Updated December 1998

Remote Sensing Centre Series

Erosion mapping using high-resolution satellite data and geographic information system. Pilot study in Brazil

by D. Lantieri, J.F. Dallemand, R. Biscaïa, S.Sohn, R.O. Potter
150 pp, 19 figures, 16 tables
RSC Series No. 56, FAO, Rome 1990.

Summary

The objective of this pilot study was to define an operational method for erosion susceptibility mapping at scales between 1:50,000 and 1:100,000 to assist the planning of a World Bank land conservation project for the whole State of Paraña, Brazil.

The technical study consisted initially of a review of the main existing methods used for erosion assessment such as the USLE or other parametric model-based methods implemented at FAO or ITC. It was shown that all these methods had their limitations and could not be applied operationally in Paraña conditions at the scales required by the study and also would not be suitable for SPOT satellite information. Finally, it was found that the best approach for erosion susceptibility mapping was to apply an empirical model integrating land conditions of Paraña. This model integrated, through a Geographic Information System (GIS) and within homogeneous rainfall zones, a land use and land management map, a soil map, a slope length map and a slope angle map. The model was built by taking into account results of hundreds of erosion measurements made on experimental plots together with extensive field experience obtained by the research institutes of Paraña, (IAPAR, EMBRAPAO ITCF), The types of parameter mappable on high-resolution satellite data (SPOT) were also utilized in the model construction.Thus:

• the land use and management maps were produced from digital and visual analysis of SPOT XS and PA data. Minimum distance supervised classification of SPOT XS data permitted the identification of the main land use and agriculture types such as forest, annual crops, coffee and sugar cane, as well as of "direct planting" fields, the only ones not being ploughed at the time of image acquisition (end of field preparation). In fact the choice of this date was found to be essential to allow the production of reliable maps. Differentiation was also made between high and low cover pastures by means of vegetation index thresholding. In the end, some visual analysis (textural and contextual) had to be made to discriminate better between some classes (e.g. sugar cane and pasture). Visual interpretation of SPOT PA digitally enhanced data (local optimization followed by a high pass filter), acquired at the end of the cropping season, was used to identify existing terraces and therefore to differentiate good and bad management fields.
• the soil map was produced by reporting visually on a SPOT stereoscopic pair (one PA and one XS image) the soil limits of the units of the minutes of the 1: 300,000 soil map drawn originally on 1:70,000 old aerial photos.
• the slope angle map was made automatically after digitizing a contour line map from available topographic 1:50,000 maps, using a GIS software package (ARCINFO).
• the slope length map was produced by measuring manually, along the direction of the slope, the length of each polygon of the angle map.

In-the discussion of the results, it is shown that the rainfall factor was not taken as an input mapping level of the model itself, since, in practice, the relationship between land use and management, soil and slope is the same as when assessing erosion susceptibility within homogeneous rainfall zones in Paraña. Some additional research was also undertaken on the modelling and mapping of real erosion and of "soil conservation measures recommended" using the same input mapping levels of the erosion susceptibility map.

The economic study was performed at two levels: On a first level, the analysis consisted of a cost-benefit comparison between six different remote sensing methods which could be used for erosion mapping and monitoring. These six methods were differentiated according to the data (SPOT, aerial photos) or method of analysis (visual, digital).

It is shown that the most cost-effective method is the one used in the pilot study and therefore based on the GIS integration of maps produced from digital and visual analyses of SPOT XS and SPOT PA digitally enhanced data preferably taken in stereoscopy, and where the slope map would be derived from digitization of topographic maps. The cost of this method, which includes data, manpower and renting equipment, ranges from US$ 5.3 to 8.06 per km2 (using FAO standard rates) depending on the type of consultancy (national or international). Benefits taken into account are accuracy of mapping, time for erosion map production and potential for other applications.

On a second level, the economic impact of producing an erosion susceptibility map was assessed for the whole State of Paraña. The benefit of having an erosion susceptibility map was evaluated as the economic value of the time saved in a soil conservation programme thanks to the use of erosion maps; this time was estimated at four months per year, the value of which is equivalent to US$19.7/ha of rehabilitated land. For the whole State of Paraña, the use of an erosion map represents an overall benefit of US$75 million in the hypothesis of the Paraña rural programme where 3.8 million ha would be rehabilitated. This estimate of the benefit was made after calculating:

1. the saving of the costs of soil erosion to agricultural production (cost of nutrients contained in soil losses) and to the environment (budget for cleaning drinkable water and for silt removal from state infrastructures) (US$38.22/ha/year);
2. the increase in agricultural productivity due to land conservation practices and particularly to direct planting (US$24/ha/year);
3. the cost of state research and rural development programmes related to soil conservation (US$l/ha rehabilitated);
4. the cost of mapping, calculated from the cost benefit analysis mentioned above (US$0.4/ha rehabilitated).

It should be noted that this overall benefit for the state is not in direct income but is diffused over the regional economy.

The results of this pilot project are now being integrated in the implementation of the Paraña World Bank- funded land conservation programme.

The above publication is available from:
Chief, FAO/SDRN
Viale delle Terme di Caracalla
00100 Rome, Italy
(e-mail: changchui.he@fao.org)