2. General Interface Description
METexpress is a web space that includes various apps for specific purposes. The current list of apps includes:
MET Upper Air
MET Anomaly Correlation
MET Air Quality
Future development of METexpress is likely to add more apps.
Each app can have multiple plot types, such as time series and vertical profiles. The available plot types include those appropriate to each app.
Note that each app will have a similar layout for selecting parameters and interacting with plots. These general features are described in detail here using the Upper Air app as an example. Individual apps are then described only in terms of their specific information and functionality.
2.1. Home Page
The METexpress home page is shown in Figure 2.1. The home page has a button for each available app. Each app lets the user create certain types of plots for a certain type of data. The user must begin by selecting the app they want to use.
2.2. User Interface Layout Common to Each App
The Upper Air app is used as an example here to describe the general layout that is carried through on all apps. The Upper Air user interface before any curves have been defined is shown in Figure 2.2. There are 3 parts to this screen:
Curve Data Parameters
After any curves have been added, the user interface screen includes a fourth area, as shown in Figure 2.3, that shows what curves have already been defined. The section for the Curve Data Parameters then allows the user to create additional curves to be added to the plot.
2.3. Plot Types
Immediately below the name of the app is a bar that allows selection of the plot type. Available types for Upper Air include Time Series, Profile, DieOff, ValidTime, Histogram, and Contour. Other apps include some plot types specific to that application domain, such as Reliability plots in the Ensemble app or GridScale plots in the Precip app. The section for each app below will describe the available plot types for that app.
The user will first want to decide what type of plot they want to produce initially. Most apps default to Time Series, but the user should change to the type of plot they want to create. If the plot type is changed after curves have been defined, the user will be prompted whether to keep all existing defined curves or to remove all the existing curves.
2.4. Curve Data Parameters
In the central area of the screen is a box to define the data parameters associated with each curve to be plotted. Blue fields are selectors and green fields are action buttons.
A default set of selections for each parameter (in blue boxes) will be shown, which will define the data to be plotted. The user can change any of these selectors. The values available for most selectors depend on what data is stored in the database, and on the values selected for other fields. For example if the DATA-SOURCE selected is GFS, the values available for REGION will be only the regions that are stored in the database for GFS.
For these parameters there is a superior-dependent relationship. For example DATA-SOURCE (which is most typically the name of a model) is superior to VARIABLE and REGION because only certain variables and regions are available for a given DATA-SOURCE. It is recommended that a user change parameters starting at the top left and progressing across then down the page in typical reading order. This ordering generally keeps the superior-dependent relationship intact. If a dependent parameter is set and the superior parameter is subsequently changed, the program attempts to maintain the dependent parameter value if it is still available for the new superior value. If the dependent value is not available for the new superior then a default value is automatically set for the dependent. This maintains consistency and prevents invalid parameter choices.
In the example Upper Air app, the selectable values are derived from the data for these parameters:
Group (as defined in the MET database)
Database (name of user database)
Data-Source (usually a model name)
Forecast lead time
Dates – date range that applies to all curves, used for plot types like time series
Curve-dates – range of dates used to filter data for a particular curve, such as seen in vertical profiles, where each curve may potentially be based on data from different date ranges
Some fields are not dependent on data. In the Upper Air app, these are Valid UTC Hour and Average, but these will vary for each app. Average and Valid UTC Hour tend to have standardized choices available across many apps.
Valid UTC Hour: This can be set to any whole number 0 to 23 for all of the apps.
Average: If the value for AVERAGE is none, then all points available will be plotted. Otherwise, all points in the selected interval are averaged into a single value. Available choices are as follows for all of the apps:
After selecting all data parameters as desired, the user would click the Add Curve action button inside the green bar, and the area outlined in red in Figure 2.4 would be added to show the parameters selected for Curve0. The color for Curve0 has been given the default value of red and will also be shown with a red line and/or symbols in the plot. The color assigned to this curve can be changed within the outlined box with the button labeled “rgb(255,0,0)”.
Note that the user has the option to supply whatever label they prefer to use for each curve, but the default is to create labels “Curve0”, “Curve1”, etc.
Editing a Curve
Once a curve is added, a user can go back and change any of its parameters. There are two ways to do this: by clicking directly on the selector value to be changed within the red outlined area, or by clicking “Edit Curve” inside the red outline. The user would then set any selectors to their updated values. If “Edit Curve” was used, or if the value of a multi-select (a selector that can have multiple values chosen) was altered, the user must finish the edit by clicking “Save Curve Changes”.
