Land cover mapping

From Wikipedia, the free encyclopedia

Land cover maps are tools that provide vital information on the Earth's land use and cover patterns. They aide policy development, urban planning, forest and agricultural monitoring etc.[1][2]

The systematic mapping of land cover patterns, including change detection most often follow two main approaches:

  • field survey
  • remote sensing satellite image processing.[3] This approach which is cost efficient, employ several image pre-processing and processing techniques to accurately map land cover patterns. These technics detect changes at various spatial scales following a series of machine learning simulations and statistical applications.

Image preprocessing is normally done through radiometric corrections, while image processing involves the application of either unsupervised classifications, or supervised classifications and vegetation indices quantification for land cover map production.

Supervised classification[]

A Supervised classification is a system of classification in which the user build a series of randomly generated training datasets or spectral signatures representing different LULC classes, and apply these datasets in machine learning models to predict and spatially classify LULC patterns, and evaluate classification accuracies.

Several machine learning algorithms have been developed for this approach. Examples include: Maximum Likelihood Classification (MLC),[4] Support Vector Machines (SVMs),[5] Random Forest (RF),[6] Decision Trees (DT),[7] K-Nearest Neighbors (KNN),[5] Multi-perceptron Artificial Neural Networks (MP-ANNs),[7][8] Minimum Distance (MD),[4] Mahalanobis Distance,[9] Spectral Angler Mapper (SAM),[10] Discriminant Analysis (DA), Fuzzy (FZ), Genetic Algorithm,[11] Subspace,[12][13] and Parallelepiped classification.[14]

Description of supervised classification algorithms[]

  • Maximum Likelihood Classification (MLC): This approach classifies overlapping signatures by estimating the probability that an image pixel with the maximum likelihood corresponds to a particular LULC type. It is also dependent on the mean and covariance matrices of training datasets and assumes statistical significance of image pixels.[4]
  • Minimum Distance (MD): A form of supervised classification that defines decision boundaries between image pixels to classify land cover.[4] The decision boundaries are formed by calculating the mean distance between class pixels, and the standard deviation of the generated training datasets to generate a parallelepiped box.
  • Mahalanobis Distance: A system of classification that uses the Euclidean distance algorithm to assign land cover classes from a set of training datasets.[9]
  • Spectral Angler Mapper (SAM): A spectral image classification approach that uses angular measurements to determine the relationship between two spectra, treating them as vectors in a q-dimensional space, with the q-dimensions representing the number of bands.[10]
  • Discriminant Analysis (DA): A system of classification in which the classifying algorithm separates groups of closely related image pixels into classes, minimizing the variance within classes, and maximizing the variance between classes following a maximum likelihood discriminant rule
  • Genetic Algorithm: A system of classification that applies genetic principles for selecting appropriate clusters of training data and classifying them under the influence of predictors (satellite image bands).[11]
  • Subspace: A classification approach in which the classifier creates low dimensional subspaces of each land cover class selected from a cluster of training points. The approach of dimensional subspace creation involves performing a principal component analysis on the training points.[12][13] Two types of Subspace algorithms exist for minimizing land cover classification errors. They include: the class-featuring information compression (CLAFIC)[15] and the average learning Subspace method (ALSM).[16]
  • Parallelepiped classification: A feature space classifier that assigns range of values for each land cover class within each image band and creates bounding boxes where pixels from each land cover class are selected for training the classifier.[14]
  • Multi-perceptron Artificial Neural Networks (MP-ANNs): A system of classification in which the classifier uses a series of neural networks or nodes to classify land cover based on backpropagations of training samples.
  • support vector machines (SVM): A classification approach in which the classifier uses support vectors to obtain optimal decision boundaries separating two or more land cover classes.
  • Random Forest (RF): An approach in which the classifier uses booistraps to create several decision trees that classify training datasets based on a number of satellite image bands.[6]
  • k-Nearest Neighbors: This approach draws k-closest samples from training datasets and classify land cover based on the distance between these samples.
  • Decision trees (DT): Like RF, DT constitutes a set of connected nodes that partition training samples into a set of land cover clusters.[7] Its advantages are that it is fast, easy to construct and interpret for smaller data, and good at excluding background or unimportant information. It is disadvantageous in that it can create overfitting especially for large datasets.

