General approach for forecasting
Pure deterministic results, with a single value for each instant of time, are avoided. The models generate forecasts from a set of input scenarios to the hydrological system on which the forecast is made (rainfall, flow, etc.), resulting in a set of results for each instant.
Different results corresponding to different models can also be generated, since, in EDAPHI, the concepts of multimodel are incorporated, in addition to the hypermodel.
The results can be analysed statistically or the concept of maximum allowable divergence can be used, whereby the forecasting lead time is given for such a time that the forecast range (deviation- the difference between extreme hypotheses at a given instant) is such that, treated as uncertainty in the forecast, it is considered acceptable from a practical point of view.Thus, the forecast lead time would be conditioned by:
Response time of the hydrological system
Maximum allowable divergence, associated with the uncertainty in the inputs to the system
General approach for simulation
An accurate forecast will depend largely on the simulation, and this can be contrasted, in the past time, at those points with a measured variables as flows, or levels.
EDAPHI incorporates several possibilities of selection of the past scenario, that can be defined by some input variables and some simulation parameters, as well as several methods of parameter calibration
For details, refer to the book "EDAPHI ..."
General approach for model developments
The specialized computer solutions are developed with free, free computer tools. In some cases they will be open source. Commercial tools can also be incorporated.
The developments that are made on this basis have important advantages, among which the following can ve highlighted:
They do not require fee licenses.
It is based on widely used solutions, so the transfer of results to users and developers is easy.
There is a large user community that offers open source.
Complementary languages and tools
Free software or open source tools are used:
Hec Programs (http://www.hec.usace.army.mil/software/): Ras, HMS, DSS-Vue
Programming in Python or one of its specialized variants
The main language is Python (https://www.python.org), with some of its variants or modes of use associated with specific applications (PyGrass in the Grass GIS or Jython in Hec-Dss-Vue, for example). Parts of code are also used in VBA, VB Script or Html.
The environment has been designed so that the user can configure the applications with formats and tools that are commonly used, such as the MS-Excel office tool.
When an application requires some GIS layer as data, the Shape format (SHP) will be adopted when it is vector type, or ASC (text) when it is matrix (raster). Both are ArcGIS software formats.
Variables (time series) as inputs are read in spreadsheet formats, either as delimited text (CSV) or as MS-Excel spreadsheet (XLSX).
Data stored in other file formats can also be accessed, including databases.
A particular case is that of the series stored in DSS files, used with Hec applications.
GRIB and AREA files can be read
The configuration parameters of the applications, which define their operation or some modeling parameter, are stored in Excel spreadsheets. Most of the modeling parameters, except those considered relevant, are included in the specific files of the auxiliary tool (case of Hec-HMS, for example), as well as configuration data of the application.
Presentation of results. Outputs
The same formats are used as for the entries
The results will be stored in XLSX or CSV format analogous to the inputs. If neccessary, spreadsheets (xlsx) are prepared with tables and graphs that depend on data linked to CSV files.
In addition to the tables contained in the XLSX spreadsheets mentioned above, synthesis results are available in HTML format.
In addition to the graphs that can be found in XLSX spreadsheets, the applications generate PNG-format graphs of time series.
The programs developed do not have graphical interfaces.
Their execution is done by line commands, which can be done with CMD files of the operating system. Thus, the programs can be called from other modules, other than EDAPHI, according to an operating system standard.
Automation can be done with the operating system task scheduler. The design of the programs takes this into account.
Installing the programs
The only limitation will be in the characteristics of the tools with which the solutions are developed (Python, HEC software, ...). But the selection of software, on which EDAPHI is based, facilitates the installation in most of the computers with Windows 10 operating system, even earlier (no limitations are known), as well as in other platforms (depending on the case of application and the necessary complementary tools).
For details, refer to the book "EDAPHI ..."