Validation: Soil maps, based on transects, grid-based randomization, maximum
likelihood and classification trees, were validated by comparing them with the
reference soil maps and independent sampling points. The reference maps included
the maximum amount of information available – that drawn from soil catenas,
sampling points along local trails, areas of low point density or high soil variability,
reference profiles, LANDSAT and SPOT images, and topographic information. The
reference soil maps were manually created based on expert knowledge and using
ArcGIS 9.2. In addition, for each area 15 % of all the sample points were randomly
selected as validation points and exclusively used for this purpose, resulting in 55
for Mae Sa Mai, 30 for Huai Bong and 55 for Bor Krai. In addition, re-classified
radiometric data were used for validation in Bor Krai. A validation of the regional
map is currently ongoing using a regional transect (Warber 2008), as well as
selected watersheds.
2.3.3 Results
Reference soil maps: For more than 75 % of the mapping points, clay illuviation
was identified as the dominant soil forming process, with soils mainly classified as
Alisols and Acrisols. Less frequent RSGs found were Cambisols, Umbrisols and
Regosols, while the remaining soil types mapped (Anthrosols, Chernozems,
Ferralsols, Fluvisols, Gleysols, Leptosols and Technosols) each represented less
than 2 % of all the sampling points. The reference soil maps revealed that soil cover
in the granitic Mae Sa Mai area is dominated by Acrisols (84 %), followed by
Cambisols (9 %), Umbrisols (4 %) and Technosols (2 %). Anthrosols, Chernozems,
Gleysols, Leptosols, and water bodies are present in the remainder of the area. In
contrast to Mae Sa Mai, Alisols prevailed in the Huai Bong sandstone area (77 %),
followed by Cambisols (13 %), Regosols (9 %), Leptosols (2 %) and Fluvisols
(0.1 %). In Bor Krai, in limestone and claystone areas the predominant mapping
units found were Alisols (64 %), Acrisols (27 %) and nudilithic Leptosols (limestone outcrops – 5 %), while Cambisols, Chernozems, Ferralsols, Fluvisols,
Gleysols, Leptosols, Luvisols and Umbrisols each comprised less than 1 %.
Local knowledge-based soil maps: The study’s elicitation of local knowledge
revealed that farmers had a good knowledge of the local soil diversity, with soils
differentiated according to soil color in all three areas. In Mae Sa Mai, the six
different local soil types were classified using topsoil thickness as a criterion also,
and in Huai Bong, the five local soil types were identified based on topsoil thickness
and stone content. Finally, in Bor Krai, eight different soil types could be distinguished, with bulk density used as an additional criterion. All the local soil maps
produced provided an overview of the respective main soil types, plus background
information on each local soil type such as crop suitability, water infiltration and
erosion risk, as provided by the farmers. The mapping of local soils and their
properties was the cheapest and most rapid method to use, but the quality and
quantity of local knowledge varied among the villages.
56
K. Stahr et al.
likelihood and classification trees, were validated by comparing them with the
reference soil maps and independent sampling points. The reference maps included
the maximum amount of information available – that drawn from soil catenas,
sampling points along local trails, areas of low point density or high soil variability,
reference profiles, LANDSAT and SPOT images, and topographic information. The
reference soil maps were manually created based on expert knowledge and using
ArcGIS 9.2. In addition, for each area 15 % of all the sample points were randomly
selected as validation points and exclusively used for this purpose, resulting in 55
for Mae Sa Mai, 30 for Huai Bong and 55 for Bor Krai. In addition, re-classified
radiometric data were used for validation in Bor Krai. A validation of the regional
map is currently ongoing using a regional transect (Warber 2008), as well as
selected watersheds.
2.3.3 Results
Reference soil maps: For more than 75 % of the mapping points, clay illuviation
was identified as the dominant soil forming process, with soils mainly classified as
Alisols and Acrisols. Less frequent RSGs found were Cambisols, Umbrisols and
Regosols, while the remaining soil types mapped (Anthrosols, Chernozems,
Ferralsols, Fluvisols, Gleysols, Leptosols and Technosols) each represented less
than 2 % of all the sampling points. The reference soil maps revealed that soil cover
in the granitic Mae Sa Mai area is dominated by Acrisols (84 %), followed by
Cambisols (9 %), Umbrisols (4 %) and Technosols (2 %). Anthrosols, Chernozems,
Gleysols, Leptosols, and water bodies are present in the remainder of the area. In
contrast to Mae Sa Mai, Alisols prevailed in the Huai Bong sandstone area (77 %),
followed by Cambisols (13 %), Regosols (9 %), Leptosols (2 %) and Fluvisols
(0.1 %). In Bor Krai, in limestone and claystone areas the predominant mapping
units found were Alisols (64 %), Acrisols (27 %) and nudilithic Leptosols (limestone outcrops – 5 %), while Cambisols, Chernozems, Ferralsols, Fluvisols,
Gleysols, Leptosols, Luvisols and Umbrisols each comprised less than 1 %.
Local knowledge-based soil maps: The study’s elicitation of local knowledge
revealed that farmers had a good knowledge of the local soil diversity, with soils
differentiated according to soil color in all three areas. In Mae Sa Mai, the six
different local soil types were classified using topsoil thickness as a criterion also,
and in Huai Bong, the five local soil types were identified based on topsoil thickness
and stone content. Finally, in Bor Krai, eight different soil types could be distinguished, with bulk density used as an additional criterion. All the local soil maps
produced provided an overview of the respective main soil types, plus background
information on each local soil type such as crop suitability, water infiltration and
erosion risk, as provided by the farmers. The mapping of local soils and their
properties was the cheapest and most rapid method to use, but the quality and
quantity of local knowledge varied among the villages.
56
K. Stahr et al.