Unsupervised classification[]

Unsupervised classification is a system of classification in which the image pixels or group of pixels are automatically classified by the software without the user applying signature files or training data. However, the user defines the number of classes for which the computer will automatically generate by grouping similar pixels into a single category using a clustering algorithm. This system of classification is mostly used in areas with no field observations or prior knowledge on the available land cover types. Examples of unsupervised algorithms include ISODATA (Iterative Self-Organizing Data Analysis Technique) and K-Means.[17]

  • ISODATA is a system of classification in which the classifier automatically group a number of closely related image pixels into clusters, and then computes the mean clusters and classify land cover based on a series of repeated iterations.
  • With K-Means, the computer automatically extract k land cover features from satellite images, and classify the overall image based on the calculated means of the extracted features.

Classification with vegetation indices[]

Vegetation indices classification is a system of classification in which two or more spectral bands are combined through defined statistical algorithms to reflect the spatial properties of a vegetation cover.

Most of these indices make use of the relationship between Red and Near Infrared (NIR) bands of satellite images to generate vegetation properties. Several vegetation indices have been developed and applied by remote sensing scientist to effectively classify forest cover and land use patterns. Prominent examples include the

  • Normalized Differential Vegetation Index (NDVI),[18][19]
  • Enhanced Vegetation Index (EVI),[20]
  • Soil Adjusted Vegetation Index (SAVI),[21]
  • Advanced Vegetation Index (AVI)
  • Canopy Shadow Index (SI),
  • Bare Soil Index (BSI),[22][23]
  • Normalized Differential Water Index (NDWI)[24]
  • Normalized Differential Built-up Index (NDBI).[25]

These spectral indices use two or more bands to accurately acquire surface reflectance of land use and cover features, thereby improving land use/cover classification accuracies.[26][27]

Description of vegetation indices[]

  • Normalized Differential Vegetation Index (NDVI): The Normalized Differential Vegetation Index is defined by the ratio between the red and near infrared bands of satellite images. i.e. NDVI = (NIR – R) / (NIR + R), where NIR = Near Infrared and R = Red. This index measures vegetation greenness, with values ranging between -1 and 1. High NDVI values represent dense vegetation cover, moderate NDVI values represent sparse vegetation cover, and low NDVI values correspond to non vegetated areas (e.g. barren or bare lands).[28]
  • Enhanced Vegetation Index (EVI): EVI is defined by the ratio between the Red, NIR and Blue bands, with a soil brightness correction factor (L) and an atmospheric aerosol correction factor (C).[29][30] It is calculated as: EVI = , where B = Blue, L = soil brightness correction factor usually given a default value of 0.5, and C1 and C2 = atmospheric aerosol correction factors.
  • Soil Adjusted Vegetation Index (SAVI): SAVI is defined by the ratio between the Red and NIR values with a soil brightness correction factor (L). It is calculated as SAVI = .
  • Canopy Shadow Index (SI): SI is defined by the square root of the Red and Green bands of satellite images. It evaluates the different shadow patterns of forest canopies based on age, structure and composition, as well as easily differentiates dense forests from grass and bare lands.[23][31] It is calculated as SI = , where G = Green band
  • Advanced Vegetation Index (AVI): AVI is defined by the cubic root of the NIR and Red bands, with its generated indices differentiating forest cover from grassland and bare land areas. It is calculated as AVI =
  • Bare Soil Index (BSI): BSI is defined by the ratio between the NIR, Red and Blue bands of satellite images. It measures the amount of bare soil and as such increases with decrease forest density.[23][31] It is calculated as BSI =
  • Normalized Differential Water Index (NDWI): Developed for quantifying the water content of plants and other earth system features. It is calculated as NDWI = (NIR – SWIR)/(NIR +SWIR), where SWIR = short wave infrared
  • Normalized Differential Built-up Index (NDBI): Developed for quantifying built-up areas in satellite images. It is calculated as NDBI = (SWIR – NIR) / (SWIR + NIR)

