Aggregation of weather indicators

From Agri4castWiki
Revision as of 10:51, 22 October 2021 by Hendrik2 (talk | contribs) (Masks)
Jump to: navigation, search

General description

The aggregation of grid weather towards administrative regions is executed for different input datasets. For each of the datasets the procedure is repeated using different weighing methods and different output resolutions. A dedicated software package called Aggregator is developed to do all calculations.

Input datasets

In the MCYFS several grid weather datasets are available: grid weather based on station observations (observed grid weather) and several ECMWF products (see acquisition). All sets are available in grid format. A selection of these datasets is aggregated to regions: observed grid weather (table WEATHER_OBS_GRID) and (re-)forecasted weather of the ERA5 and HIS/OPE model (table WEATHER_<MODEL>_GRID).

Weighing methods

Grid cells (or part of grid cells) that fall inside a region are used to calculate the aggregated value of a specific weather indicator. Besides using all grid cells in a region, the aggregation is repeated using different selections of grid cells in each region. The selection is based on landcover masks. In other words: the aggregated indicator emphasizes the phenomena that are most relevant for the selected mask while neglecting the phenomena not relevant for the selected mask.


For Europe weather data is aggregated to land cover types only:

  • No mask, all mixed (1000)
  • Irrigated cropland (1001)
  • Rainfed cropland (1002)
  • Arable Land (irr+rainfed) (1003)
  • agricultural areas (arable + mosaic > 50%) (1004)
  • rice (1009)
  • Fruit trees and berry plantation (1023)
  • Olives Groves (1024)
  • Pasture Forage and Grassland (1025)
  • Winter Spring Crops (1040)
  • Summer Crops (1041)
  • Fodder Crops (1045)
  • Permanent Grassland (1046)

These masks are based on several landcover products like CORINE land cover and Globcover.

Output resolutions

The output is calculated for different types of regions.

In case of Europe:

  • 4 levels of administrative regions (NUTS0-3)

In case of other regional windows like Russia and Kazakhstan (RUK):

  • 3-4 levels of administrative regions (RUK has a specific sub national level grouping level 1 regions)

The results are written to the tables

Replace <MODEL> with one of the model abbreviations for which aggregated indicators are calculated (ERA5, HIS/OPE).