9.6   Creating a gridded composite

Each Level 2 image consists of a single scene taken from the measurements made by one of the sensors during one overpass by the satellite. To study how oceanographic phenomena develop with time, you need images from several days, weeks or months, covering the same area at the same spatial resolution. Such images are usually referred to as Level 3 data, gridded data, or gridded composites. Level 3 images consist of geophysical data from more than one overpass collected together in one image, and plotted on a common geographical grid.

A step by step approach

To create a gridded composite of oceanographic data from ENVISAT Level 2 images you have to carry out the following main steps:

1. Select the temporal resolution (the appropriate time interval) for the composite.
2. Prepare the individual scenes by applying masks to select only valid geophysical data.
3. Select the area to be covered, the spatial resolution to use, and an appropriate coordinate grid for the new Level 3 image.
4. Co-register the images by resampling each scene to the chosen grid.
5. Combine all the scenes to calculate average values for each pixel over the chosen time period and write the result to a new image.

These steps are general and may be applied to oceanographic data from any satellite sensor. Here you will use the method to create a gridded composite of MERIS chlorophyll concentrations (algal_1) from a study area in the Benguela upwelling area on the south western coast of Africa.

Selecting a time interval

The number of days used to create a composite gives the temporal resolution of the gridded data. In any time series study it is important to choose an appropriate temporal resolution for the oceanographic features you want to investigate. This means reconciling two opposing considerations:

1. The need to make spatial coverage more complete by increasing the time interval used for the gridded composites.
2. The requirement for a temporal resolution that does not mask important developments in the features of interest, but is in keeping with the time scales on which these features evolve.

For a study of seasonal change a monthly (or even a 3-monthly) composite may be sufficient. Other phenomena, such as eddies, coastal upwelling events or algal blooms change on shorter time scales. Increasing the time interval used for composites beyond a few days is likely to improve image quality, but may well average out some of the changes you want to study.

The minimum number of days

Sensors onboard polar orbiting satellites do not usually have swaths of sufficient width to cover the whole of the Earth’s surface in one day. As a result there are gaps in the daily coverage, particularly at the equator. (figure (13K)).

Unless your study area is at a high latitude with overlapping swaths, you will need to combine images from several consecutive days in order to avoid gaps. This is the case for instance for the Benguela study area ( roughly 10 to 40 ^o S and 10 to 30 ^o E.

The minimum number of days required to cover a study area fully depends not just on latitude, but also on the swath width of the sensor. For instance: Meris has a wider swath than AATSR, and will therefore need a shorter time period to give full cover of the region of interest.

Is the minimum period sufficient?