Accuracy assessment, key concerns and improve procedures involved in NLCMS
Download as PPTX, PDF0 likes13 views
The document assesses the accuracy of Nepal's land cover map, highlighting critical validation processes and results from accuracy assessments conducted in 2011. It reports an overall classification accuracy of 79.20% with varying performance across different land cover classes, indicating concerns about mixed pixels and the need for improved training and validation methods. Key challenges include the interpretation of data and the impact of environmental factors like clouds and snow on the accuracy of the classifications.
Accuracy assessment, key concerns and improve procedures involved in NLCMS
- 1. International Centre for Integrated Mountain Development Kathmandu, Nepal Accuracy assessment, key concerns and improve procedures involved in NLCMS Kabir Uddin Kabir.Uddin@icimod.org
- 2. Validation of Nepal land cover map What is the accuracy of the results Have you done validation Most common questions from the remote sensing expert about results Background The accuracy of image classification is highly important to the monitoring and investigation of national level land cover mapping. During the image classification many mixed pixels are located in between different classes, the accuracy of edge pixels tend to be lower than that of interior pixels. This scenario leads to the increment of heterogeneity to the classification accuracy of each class and to the increase of uncertainty of accuracy assessment. Nepal land cover validation process helped us to assess how well the classification carried for Nepal land cover mapping. Understand how to interpretation using was the usefulness
- 3. Validation of developed land cover: sample collection 4 km interval 55358 assessment plot generated for Nepal Forest Grassland Other wooded area Cropland Waterbodies Riverbed Glacier Snow Built-up area Baresoil
- 4. Validation of developed land cover 45173 number of quality point collected or used against of land cover classes 796 addition point collected for riverbed, built-up, waterbodies, grassland, and baresoil Availability of 2011 high resolution images Shadow and cloud on the 2011 high resolution images Variation of snowing area
- 5. Land cover map of Nepal Year Forest area (%) OWL (%) Total (%) 2018 40.40 04.11 44.52 2015 40.32 03.61 43.94 2011 39.83 03.14 42.97 2010 39.59 03.06 42.66 2000 39.79 03.18 42.97 Forest area (%) OWL (%) Total (%) 0 5 10 15 20 25 30 35 40 45 50 2000 2010 2011 2015 2018
- 6. Accuracy assessment of Nepal land cover ERROR MATRIX for all classes of 2011 land cover Overall Classification Accuracy (2011) = 79.20% Overall Kappa Statistics = 0.7018 Class name Waterbodies Glacier Snow Forest Riverbed Built-up area Cropland Baresoil Barerock Grassland OWA Total Waterbodies 181 0 0 2 11 0 5 0 0 0 2 201 Glacier 0 1516 8 0 0 0 0 0 0 0 0 1524 Snow 0 3 2686 0 0 0 8 0 263 0 1 2961 Forest 1 0 0 18996 0 6 558 0 0 0 266 19827 Riverbed 46 0 2 12 337 4 209 12 1 1 4 628 Built-up area 0 0 0 0 0 263 21 0 0 0 1 285 Cropland 3 0 0 839 3 498 10754 0 0 6 406 12509 Baresoil 0 0 0 0 0 0 0 0 0 0 0 0 Barerock 0 0 855 6 0 0 0 0 935 0 4 1800 Grassland 5 0 240 858 3 25 587 0 7 147 951 2823 OWA 0 0 0 2360 0 1 447 0 0 2 591 3401 Total 236 1519 3791 23073 354 797 12589 12 1206 156 2226 45969 Class name Producers Accuracy Users Accuracy Waterbodies 76.69% 90.05% Glacier 99.74% 99.48% Snow 70.85% 90.71% Forest 82.33% 95.81% Riverbed 92.84% 53.66% Built-up area 33.00% 92.28% Cropland 85.42% 85.97% Baresoil --- --- Barerock 77.53% 51.94% Grassland 94.23% 5.21% OWA 26.55% 17.38%
- 7. IPCC classes of 2011 land cover Land cover map of Nepal
- 8. Accuracy assessment of Nepal land cover ERROR MATRIX for IPCC classes of 2011 land cover Overall Classification Accuracy (2011) = 87.70% Overall Kappa Statistics = 0.8033 Class_Name Wetlands Other land Cropland Forest land Grassland Settlements Total Wetlands 575 14 216 22 1 4 832 Other land 0 6275 8 12 0 0 6295 Cropland 6 0 10716 1223 6 499 12450 Forest land 1 0 1070 22341 2 10 23424 Grassland 8 240 560 1700 147 24 2679 Settlements 0 0 19 1 0 260 280 Total 590 6529 12589 25299 156 797 45969 Class name Reference Totals Classified Totals Number Correct Producers Accuracy Users Accuracy Wetlands 599 832 575 95.99% 69.11% Other land 6529 6295 6275 96.11% 99.68% Cropland 12589 12450 10716 85.12% 86.07% Forest land 25299 23424 22341 88.31% 95.38% Grassland 156 2679 147 94.23% 5.49% Settlements 797 280 260 32.62% 92.86% Total 45969 45969 40314 Wetlands 0.6870 Other land 0.9963 Cropland 0.8082 Forest land 0.8972 Grassland 0.0517 Settlements 0.9273 KAPPA (K^) STATISTICS Overall Kappa Statistics = 0.8033 Conditional Kappa for each Category.
- 9. Key concerns and improvements
- 10. Key concerns and improvements
- 11. Key concerns and improvements
- 12. Key concerns and improvements Training and validation sample collection
- 13. Key concerns and improvements Understanding of ground reality
- 14. Key concerns and improvements Understanding of ground reality
- 15. Key concerns and improvements Understanding of ground reality
- 16. Key concerns and improvements Understanding of ground reality
- 17. Key concerns and improvements Understanding of ground reality Utilization of annual normalized built- up index