Skip to content

Latest commit

 

History

History
154 lines (123 loc) · 5.92 KB

README.md

File metadata and controls

154 lines (123 loc) · 5.92 KB
Raw

Himan

Himan is a weather parameter post processing tool.

Properties

  • Forecast provider agnostic post processing
  • Able to produce several useful weather parameters not produced by the forecast models themselves
  • Main design principles are performance and maintainability
  • Written in C++11
  • Partial support for GPU's through Cuda
  • Actively developed and supported by the Finnish Meteorological Institute
  • yum repository available for RHEL8 and RHEL9
  • Licensed under MIT license

Parameter list

Himan is able to produce the following parameters, given required source data:

  • cloud code
  • dewpoint
  • fog type
  • pressure at hybrid level
  • (metric) height of hybrid level
  • icing in the atmosphere
  • icing at sea (sea spray icing)
  • stability indices
    • k-index
    • lifted index
    • vertical totals index
    • cross totals index
    • total totals index
    • showalter index
  • height of 0 and -20 isotherms
  • precipitation form using pressure level information (less accurate)
  • precipitation form using hybrid level information (more accurate)
  • precipitation type (continuous, showers)
  • relative humidity
  • accumulated precipitation and radiation to rates and powers
  • potential temperature
  • pseudo-adiabatic potential temperature
  • equivalent potential temperature
  • vertical velocity in meters per seconds
  • speed and direction from vector components
  • density of dry air
  • liquid and solid precipitation from mixing ratios
  • absolute humidity
  • cape and cin & lpl lcl, lfc, el height and temperature
    • surface based
    • lowest 500m
    • most unstable
  • weather symbol (two different versions)
  • monin-obukhov length
  • wind gust
  • qnh pressure
  • cloud ceiling in feet
  • probability of thunderstorms
  • snow depth
  • toplink winter weather indices 2 & 3
  • modified clear air turbulence indices 1 & 2
  • lapse rate
  • cloud types
  • visibility
  • fractiles from ensemble
  • probabilities from ensemble
  • probability of precipitation
  • turbulent kinetic energy
  • unstaggering of Arakawa C grids
  • fog intensity
  • wind (bulk) shear
  • effective bulk shear
  • storm relative helicity
  • energy-helicity index
  • bulk richardson number
  • frozen soil depths (bottom and top)
  • growing degree days
  • frost sum
  • tropopause height
  • cloud layers for taf generation
  • smartsymbol (weather symbol)
  • weather number
  • low visibility procedures probability
  • snow drift
  • frost probability
  • tropopause height
  • nearby weather (symbol)
  • cloud top
  • surface roughness
  • potential evaporation
  • precipitation phase
  • and many more...

Supported file formats

Himan supports a number of different input/output file formats:

  • grib version 1 and 2
  • querydata (FMI internal dataformat)
  • geotiff
  • csv for point data

Data can be read from and written to either traditional POSIX compatible file system or Amazon S3 and any compatible system.

Architecture

Himan is split into four separate but not independent components:

  • executable (himan-bin)
  • library (himan-lib)
  • plugins (himan-plugins)
  • scripts (himan-scripts)

The binary (himan) is a simple frontend and does little else but parse the command line options and initiate the execution of plugins listed in the configuration file.

The library (libhiman.so) contains common code that is used by the plugins. This common code includes for example interpolation routines, json parser, metadata classes such as parameter and level, meteorological formulas and so forth.

The plugins (libX.so) contain the actual core and idea of Himan. Each plugin is a shared library that exposes only one function that himan executable is calling. Everything else is free game for the plugin. So there is a very large degree of freedom for a plugin to do whatever it needs to do. All plugins share code from a common parent class which reduces the amount of boiler plate code; the common code can be overwritten if needed.

Plugins are split into two types: helper (auxiliary) plugins and core ("compiled") plugins. The helper plugins provide several necessary functions for the other plugins to interact with the environment. These functions are for example fetching and writing data, accessing databases, storing and fetching data from cache and so forth. Two separate but important helper plugins are hitool and luatool. Hitool exposes functions that examine the state of the atmosphere in a certain way. For example it can be used to get minimum/maximum value of some parameter in a given vertical height range. Height can be specified in meters or in Pascals. Luatool plugin is a shallow lua-language wrapper on top of Himan. With luatool one can create lua-language scripts which are much faster to write and more flexible than the C++-based plugins, especially if the given task is light weight.

The purpose of the core plugins is to provide some output data given required input data. The range of operations stretches from very simple (calculate radiation power from accumulation) to more complex algorithms (determine the precipitation form). Certain plugins that operate with hybrid levels are also optimized for Nvidia GPU processors. This an addition to the normal CPU based functionality: Himan does function perfectly without GPU's.

Himan also supports scripting through lua: for some post processing operations building a new cpp-plugin is a too heavyweight operation. For this reason Himan has a plugin that interprets lua-language scripts.

The name

Himan stands for HIlojen MANipulaatio, "grid manipulation" in Finnish.

Getting started

A docker repository at quay.io for Himan:

https://quay.io/repository/fmi/himan

docker pull quay.io/fmi/himan

Getting started

Contributing

  • Feel free the submit issues or feature requests
  • Pull requests are accepted
  • CLA is required for code contribution

Communication

You may contact us from following channels:

Our public web pages: http://en.ilmatieteenlaitos.fi/open-source-code