Soil attributes mapping: an experience from the dry areas by Feras M. Ziadat
0 likes466 views
The document discusses the challenges and advancements in soil attribute mapping, particularly in dry areas. It highlights the limitations of traditional soil maps and demonstrates how modern technologies like GIS and remote sensing can improve predictions of soil characteristics. The results indicate that predictive models based on terrain attributes yield more accurate and detailed spatial distributions of soil properties compared to conventional mapping methods.
Soil attributes mapping: an experience from the dry areas by Feras M. Ziadat
- 1. 1 Soil attributes mapping: an experience from the dry areas Inception Workshop on Regional Soil Partnership and MENA Soil Information System 1-5 April 2012 Amman – Jordan Feras M. Ziadat
- 2. 2 Soil Survey Data Limitation of soil maps: The coverage, especially with enough details, in dry areas is usually limited. The cost of extending this coverage is high. Example: Jordan … 1:250,000: The whole country 1:50,000: 20 % of the country 1:10,000: 2 % of the country
- 3. 3 Soil Survey Data Limitation of soil maps: The purity of mapping units is usually low, leads to erroneous conclusions when used for site-specific decisions The accuracy of site-specific suitability using 1:10,000 scale soil map was only 60-70% ALTERNATIVE SOURCES OF SOIL INFORMATION Modern environmental applications and modelling require information about the spatial distribution of lateral and vertical soil attributes
- 4. 4 Alternative sources of soil information Developments in new technologies; GIS and remote sensing provide new approaches to meet the demand of resource-related modelling. DEMs are used to derive landscape attributes, which is utilized in landforms characterization. The derived topographic attributes control the soil differentiation through the form-process connections that is often addressed in geomorphology.
- 5. 5 Prediction of soil attributes using terrain attributes Terrain attributes derived from 20-m resolution DEM were utilized to predict soil attributes by implementing different statistical and clustering techniques. Correlation between soil attributes and terrain attributes for 2193 field observations
- 6. 6 Correlation and regression coefficients between soil attributes and terrain attributes (example for soil depth Significant but low correlation Low regression coefficients
- 7. 7 Regression coefficients between soil attributes and terrain attributes within selected sub-watersheds Rational: stream lines divided the watershed into two subdivisions Each sub-division is constitutes of endless number of catena Within each sub-division there is a unique relationship between soil and topography
- 8. 8 Better regression coefficients between soil attributes and terrain attributes within selected sub-watersheds
- 9. 9 However, watershed subdivision was only serial number, cannot be extrapolated Watershed subdivision were classified according to area and slope using unsupervised classification
- 10. 10 26 classes were optimum for this study area highest regression coefficient between soil attributes and terrain attributes within each of the 26 classes
- 11. 11 Linear models that utilize terrain attributes to predict soil attributes within each class were generated. Example: Soil depth
- 12. 12 Regression models were applied for each class and predicted soil attributed were amalgamated for the whole study area.
- 13. 13 Accuracy of prediction Soil Attribute Confidence of Prediction Prediction Accuracy Erosion type ± One class 48.4% Water holding capacity ± 50 mm/cm 75.8% Surface cover percent ± 10 % 78.7% Soil depth ± 50 cm 89.3% Soil texture ± One class 90.3% Surface cover type ± One class 94.5% Erosion class ± One class 98.0%
- 14. 14 The prediction model was applied for other study area to predict surface and sub-surface soil properties with good results even for un-sampled watershed subdivisions
- 15. 15 Root mean square error (RMSE) of the predicted soil characteristics was generally better than those derived from traditional soil map (scale 1:5,000)
- 16. 16 Spatial distribution of the predicted soil characteristic (carbonate content of horizon A) is better than those of the traditional soil map
- 17. 17 Application in Ethiopia Using SRTM 90m NDVI from ASTER images At two different dates 180 observations used for Prediction and 40 for Accuracy assessment
- 18. 18 Acceptable results given the availability of input data at Global level However, number of observations used is 180 for 60 sq. Km area
- 19. 19 Testing accuracy using few observations to optimize field work required
- 20. 20 Future Design user-friendly toolkit to predict soil attributes : Stand alone Sub-model within SWAT Encourage applications at various levels and purposes
- 21. 21 Example of application: suitability analysis The accuracy of the suitability classification derived from predicted soil attributes is comparable with the accuracy of classifications derived from traditional 1:10,000 soil maps (for range crops)
- 22. 22 The spatial distribution of the suitability classification derived from the predicted soil attributes indicated more realistic pattern
- 23. 23 Concluding remarks Modelling of the soil-landscape relationships within small watershed subdivisions is a promising approach to predict soil attributes for large areas. The spatial distribution of the predicted soil attributes is provided in more detailed form than what the soil map provides.
- 24. 24 Concluding remarks The digital output is an advantage that facilitates the incorporation into modern modelling techniques at various scales. The accuracy and spatial distribution of the suitability classification derived from predicted soil attributes compares favourably with those derived from traditional soil maps.
- 25. 25 Thank You …