climate gradients in northern Thailand could be hypothesized based on the fact that
the major mountain chains run rather perpendicular to the dominant monsoon wind
direction. Another area which demands caution is the prediction of Luvisols via
topographic height. Basic intrusions, which were decisive in the development of
Luvisols in Bor Krai, could potentially occur at any elevation in other training
areas. Soil predictions using the random forest approach are very promising, as this
approach requires less training data than the classification tree and maximum
likelihood approaches. In particular, here the introduction of gamma-ray data,
which contained real soil information, increased the performance of the random
forest-based map. In combination with the catena concept, the random forest
approach appears suitable for all scales; however, a sound validation is still lacking.
In particular, the high weighting given to topographic information can lead to an
exceptional appearance of the resulting soil map; therefore, topographic rules used
to determine RSG have to be validated in greater number at petrographically
different sites.
2.4 How Useful Are Ethnic Minority Soil Knowledge Systems?
Case Studies from Vietnam and Thailand
4
A 300 % increase in population over the last 30 years, combined with a lack of offfarm job opportunities, has aggravated land scarcity in the rural mountainous areas
of north-west Vietnam and north-west Thailand, leading very often to land use
intensification. Swidden farming has been replaced by permanent cropping, with
maize being the most lucrative cash crop, but one that has dramatic environmental
effects, on- and off-site. Several studies (Dung et al. 2008; Pansak et al. 2008;
Chaplot et al. 2005; Toan et al. 2004; Wezel et al. 2002) have highlighted that it is
essential for more sustainable land use systems to be introduced in these areas, in
order to better protect the environment.
Farmers are aware that intensive maize cultivation is closely connected with soil
erosion and decreases in soil fertility, and this problem has also been picked up by
governmental agencies. However, the search for measures and policies to address
this problem has mostly been based on simple economic criteria, and has thus
disregarded any efficient comprehensive analysis of the natural and socio-economic
context,
5 neglecting local people and communities in terms of an assessment of
their needs and interests, and failing to integrate indigenous knowledge – at least in
4 This section was prepared by Gerhard Clemens, Ulrich Schuler, Heinrich Hagel, Bui Le Vinh and
Karl Stahr.
5 An example of this can be found in a report of farmers in Ye ˆn Cha ˆu district commune
participating in a workshop in 2009, who said there was pressure on the commune to replace a 7
year-old teak plantation on a hill top, planted as part of a re-forestation program, with a rubber
plantation. This pressure to grow rubber in Son La province took place before any feasibility study
had been carried out.
64
K. Stahr et al.
the major mountain chains run rather perpendicular to the dominant monsoon wind
direction. Another area which demands caution is the prediction of Luvisols via
topographic height. Basic intrusions, which were decisive in the development of
Luvisols in Bor Krai, could potentially occur at any elevation in other training
areas. Soil predictions using the random forest approach are very promising, as this
approach requires less training data than the classification tree and maximum
likelihood approaches. In particular, here the introduction of gamma-ray data,
which contained real soil information, increased the performance of the random
forest-based map. In combination with the catena concept, the random forest
approach appears suitable for all scales; however, a sound validation is still lacking.
In particular, the high weighting given to topographic information can lead to an
exceptional appearance of the resulting soil map; therefore, topographic rules used
to determine RSG have to be validated in greater number at petrographically
different sites.
2.4 How Useful Are Ethnic Minority Soil Knowledge Systems?
Case Studies from Vietnam and Thailand
4
A 300 % increase in population over the last 30 years, combined with a lack of offfarm job opportunities, has aggravated land scarcity in the rural mountainous areas
of north-west Vietnam and north-west Thailand, leading very often to land use
intensification. Swidden farming has been replaced by permanent cropping, with
maize being the most lucrative cash crop, but one that has dramatic environmental
effects, on- and off-site. Several studies (Dung et al. 2008; Pansak et al. 2008;
Chaplot et al. 2005; Toan et al. 2004; Wezel et al. 2002) have highlighted that it is
essential for more sustainable land use systems to be introduced in these areas, in
order to better protect the environment.
Farmers are aware that intensive maize cultivation is closely connected with soil
erosion and decreases in soil fertility, and this problem has also been picked up by
governmental agencies. However, the search for measures and policies to address
this problem has mostly been based on simple economic criteria, and has thus
disregarded any efficient comprehensive analysis of the natural and socio-economic
context,
5 neglecting local people and communities in terms of an assessment of
their needs and interests, and failing to integrate indigenous knowledge – at least in
4 This section was prepared by Gerhard Clemens, Ulrich Schuler, Heinrich Hagel, Bui Le Vinh and
Karl Stahr.
5 An example of this can be found in a report of farmers in Ye ˆn Cha ˆu district commune
participating in a workshop in 2009, who said there was pressure on the commune to replace a 7
year-old teak plantation on a hill top, planted as part of a re-forestation program, with a rubber
plantation. This pressure to grow rubber in Son La province took place before any feasibility study
had been carried out.
64
K. Stahr et al.