eumetsat logo     bc logo     fub logo
empty





Generation of IASI/AVHRR Data Set

IASI Sample Classification

The data set of IASI samples is derived by visual analysis of globally distributed data.
The data set serves as input for the optimisation and validation of the IASI cloud detection. Based on the output of the IASI cloud detection tests, each collected IASI sample will be classified concerning its cloudiness as being a clear-sky, partly cloudy and cloudy footprint based on the visual inspection of the AVHRR pixels covered by an IASI footprint.

In order to ensure the representativeness of the data-set for the natural variability of those parameters which affect the radiation measurement from a space borne instrument additional classification are performed for each IASI footprint:

  • the surface type (discrimination of 20 types),
  • the climate zone (Köppen classification over land, geographical bands over sea),
  • the season and
  • day and night discrimination.

Cloudiness

For the assessment of the cloudiness of an IASI sample four cloud classes was used for the visual classification of the collected IASI samples:
  •   clear - if the shape of the IASI footprint contains exclusively AVHHR pixels which are unambiguously cloud free, 28% of all cloudy samples,
  •   partly cloudy low - if the shape of the IASI footprint contains both cloudy and cloud free AVHRR pixels, less than ~ 20% cloudy pixels, 26% of all cloudy samples ,
  •   partly cloudy high - if the shape of the IASI footprint contains both cloudy and cloud free AVHRR pixels, more than ~ 20% cloudy pixels, 26% of all cloudy samples,
  •   cloudy - if the shape of the IASI footprint contains exclusively AVHHR pixels which are unambiguously cloudy, 20% of all cloudy samples.
totop

Surface Types

The surface type database which was used as ancillary information in this project is the IGBP scene types in the CERES/SARB surface map. The data are freely available from the IGBP Website or from the MODIS Website.
A technical description of the dataset, as it is used in the IAVISA tools is given in chapter IGBP Surface Type Map.

The binary file IGBPa_1198.map was downloaded.
The map properties are:

  • 8-bit binary data made on an SGI workstation
  • size is 2160 points in longitude, 1080 points in latitude or 1/6 degree equal angle
  • begin at the North Pole, Greenwich Meridian

The web pages explain the use of the International Geosphere/Biosphere Programme (IGBP) scene types in the CERES/SARB surface map. The map below shows the 18surface types used by the SARB group to identify surface properties of a given region. Scene types were delineated by IGBP. This map is determined using a 1km IGBP scene types supplied by USGS. An 18th scene type (Tundra) is added to distinguish the rocky/barren scene of northern climes vs. that of other deserts. These scene types map the geographic region to long wave and shortwave spectral emissivity and albedo properties defined within the spectral ranges of the Fu & Liou Code. The spectral properties page has a complete description of data source. The distribution of the surface types in the map is as follows:

ID
Surface Type
Pixels
Selected Samples
1
Evergreen Needle Forrest
33170
603
2
Evergreen Broad Forrest
35114
876
3
Decid. Needle Forrest
10500
89
4
Decid. Broad Forrest
11366
324
5
Mixed Forest
35405
602
6
Closed Shrubs
10817
329
7
Open Shrubs
78789
1484
8
Woody Savannas
46660
768
9
Savannas
28638
656
10
Grassland
42696
886
11
Wetlands
5789
147
12
Crops
58910
1294
13
Urban
275
44
14
Crop/Mosaic
60314
1436
15
Snow/Ice
272144
1443
16
Barren/Desert
57968
1379
17
Water
1517554
12150
18
Tundra
26691
413

totop

Climate Zones

koeppenThe most widely used system of climate classification is that developed by the German climatologist Koeppen (1936) - virtually all more recent classifications are refinements or variantions of the "Koeppen system". The Koeppen climate classification distribution grid derived by the Food and Agriculture Organization (FAO) from the International Institute for Applied Systems Analysis (IIASA) data sets enables a determination of the general climate of any geographical entity (national or sub-national). Koeppen categories are based on the annual and monthly averages of temperatures and precipitation. The map data are freely available.

Five major climatic types are recognized in this system, each type being designated by a capital letter:

     
Selected Samples
A Tropical Moist Tropical Climates are known for their high temperatures all year round and for their large amount of annual rainfall.
7499
B Dry Dry Climates are characterized by little rain and a large daily temperature range. Two subgroups, S – semi-arid or steppe, and W - arid or desert, are used with the B climates.
3237
C Temperate In Humid Middle Latitude Climates land/water differences play a large part. These climates have warm, dry summers and cool, wet winters.
8150
D Cold Continental Climates can be found in the interior regions of large land masses. Total precipitation is not very high and seasonal temperatures vary widely.
2205
E Polar Cold Climates describe this climate type perfectly. These climates are part of areas where permanent ice and tundra are always present, and only about four months of the year have above freezing temperatures.
3832

No-data gaps in the Köppen map are filled by the software with Fallback values according to the following scheme based on the pixel’s latitude value.
Lat Range
Regions
Köppen Code
-90° ÷ -66.5°
South Pole ÷ Antarctic Circle
E - Polar
-66.5° ÷ -23.5°
Antarctic Circle ÷ Tropic of Capricorn
C - Temperate
-23.5° ÷ 23.5°
Tropic of Capricorn ÷ Tropic of Cancer
A - Tropical
23.5° ÷ 66.5°
Tropic of Cancer ÷ Arctic Circle
C - Temperate
66.5° ÷ 90.0°
Arctic Circle ÷ North Pole
E - Polar

totop

Seasons

In the context of the current study, climatological seasons are used. I.e., for the Northern hemisphere, spring begins by convention on 1 March, summer on 1 June, autumn on 1 September and winter on 1 December.
Start Date
Day of year
Northern
Southern
1 March
59
Spring
Autumn
1 June
151
Summer
Winter
1 September
243
Autumn
Spring
1 December
334
Winter
Summer
Season
Selected Samples
Spring
5862
Summer
6930
Autumn
6662
Winter
5469

totop

Day-Night classification

The information about day or night condition during time of the measurement was taken from the Scene Meta data. If the sun zenith angle is below 90° a night measurement is assumed, a day measurement otherwise.
Daytime
Selected Samples
Day
13455
Night
11468

totop