OBIS: Time series data

Revision as of 15:08, 20 February 2013

The section time series data is one of the core features of OBIS. OBIS allows the management, visualization and download of time series data. The section "Time series data" gives first an overview of all datasets. The measured data are directly linked to the stations, where they have been measured.

View list of available datasets

On the screen most important information are listed:

• Dataset title consists of measured parameter and station name, e.g. "Climate Benguela"
• Start of time series (yy-mm-dd)
• End of time series (yy-mm-dd)
• Time step, e.g. daily, monthly
• TS Measurement + Data Gaps, e.g. precipitation, discharge
• TS Data contains the link to access the data [show]

Metadata

With a click on "Details" you can see all data linked to the dataset. First metadata are listed. They are grouped in Title, Time series information, Dataset and Station details.

If you chose the register "TS Measurements + Data Gaps" time series are listed. In the example Luanda eight time series are available.

View and download time series data

The section "overall statistics and actions" lists the number of entries in total and in brackets the number of entries for each parameter. You get an overview of the completeness of the time series or number of gaps. Start and end date of the time series is shown. For the export of data you have two possibilities: If you are interested in the entire dataset, use the download link "database data" and you get a txt-file containing all data and metadata. The example "climate Luanda" file opened in an editor (see figure below), shows you station information and time series information, lists the parameters in their correct order and gives you the monthly data. Using this download link you get the data as they are stored in the database (no default values, no date gaps). Using the download link "file" the original file will be exported as it has been uploaded to the database.

In the section "Interval statistics and actions" you have the possibility to view the time series data in a plot. You can set a time interval and parameters of interest. Therefore enter the time interval in the entry fields, activate the checkboxes of the parameters of interest and confirm the selection by clicking "set interval & parameter". By clicking on the link "show plot" a graph appears visualizing time series data for the timeframe and parameters selected before. By clicking on the link "database data" you get the time series for this respective timeframe and parameters. You can also chose the function "Export aggregated" for download. This feature allows to export e.g. yearly data, average data, max or min values or sum. Tick the checkboxes of the required output and click on "download".

Visualization of a txt-file in PSPad (download via the link "database data")

The export delivers a tab stop splitted text file. The header contains most important metadata information: Name of the RBIS (e.g. Okavango Basin Information System), Date of download, user name, dataset title, station name, coordinates and spatial reference system, elevation, time series ID, metadata ID and column description. If the data had or have data gaps you will further get the information of applied interpolation methods, date of interpolation and a declaration of completeness.

Time series data analysis

Data gaps

Data gaps will be automatically identified during the import (see Time Series data Import).

The percentage next to the number of entries in the first row shows the completeness of the time series for each parameter. In the Interval statistics section you will find behind every parameter the number of missing values and the sum of gaps. These values are linked to the full list of data gaps. The color of the following "I" shows if data gaps are already filled (green), partly filled (blue) or not filled (red). To fill data gaps please follow the linked "I". The Gap filling Toolbox will open in a new tab and provide several methods to fill data gaps.

Find nearby stations

RBIS allows to search for nearby stations with respect to one parameter (e.g. mean temperature). You can search for neighboring stations by distance (in m), elevation (in m), coefficient of determination (rsquared) and the time interval. The number of results can be limited and ordered either by distance or the coefficient of determination. To simulate the same behavior as during an interpolation process (see Gap filling Toolbox) tick on "Exclude raw data". This will exclude all time series data datasets marked as raw data.

Notice: Before you can use the coefficient of determination you have to calculate it. Click the link on the bottom named as "update calculation of r²".

The table lists nearby stations, their distance, dataset title, elevation, coefficient of determination (rsquared) and start and end date of time series. If the number of your listed datasets is less than 8, a plot can be generated, where all data gaps are marked in red. This is very helpful to identify possible neighboring stations to fill data gaps.

Show gap information will generate a list as it would generated during the interpolation process (see Gap filling Toolbox). In comparison to the above list in addition the time interval and unit have to be equal and date have to be available in respect to each single gap interval.