See also[]

References[]

  1. ^ Wessels, Konrad J; Reyers, Belinda; van Jaarsveld, Albert S; Rutherford, Mike C (April 2003). "Identification of potential conflict areas between land transformation and biodiversity conservation in north-eastern South Africa". Agriculture, Ecosystems & Environment. 95 (1): 157–178. doi:10.1016/s0167-8809(02)00102-0. ISSN 0167-8809.
  2. ^ Gebhardt, Steffen; Wehrmann, Thilo; Ruiz, Miguel; Maeda, Pedro; Bishop, Jesse; Schramm, Matthias; Kopeinig, Rene; Cartus, Oliver; Kellndorfer, Josef; Ressl, Rainer; Santos, Lucio (2014-04-30). "MAD-MEX: Automatic Wall-to-Wall Land Cover Monitoring for the Mexican REDD-MRV Program Using All Landsat Data". Remote Sensing. 6 (5): 3923–3943. Bibcode:2014RemS....6.3923G. doi:10.3390/rs6053923. ISSN 2072-4292.
  3. ^ Cracknell, Matthew J.; Reading, Anya M. (February 2014). "Geological mapping using remote sensing data: A comparison of five machine learning algorithms, their response to variations in the spatial distribution of training data and the use of explicit spatial information". Computers & Geosciences. 63: 22–33. Bibcode:2014CG.....63...22C. doi:10.1016/j.cageo.2013.10.008. ISSN 0098-3004.
  4. ^ a b c d Press, Forex. "Analysis of Supervised Image Classification Method for Satellite Images". Cite journal requires |journal= (help)
  5. ^ a b Lo, C. P.; Choi, Jinmu (July 2004). "A hybrid approach to urban land use/cover mapping using Landsat 7 Enhanced Thematic Mapper Plus (ETM+) images". International Journal of Remote Sensing. 25 (14): 2687–2700. Bibcode:2004IJRS...25.2687L. doi:10.1080/01431160310001618428. ISSN 0143-1161. S2CID 129129271.
  6. ^ a b Mellor, Andrew; Haywood, Andrew; Stone, Christine; Jones, Simon (2013-06-04). "The Performance of Random Forests in an Operational Setting for Large Area Sclerophyll Forest Classification". Remote Sensing. 5 (6): 2838–2856. Bibcode:2013RemS....5.2838M. doi:10.3390/rs5062838. ISSN 2072-4292.
  7. ^ a b c Beucher, A.; Møller, A.B.; Greve, M.H. (October 2019). "Artificial neural networks and decision tree classification for predicting soil drainage classes in Denmark". Geoderma. 352: 351–359. Bibcode:2019Geode.352..351B. doi:10.1016/j.geoderma.2017.11.004. ISSN 0016-7061. S2CID 134063283.
  8. ^ Silva, Leonardo Pereira e; Xavier, Ana Paula Campos; da Silva, Richarde Marques; Santos, Celso Augusto Guimarães (March 2020). "Modeling land cover change based on an artificial neural network for a semiarid river basin in northeastern Brazil". Global Ecology and Conservation. 21: e00811. doi:10.1016/j.gecco.2019.e00811. ISSN 2351-9894.
  9. ^ a b Khan, Umair; Minallah, Nasru; Junaid, Ahmad; Gul, Kashaf; Ahmad, Nasir (December 2015). "Parallelepiped and Mahalanobis Distance based Classification for forestry identification in Pakistan". 2015 International Conference on Emerging Technologies (ICET). IEEE: 1–6. doi:10.1109/icet.2015.7389199. ISBN 978-1-5090-2013-3. S2CID 38668604.
