Soil Survey Design For Management-Scale Digital Soil Mapping In A Mountainous Southern Philippine Catchment

We are devising sustainable land use strategies for sloping uplands (>= 18%) for the southern Philippines, with the current focus on the Cabulig catchment (22,900 ha). Digital soil mapping (DSM) will be used to create soil property maps at a suitable scale for on-ground planning and management (i. e., 1: 100,000 scale). This paper describes the development and testing of a DSM sampling design to satisfy the dual needs of statistically-based sampling rigour for reliable soil estimations and field practicality in remote and rugged landscapes. The design combines the elements of: (i) conditioned Latin Hypercube sampling (cLHS), which performs maximally stratified random sampling in the presence of environmental covariates; (ii) sampling along toposequence-based transects for field efficiency; and (iii) nested interval sampling along the transects to 1,500 m in length to span the catchment ranges of soil variability. The geographic coordinates produced by cLHS (seed points) defined the location of transects from a random transect generating algorithm. Environmental covariates included satellite remote sensing and digital elevation models. Here we describe the outcome of a 2 1/2 week trial conducted late last year to test the design in preparation of an extensive operational survey. The key findings are that 20 cLHS-derived sites gave a suitable geographic and environmental variable distribution of the catchment, offering the possibility of strong soil model development. Consideration should be given to reviewing the covariates used in the cLHS component so as to be more representative of the catchment's pedogenic characteristics. If so, the legacy of existing samples should be preserved in the modification. The terminal sampling interval (1,500 m) proved overly costly in terms of field effort, hence could be shortened to 500 m. This offers the possibility of more transects to be sampled in the time allowed. Finally, in the field, the lack of understanding of cLHS site selection "logic" proved both intellectually and practically unsatisfactory. Ideas to address this issue are given.