Defining Multiple Curves
In Figure 2.4 “Curve0” has been defined. The user can then repeat the selection of curve data parameters followed by Add Curve for as many curves as they wish to show on the plot. Each new curve will be assigned a new color.
An example of the interface screen with 2 curves defined is shown in Figure 2.5 Curve0 data parameters are shown in the red box and Curve1 data parameters are shown in the blue box.
2.5. Plot Parameters
The Plot Parameters appear in the box below the Curve Data Parameters box and apply to all curves defined in the Curve Data Parameters. The user can select the dates to be included in the plot, plus the plotFormat. Note that MATS also has an option to define QC parameters, but these are not readily supported from the METviewer database.
The plotFormat defines whether or not to show differences between curves. It has 3 options: “show matching diffs”, “pairwise diffs” or “no diffs”.
Show matching diffs takes the difference of each of the curves from Curve0. For example, if a user had defined Curve0, Curve1, Curve2, and Curve3, and selected “Show matching diffs”, the plot would have three difference curves: Curve1-Curve0, Curve2-Curve0, and Curve3-Curve0.
Pairwise diffs takes the difference of each pair of adjacent curves. So, if a user had defined Curve0, Curve1, Curve2, and Curve3, and selected “pairwise diffs”, the plot would have two difference curves: Curve1-Curve0, and Curve3-Curve2.
All difference curves calculate differences only where the data is matching.
No diffs is the default and will not include any difference curves in the plot.
2.6. Saving and Restoring Settings
When all parameters have been set as desired – including curve data parameters, plot parameters, and plot types - the user has an option to Save All Curve Settings. If selected, the user will be prompted to provide a name for the saved settings. In future the user has the option for “Restore Settings”. If selected, they will be prompted for the saved-settings name. After selecting saved-settings, all curve definitions, plot parameters, and plot type will be reset from the saved-settings definitions.
2.7. Generating a Plot
When the user is ready to generate a plot, the user would click either the “Plot Unmatched” or the “Plot Matched” action button directly below the outlined boxes with curve definitions.
Plot Unmatched will plot points for all data available for each defined curve.
Plot Matched will only plot points at x-axis values (y-axis for vertical profiles) where data values exist for all defined curves. In addition, only database values that exist at the same time period and same vertical level for all defined curves will be considered when computing y-axis values (x-axis for vertical profiles).
Hide Gaps: Note that when data points are missing, by default a line is not shown connecting the surrounding points. The user can force METexpress to draw lines between points in all cases, by selecting Hide Gaps.
Figure 2.6 shows the plot screen after the user selects Plot Unmatched or Plot Matched with a single curve defined. The same curve data as defined in Figure 2.6 is used to generate the plot in Figure 2.7, but with Hide Gaps selected.
Figure 2.8 shows a sample plot with 2 curves: Curve0 and Curve1 in red and blue, and has also added a difference curve in yellow showing the difference between Curve0 and Curve1 and error bars on all curves.
2.8. User controls on plot screen
After the plot has been produced, the user can interact with the plot in various ways. For example, the user might want to focus on one particular area of the plot to examine more detail, or the user might want to change titles, line widths, etc to produce a plot suitable for publication. Figure 2.9 highlights the tools available to interact with a plot.
The user can choose to dynamically zoom in and out on the plot. In order to zoom in on a plot, the user can click and drag a box over the area of interest on the graph, and the zoom level will be automatically adjusted to reflect the bounds of the box. To zoom back out, the user should double click on the plot or click the blue refresh button to the far right the dark blue toolbar.
METexpress plots are produced with the graphing package Plotly, and as such inherit a suite of basic graph controls from Plotly. These tools appear in the upper right corner of the plot when a user hovers their cursor over the graph area. From left to right, the Plotly tools allow a user to:
Icon 1) Save a low-resolution PNG of the plot (METexpress has a separate capability to save high-resolution PNGs, as described below).
Icons 2-3) Toggle between clicking to zoom (default) and clicking to pan.
Icons 4-5) Toggle between using a box to zoom or pan (default) and using a lasso.
Icons 6-7) Zoom in or out while maintaining the current center point of the graph.
Icon 8) Have Plotly guess what the appropriate zoom level is to best frame the plot contents.
Icon 9) Reset the x- and y- axes to their original zoom and pan settings.
