forecast from 75 forecasts Orange Line (f2) = meridian of 75 forecasts Red Line (f3) = weighted mean 1 of 75 forecasts Green Line (f4) = wieghted mean 2 of 75 forecasts

Special case offshore: Wind speed-/wave or sign. wave height Orangene Linie (f2) ´= Wind gusts in [m/s] Rote Linie (f3) = Doirection on neg axis (Nothd=0, NE=-2, East=-4, SE=-3, South=-4, SW=-5, West=-6 NW=-7 North=-8)

Lines u1..u4 are user lines, i.e. here the user can visualise any time series that has been uploaded to the platform and is within the unit band of the original graph.