Gap filling toolbox

Internal interpolation methods

RBIS includes different internal interpolation methods. The methods are explained in the section "Internal Interpolation Method Set". You can chose between following internal interpolation methods:

• inverse distance weighting

Inverse distance weighted methods are based on the assumption that the interpolating surface should be influenced most by the nearby points and less by the more distant points. The interpolating surface is a weighted average of the scatter points and the weight assigned to each scatter point diminishes as the distance from the interpolation point to the scatter point increases.

Inverse distance weighting (IDW) uses the maximum given number of stations. If this number is not available the number is reduced to the given minimum number of stations.

[1] IDW (5-3) Inverse distance weighting (IDW) (max 5 stations - min 3 stations are used)

[1a] IDW & elevation correction (5-3) Inverse distance weighting (IDW) with elevation correction (max 5 stations - min 3 stations are used). Elevation correction is done by adjusting the values of the used stations to the elevation of the source station. These corrected values are used for IDW. If IDW with elevation correction is not possible only IDW is used instead. The default exponent is 2 otherwise it can be changed by the value of the parameter field.

• linear regression

Linear regression is an approach to modeling the relationship between a scalar dependent variable y and one or more explanatory variables denoted X. The case of one explanatory variable is called simple regression. More than one explanatory variable is multiple regression.

[2] linear regression with the values of the station with the best r² is used.

[2a] linear regression (0.7) Linear regression with the values of the station with the best r² in a radius of 10 stations is used if r² is at least 0.7 or greater.

[2b] linear regression (0.7/30km) Linear regression with the values of the station with the best r² in a radius of 10 stations which is at least 0.7 or greater is used. The maximum distance of this station is 30km.

• linear interpolation

Linear interpolation is a method of curve fitting. If two known points are given (value before and after gap), the linear interpolant is the straight line between these points. The default exponent is 2, otherwise it can be changed by the value of the parameter field.

[3] linear interpolation

[3a] linear int/regr. (l3/0.7) Linear interpolation is used if gap length is not more than 3. Gaps greater 3 are closed by linear regression with the values of the station with the best r² in a radius of 10 stations is used if r² is at least 0.7 or greater.

• nearest neighbor

Nearest-neighbor interpolation (also known as proximal interpolation or, in some contexts, point sampling) is a simple method of multivariate interpolation in one or more dimensions. Interpolation is the problem of approximating the value of a function for a non-given point in some space when given the value of that function in points around (neighboring) that point. The nearest neighbor algorithm selects the value of the nearest point and does not consider the values of neighboring points at all, yielding a piecewise-constant interpolant.

[4] nearest neighbour

screen method selection

Define rules for interpolation methods

The user selects one method or can chose a combination of interpolation methods. In addition the user has the possibility to define rules for the application of the method by clicking on "Internal Interpolation Method Set". A description of the method is given here. With support of selected rules the interpolation method can be adapted. Here values can be set for the parameters: length of data gap until the method shall be applied, maximum and minimum station number, coefficient of determination r2 and maximum distance. To set a maximum length of data gap is especially recommended for the linear interpolation method as this method is not suitable for larger data gaps.

Interpolation cannot be conducted in case that no matching time series were found or no matching method was found.

screen internal interpolation method set

Default values have been set, e.g. maximum number of stations is 5, minimum nuber is 3 and the minimum coefficient for determination was set to 0.7.

The section "Internal Limit values" offers a table containing the value range of all parameters, e.g. humidity: minimum value is 0, maximum value is 100.

External interpolation

The section "External Interpolation" offers the possibility to conduct an interpolation out of the OBIS and upload the file into OBIS. If you use an external method you need to chose the method in the selection menue or describe a new method. The import function takes only those values which lie within a previously described data gap, even if the timeseries is longer. The values within the data gap were treated by OBIS like interpolated values. The gap interval can only be filled completely or not. If only one value in the timeseries is missing, it will not be used to close the gap.

Further you need to specify the upload file. You can chose add to complement the timeseries or new to add new data after you have previously deleted already exitsing data. This feature allows to close still existing gaps without losing former interpolation results.

[Back to tutorial main page]