Here, a spatial or local average of a user defined inpedendent variable is modeled. A GAM (Generalized Additive Model) or GAMM (Generalized Additive Mixed Model) is fitted here.

Data selection: Under "Data" source the user can select if he likes to use the data selected in the "Data" tab or to upload an .xlsx or .csv file. If the latter the user has to make sure that the first row of his tabular data set contains the variable/field names and select an appropriate coordinate format of the three choices: "decimal degrees", "degrees decimal minutes" or "degrees minutes seconds". If the option .csv is selected the user also has to provide the column and decimal separators as they may differ between different language versions of Excel.

Variable selection: Here the user has to specify the independent variable, its uncertainty in standard deviations (optional), the longitude and latitude variable as well as the site/ID variable. The latter is optional but recommended. If selected, a random intercept is added to the model, which is able to capture dependencies within one site.

Model options: The "Number of basis functions" governs the number of basis functions used for the approximation of the surface, so called approximate thin-plate-splines (see e.g. doi:10.1007/3-540-47977-5_2 or doi:10.1111/1467-9868.00374 for further information). A higher number delivers more exact results and enables the estimation of more complex 2-D surfaces, but uses much more computational power. In general, there is little gain from very high numbers and there is no additional gain from selecting a higher number than there are unique combinations of longitude and latitude. By selecting the corresponding check-box, a Bayesian model is possible as well. However, the computation time may be quite high for large data sets. The user can select how much MCMC (Markov Chain Monte Carlo) iterations should be computed. Higher numbers deliver more exact results at the expense of computation time (usually 1000 iterations already give a pretty usable results for testing). If a Bayesian model is selected, the user can also modify the amount of smoothing. The model already selects an optimal amount of smoothing itself, but with this option the user gets the possibility to estimate a more or less smoother surface. The model can be estimated by clicking on the "Run" button.

The estimated spatial average is displayed in the center. An export button can be found directly below, which offers various formats for plot export. With the two sliders below the map section can be selected.

After estimating the model, several options to modify the displayed plot appear on a rightern panel. If the dependent variable is range restricted between 0 and 1 or 0 and 100, the user can select this on the top of the panel. Directly below, there is the choice to restrict the display the estimates on the terrestrial or aquatic areas or the complete surface. Furthermore, the location marks (if selected the size can be influenced bya slider), grid, scale and north arrow can be removed from the map. The "Apply convex hull" checkbox marks, whether estimates of locations within the convex hull of the data points should be displayed. In general, this is indicated as extrapolation beyond areas with available data is highly speculative. The slider "Display up to max standard error" governs the area where we have a standard error below the selected mark to prevent areas with low certainty being shown. Right under the colour options, a slider "Plot resolution" is available. A higher number delivers a higher quality picture of the coloured area with less pixaleted borders but needs much more drawing time. A lower value is recommended for setting the right picture options and map section and a high value for publication.

The options on the bottom of the rightern panel enable the user to get an estimate of a specified location ("Center point latitude/longitude") within a certain radius. The estimate with standard error will appear below the x-axis in the displayed plot. For getting estimates for a batch of points the "Batch Estimes" button opens a pop-up window, where a user-specified data set can be uploaded. A local estimate including standard error and 95 percent uncertainty intervals is added and can be exported. Alternatively, these estimates can be exported into the "FRUITS"-app (not functional yet).