  10. ^ a b Kruse, F. A.; Lefkoff, A. B.; Boardman, J. W.; Heidebrecht, K. B.; Shapiro, A. T.; Barloon, P. J.; Goetz, A. F. H. (1993). "The spectral image processing system (SIPS)-interactive visualization and analysis of imaging spectrometer data". AIP Conference Proceedings. AIP. 283: 192–201. Bibcode:1993AIPC..283..192K. doi:10.1063/1.44433.
  11. ^ a b Maulik, Ujjwal; Bandyopadhyay, Sanghamitra (September 2000). "Genetic algorithm-based clustering technique". Pattern Recognition. 33 (9): 1455–1465. Bibcode:2000PatRe..33.1455M. doi:10.1016/s0031-3203(99)00137-5. ISSN 0031-3203.
  12. ^ a b Sun, Weiwei; Ma, Jun; Yang, Gang; Du, Bo; Zhang, Liangpei (June 2017). "A Poisson nonnegative matrix factorization method with parameter subspace clustering constraint for endmember extraction in hyperspectral imagery". ISPRS Journal of Photogrammetry and Remote Sensing. 128: 27–39. Bibcode:2017JPRS..128...27S. doi:10.1016/j.isprsjprs.2017.03.004. ISSN 0924-2716.
  13. ^ a b Sun, Weiwei; Du, Bo; Xiong, Shaolong (2017-05-01). "Quantifying Sub-Pixel Surface Water Coverage in Urban Environments Using Low-Albedo Fraction from Landsat Imagery". Remote Sensing. 9 (5): 428. Bibcode:2017RemS....9..428S. doi:10.3390/rs9050428. ISSN 2072-4292.
  14. ^ a b Mei Xiang; Chih-Cheng Hung; Minh Pham; Bor-Chen Kuo; Coleman, T. (2005). "A parallelepiped multispectral image classifier using genetic algorithms". Proceedings. 2005 IEEE International Geoscience and Remote Sensing Symposium, 2005. IGARSS '05. Seoul, Korea: IEEE. 1: 482–485. doi:10.1109/IGARSS.2005.1526216. ISBN 978-0-7803-9050-8. S2CID 37014767.
  15. ^ Gülmezoğlu, M. Bilginer; Dzhafarov, Vakıf; Edizkan, Rifat; Barkana, Atalay (April 2007). "The common vector approach and its comparison with other subspace methods in case of sufficient data". Computer Speech & Language. 21 (2): 266–281. doi:10.1016/j.csl.2006.06.002. ISSN 0885-2308.
  16. ^ Laaksonen, Jorma; Oja, Erkki (1996), Malsburg, Christoph; Seelen, Werner; Vorbrüggen, Jan C.; Sendhoff, Bernhard (eds.), "Subspace dimension selection and averaged learning subspace method in handwritten digit classification", Artificial Neural Networks — ICANN 96, Berlin, Heidelberg: Springer Berlin Heidelberg, 1112, pp. 227–232, doi:10.1007/3-540-61510-5_41, ISBN 978-3-540-61510-1, retrieved 2021-04-13
  17. ^ Abbas, A.; Minalla, N.; Ahmad, N.; Abid, S.; Khan, M. K-means and ISODATA clustering algorithms for landcover classification using remote sensing. Sindh Univ. Res. J. SURJ (Sci. Ser.) 2016, 48, 315–318
  18. ^ Pettorelli, Nathalie; Vik, Jon Olav; Mysterud, Atle; Gaillard, Jean-Michel; Tucker, Compton J.; Stenseth, Nils Chr. (September 2005). "Using the satellite-derived NDVI to assess ecological responses to environmental change". Trends in Ecology & Evolution. 20 (9): 503–510. doi:10.1016/j.tree.2005.05.011. ISSN 0169-5347. PMID 16701427.
