Saturday, 15 July 2006

Land Degradation Surveillance: A Spatial Framework for Characterization, Research and Development.

Markus Walsh, K. D. Shepherd, A. Awiti, and T-G Vagen. World Agroforestry Centre (ICRAF), ICRAF House, PO Box 30677-00100, Nairobi, Kenya

The World Agroforestry Centre has developed a Land Degradation Surveillance Framework that is modeled on epidemiology approaches used in medical diagnostics. The framework provides a basis for:

        Rapid quantitative diagnosis of constraints to soil, plant and livestock health in a target area, including quantification of environmental and socio-economic risk factors;

        Systematic location of field experimentation to sample the diversity of conditions in an area, so that results can be generalized within the target area; and

        Baselines for scientifically-rigorous impact assessment of development interventions, controlling for spatial and temporal confounding effects.

The approach first uses freely-available satellite data and GIS databases to assess variation in the target area. A ground sampling protocol is used to assess baseline soil and vegetation conditions. Baseline ground sampling is based on 10 x 10 km blocks, which are operationally convenient for conducting ground surveys and monitoring field experiments and at the same time are large enough to sample considerable variation. The location of the blocks can be random, or based on regions or gradients of interest. The blocks are spatially stratified into sub-blocks, within each of which a cluster of 1 km radius is randomly located. The cluster consists of ten 1000 m2 plots located at random. Each plot is characterized in the field for landscape features, vegetation features (FAO land cover class and cover scores), land use, and water infiltration rate; and pooled topsoil and subsoil samples are taken. The cluster locations are also used for household surveys and more detailed sampling for crop and livestock health.

Soil, crop and livestock health are assessed using near-infrared spectroscopy (IR) on soil, crop tissue, and livestock faecal samples in conjunction with visual scores of condition. This provides low cost unbiased prevalence data at the population level on soil constraints, crop nutritional and pest/disease constraints, and livestock health and nutritional constraints. Subsets of samples are sent for specialized analysis (e.g. delta 13C to track historic land use changes; 137Cs to quantify recent erosion) and calibrated to infrared analyses to provide information on environmental risk factors at the landscape-level.

Information on constraints and risk factors are spatially interpolated by direct calibration of the georeferenced ground data to Landsat or QuickBird imagery. QuickBird imagery also provides a baseline on conditions within the blocks at the start of the project (e.g. woody cover, housing units and infrastructure, cultivated areas, surface erosion features). Response data from field experimentation (e.g. crop response to nutrient inputs) can then be related to site indices based on soil and crop tissue infrared scans. Diagnostic and advisory information (e.g. dominant constraints and their hot spots; response domains) is fed back to local extension providers and farmer community groups.

Before-after-control-impact-pair designs are used to monitor development interventions. This allows rigorous assessment of intervention impacts, correcting for spatial confounding and baseline drift.


Back to 3.5P New Methods for Large-Area Assessment of Soil Degradation - Poster
Back to WCSS

Back to The 18th World Congress of Soil Science (July 9-15, 2006)