Icon 10) Turn on/off lines that trace the cursor’s position on the x- and y- axes.
Icons 11-12) Toggle between only displaying the data tooltip when the cursor is over a point (default), or always displaying the nearest tooltip to the user.
Icon 13) Visit Plotly’s website.
Beyond the inherited Plotly controls, a number of buttons on the plot screen allow the user to interact with the displayed plot.
The Back button will return the user to the main app page, where the curve parameters were defined. This would allow the user to modify the plot definitions and re-plot.
The Preview button will copy the interactive plot into a separate window, to store it for comparison with future plots. The plot preview window also has the functionality to save each plot as a PDF or PNG.
The Data Lineage button displays all of the plot parameters, data queries, point values, and other metadata in a JSON format.
The Axes button allows users to customize the plot axes. This includes options to change the x- and y-axis limits, labels, label font sizes, and tick font sizes, as well as the plot legend font size and font color, and the grid line weight and color. These modifications all occur instantly, without the user having to regenerate the plot.
The Y Linear/Log button appears for profile plots, and allows users to toggle between having a linear vertical scale (the default), and a logarithmic one.
The Merge Y Axes button appears for all types of line plots that can have more than one Y axis, and allows users to condense all curves on those plots onto one Y axis. Clicking the button again will revert the graph to its default Y axes.
The Merge X Axes button appears for profile plots, and allows users to condense all curves on those plots onto one X axis. Clicking the button again will revert the graph to its default X axes.
The Equi-space X button appears for threshold plots, and allows users to toggle between having a linear x-axis scale (the default), and equally spacing all of the returned threshold values.
The Curve Styles button allows users to customize the appearance of the curves in the plot. Here, users can change each curve’s color, line style, line weight, marker style, and marker weight.
The Colorbar button appears for contour plots, and allows users to customize the appearance of the contours. Here, users can change the colorbar label, colorbar range, colorbar stride, colormap being used, the direction of the chosen colormap (normal or reversed), and whether to interpolate over any null points in the graph.
The Show/Hide button allows users to show/hide individual curves, the markers on individual curves, the error bars on individual curves (if applicable), the bars of individual histograms (if plotting histograms), the mean/median/stdev annotation corresponding to individual curves, and the entry for each curve in the plot legend.
The Edit Legend button allows users to customize the legend text for each curve. By default, the legend displays the curve label, followed by each of the curve parameters contributing to that plot trace.
The Filter Points button allows users to hide certain points on the graph, for quality control, to increase the time interval between each point, etc. Users uncheck independent variable values that they wish to hide when the Filter Points Modal appears.
The Re-sample button allows users to change the zoom level of the x-axis and then re-query the database with the new x-axis limits. This has two advantages. Firstly, if the user initially asked for too narrow a time period on a time series, they can increase the x-axis limits to their new time range of interest, click the Re-sample button, and receive a new plot with the full time range, all without having to return to the main app page. Secondly, if the user initially requested a large time period, METexpress will downsample the returned data with the Largest Triangle Three Buckets algorithm, in order to prevent huge datasets from slowing down the user’s web browser. The user can zoom in on a time period of interest, click Re-sample, and receive the full, un-downsampled dataset for that smaller time range.
The Re-cache button forces METexpress to re-query the database in the event of new data being ingested. For the most part, METexpress will cache requested plots for eight hours, and if another user requests the identical curve/plot parameters within that eight hours, METexpress will return its cached plot, instead of re-generating it. Re-cache forces it to actually re-retrieve the data from the database and remake the plot.
The Text button changes from the plot page to a text page, which displays all of the data on the graph in tables, along with statistics relating to each point, as shown in Figure 2.10 for the plot in Figure 2.8. These data tables can be exported as CSV files by clicking the “Export” button at the top (next to the “Data Lineage” button), so that users have the option of creating their own plots from the queried data.
The blue refresh button to the right of the text button will reset the graph to its original pan and zoom settings, as well as reverting any customization that the user has done to the axes or curves.
Tool-tips: On the graph itself, each point on each curve has an interactive tool-tip. When a user mouses over a point, the tool-tip pops up, providing details about that point’s x- and y-value and statistics.
Error bars: When the user clicks the “Plot Matched” action button instead of “Plot Unmatched”, the points on the graphs will display error bars, which are useful for comparing multiple curves. It should be noted, however, that error bars will not display if each point only has one database value contributing to it, as error bars are based on standard deviation, and the standard deviation of one value is zero.