Testing FAO's ECONET approach in monitoring land cover changes in Pakistan
Few years ago FAO developed and tested in several African countries the ECONET approach in monitoring
land cover changes. It builds upon a systematic sampling of the area of interest against
the traditional "wall-to-wall" mapping, with the advantage of reducing drastically time, costs and overall work for
regularly monitoring the same area. The recent availability of Sentinel-2 data has suggested FAO and SUPARCO to test
such approach in a pilot area of Pakistan.
Land cover data
Land-cover data and the relevant changes are of primary importance in satisfying the ever-increasing
demand for reliable data to support studies and application such as natural resources evaluation and
conservation, agriculture development, improvement of environmental policies. Data are essential for
the comprehension and analysis of natural and human driven phenomena, including for example climatic
change, land planning, disaster management and agriculture development monitoring.
Traditionally, land cover mapping is performed adopting the so called "wall-to-wall" approach, which
means mapping the entire area, from sub-national to regional and global scale. This approach results
in accurate and reliable data, but with time and costs that makes it not sustainable in many countries
and areas of the world.
The ECONET approach in monitoring land cover changes has been firstly proposed by FAO few years ago
following the success of the Global Land Cover Network Programme. It is sample based and is a cost-effective
solution to assess the land cover changes accuracy using high resolution Google Earth images: it has
proved consistent in results versus a more traditional "wall-to-wall" mapping, but with the advantage of
reducing drastically time, costs and overall work for implementation. With this sampling approach,
statistical estimates can be derived while retaining the advantages of fine spatial detailed data.
Systematic sampling has been successfully used for regional, continental, and global tree cover change
An ECONET based mapping and analysis has two major outcomes: 1) the production of very detailed information
on land cover, based on photo-interpretation of a wide number of sample areas evenly distributed, and 2) the
setup of a collaborative community involved to implement and spread a fully harmonized approach, supporting
and making accessible reliable and comparative land cover data required by local, national and international
activities. The ECONET approach heavily relies on the FAO's Land Cover Meta-Language (LCML) standard (ISO)
and Land Cover Classification System (LCCS) tools.
- Create a database for the description of land cover with detailed information to be used for a wide
range of activities;
- Serve as multi-statistical source of information at any level of detail or complexity. Using specific
software any end-user will be able to extract statistical information for a given geographical area;
- Work as a reference data-set for any future mapping activity;
- Contribute to create a common shared awareness on the importance and value of natural resources.
A pilot area of Pakistan is being tested in the application of the ECONET approach taking advantage of
the Sentinel-2 mission data that are available freely at a medium high spatial resolution (10m). Few points:
- The initiative in Pakistan is mainly for demonstration purpose;
- Only agriculture is considered (limited number of classes);
- Work done only at sub-country level (limited number of tiles);
- Only local experts involved (no further interaction to increase the level of information);
- Focus on change detection to monitor agriculture expansion.
The data production will be based on the photo-interpretation and analysis of very high resolution satellite
imageries of 10 km x 10 km portion of land surface (tiles), distributed according to a fixed grid. The design
of the sample grid has to ensure a sound statistical basis for the validity of the results and will follow
basic zonation criteria: the tile density for Pakistan will be at ½ x ½ degree intersections.
Dot density per tile
A secondary grid inside the 10 by 10 sample tiles has been generated. Each dot (square size) has a size of 25 by
25 m and are sited at a distance of 250 m each. In total 1880 points are generated for each tile. A less dense dot
grid is also generated at a distance of 500 m, dot size of 35 x 35 m, for a total number of dots of 480 per tile.
The analysis of the changes is carried out by using Google Earth high res images and recent Sentinel-2 images
(10 m). With LCCS3, any land cover features can be described using a set of attributes (elements). Each polygon in the
database is not indicated by a traditional class (as part of a predefined legend) but by a series of attributes
(elements of LCML language). This will ensure that future analysis to verify change trends will be made on the
same areas, thus permitting alid statistical estimation of changes.
Sentinel-2 image: false color composite (843)
of prevalent agriculture area (vegetation in red)
A preliminary legend of 22 classes of land cover has been generated based on LCML/LCCS "object oriented" standard
Each category is systematically described both with a traditional text description (see table above) but also with
a more accurate graphical schema that shows the combination of LCML elements (with their properties and characteristics)
used to generate each of the land cover classes according to the rules defined in the ISO standard UML of the Land
Cover Meta-Language. Here is an example of class description, in particular the Irrigated crop and Mango orchard:
Description: "Irrigated crop and Mango orchard"
LC class Schema:
Screenshot of high res image (Google Earth) for the class above:
Sep 2, 2015 - A new amendment (no. 3) was signed by USDA and FAO to formalize a no-cost
extension of the project GCP/PAK/125/USA from Oct 1, 2015 to Sep 30, 2016.
Aug 8, 2016.
E-bulletin 11, Apr-Jun 2016.
Aug 5, 2016.
E-learning courses on agriculture monitoring and statistics.
Aug 5, 2016.
FAO's ECONET approach and land cover change in Pakistan.
Aug 5, 2016.
Training on Carbon Sequestration.
Aug 5, 2016.
CRS Punjab bulletins Apr 2016.
Apr 22, 2016.
Cotton Production & Analytics.
Mar 1, 2016.
Testing Sentinel-2 images for crop monitoring in Pakistan.
Feb 20, 2016.
Crop Information Portal training.
Jan 20, 2016.
Sugarcane Production & Analytics.
Jan 20, 2016.
Training on REDD+ Technology for Forest Department Officers of Punjab.
Oct 29, 2014.
Agriculture Information System project: roll-out workshop.
Aug 13, 2015.
RS/GIS Training in Forest Management, Lahore, 4-8 May 2015.
FAO Pakistan website
Forum: "Pakistan Agriculture Sector". >>>>
SUPARCO, the National Space Agency of Pakistan
Government of Sindh: Agriculture Department
Agri Punjab >>>>