Pengembangan model tanah-lanskap untuk memprediksi penyebaran sifat tanah di wilayah tropika (studi kasus di Pulau Jawa)
Developing soil-landscape model to predict soil property distribution in tropical region: case study in Java Island
View/ Open
Date
2012Author
Sulaeman, Yiyi
Sutandi, Atang
A. Rachim, Djunaedi
Barus, Baba
Metadata
Show full item recordAbstract
Soil property and its distribution are required as a basic input in planning, managing, and monitoring land resources as well as in assessing soil for improving crop productivity and in evaluating the impact of land use to environment. Yet, data collection consumes much money and time so that a new mapping approach is required. Some current advances in database technology, spatial mathematics, geospatial analysis, and computer technology can explore spatial pattern in legacy soil data more efficiently. A digital soil mapping makes use the available techniques and revitalizes data, but it has not been tested yet in the tropical region. This research aimed: (i) to develop soil-landscape models using legacy soil data from Java, (ii) to evaluate model transferability in other region, and (iii) to map soil property in new area using modified digital soil mapping framework based on legacy data. The research covers (i) legacy data collection and harmonization from Java Island, (ii) model development using stepwise regression and tree regression, (iii) model transferability evaluation in the Upper Cisadane Watershed in Bogor, West Java Province and in the Upper Sampean Watershed in Bondowoso, East Java Province, and (iv) soil map creation using a digital soil mapping technique in Upper Cisadane watershed. The study resulted in 96 soil-landscape models that have good predictive power to predict selected soil properties. The best model were models to predict thick of A horizon, depth to B horizon, clay percentage, pH, and cation exchange capacity. Model transferability in both watershed varied depending upon the combination of covariates and watershed characteristics. Models to predict soil depth was demonstrated to map soil depth distribution in the Upper Cisadane Watershed. The method, models, and dataset from this research could be used to assist in accelerating soil survey and mapping.
Collections
- DT - Agriculture [727]