  19. ^ Pettorelli, Nathalie; Gaillard, Jean-Michel; Mysterud, Atle; Duncan, Patrick; Chr. Stenseth, Nils; Delorme, Daniel; Van Laere, Guy; Toïgo, Carole; Klein, Francois (March 2006). "Using a proxy of plant productivity (NDVI) to find key periods for animal performance: the case of roe deer". Oikos. 112 (3): 565–572. doi:10.1111/j.0030-1299.2006.14447.x. ISSN 0030-1299.
  20. ^ JIANG, Z; HUETE, A; DIDAN, K; MIURA, T (2008-10-15). "Development of a two-band enhanced vegetation index without a blue band". Remote Sensing of Environment. 112 (10): 3833–3845. Bibcode:2008RSEnv.112.3833J. doi:10.1016/j.rse.2008.06.006. ISSN 0034-4257.
  21. ^ Huete, A.R (August 1988). "A soil-adjusted vegetation index (SAVI)". Remote Sensing of Environment. 25 (3): 295–309. Bibcode:1988RSEnv..25..295H. doi:10.1016/0034-4257(88)90106-x. ISSN 0034-4257.
  22. ^ Rikimaru, A., 1999. The concept of FCD mapping model and semi-expert system. FCD mapper user’s guide. International Tropical Timber Organization and Japan Overseas Forestry Consultants Association. Pp 90.
  23. ^ a b c Rikimaru, R., Roy, P.S. and Miyatake, S., 2002, Tropical forest cover density mapping. Tropical Ecology, 43, pp. 39–47.
  24. ^ Gao, Bo-cai (December 1996). "NDWI—A normalized difference water index for remote sensing of vegetation liquid water from space". Remote Sensing of Environment. 58 (3): 257–266. Bibcode:1996RSEnv..58..257G. doi:10.1016/s0034-4257(96)00067-3. ISSN 0034-4257.
  25. ^ Zha, Y.; Gao, J.; Ni, S. (January 2003). "Use of normalized difference built-up index in automatically mapping urban areas from TM imagery". International Journal of Remote Sensing. 24 (3): 583–594. Bibcode:2003IJRS...24..583Z. doi:10.1080/01431160304987. ISSN 0143-1161. S2CID 129599221.
  26. ^ Tso, Brandt; Mather, Paul M (2001). Classification Methods for Remotely Sensed Data. Abingdon, UK: Taylor & Francis. doi:10.4324/9780203303566. ISBN 978-0-203-35581-7.
  27. ^ Shaban, M. A.; Dikshit, O. (January 2001). "Improvement of classification in urban areas by the use of textural features: The case study of Lucknow city, Uttar Pradesh". International Journal of Remote Sensing. 22 (4): 565–593. Bibcode:2001IJRS...22..565D. doi:10.1080/01431160050505865. ISSN 0143-1161. S2CID 128572668.
  28. ^ Wegmann M, Leutner B, Dech S (2016) Remote sensing and GIS for ecologists: using open source software. Pelagic Publishing, Exeter, UK
  29. ^ Hui Qing Liu; Huete, A. (March 1995). "A feedback based modification of the NDVI to minimize canopy background and atmospheric noise". IEEE Transactions on Geoscience and Remote Sensing. 33 (2): 457–465. doi:10.1109/36.377946. ISSN 0196-2892. S2CID 28380065.
  30. ^ Xue, Jinru; Su, Baofeng (2017-05-23). "Significant Remote Sensing Vegetation Indices: A Review of Developments and Applications". Journal of Sensors. 2017: 1–17. doi:10.1155/2017/1353691.
  31. ^ a b Baynes, Jack (January 2004). "Assessing forest canopy density in a highly variable landscape using Landsat data and FCD Mapper software". Australian Forestry. 67 (4): 247–253. doi:10.1080/00049158.2004.10674942. ISSN 0004-9158. S2CID 84900545.
Retrieved from ""