Methods of performance evaluation for the supervised classification of satellite imagery in determining land cover classes

Souza, Carlos Eduardo Alves de; Mercante, Erivelto; Prudente, Victor Hugo Rohden; Justina, Diego Domingos Della

doi:10.4067/s0718-16202013000200016

Cited by 9 publications

(5 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Afterward, we collected samples by visual interpretation of these images. We applied the supervised classifier Spectral Angle Mapper (SAM) over the stack of these selected images (Souza et al, 2013). We extracted the mask of water/no water, along with the elevation and azimuth solar angles information from the metadata, and we used it as input to the ATSA algorithm.…”

Section: Automatic Time-series Analysis (Atsa)mentioning

confidence: 99%

Accuracy Assessment of Cloud Mask Detection Algorithms for Cbers-4 Wfi Imagery

Oldoni

Sanches²,

Picoli³

2022

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

Abstract. Clouds limit the potential use of optical images. Proper clouds and cloud shadows detection are necessary steps for optical image applications. Few algorithms are flexible in detecting clouds in images with a limited number of bands, such as the Wide-Field Imager (WFI) sensor on board the China-Brazil Earth Resources Satellite (CBERS-4), which has four spectral bands (blue, green, red, and near-infrared). Therefore, this work aims to assess the accuracy of two cloud detection algorithms: CMASK and ATSA, and evaluate the influence of the ATSA configuration parameters. We selected four regions in Brazil for our analysis. In all cases, ATSA presented overall accuracy (OA) superior to CMASK. While the ATSA obtained OA greater than 0.91 for all analyzes, the OA from CMASK did not exceed 0.84. CMASK presented commission errors for the No Clear class (combination of Cloud and Cloud shadow) and inclusion errors for the Clear class close to zero. However, many errors of omission of clouds misclassified as the Clear class was observed. The ATSA algorithm presented a better balance between inclusion errors and omission errors. Our results can be used as guidance for choosing a cloud mask algorithm for the CBERS-4 WFI images and for analysis considering the images from WFI on board CBERS-4A and Amazonia-1, as they have similar characteristics.

show abstract

Section: Automatic Time-series Analysis (Atsa)mentioning

confidence: 99%

Accuracy Assessment of Cloud Mask Detection Algorithms for Cbers-4 Wfi Imagery

Oldoni

Sanches²,

Picoli³

2022

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

show abstract

“…The more visited each land cover, the better. Then the data were analyzed with the Kappa coefficient to determine the percentage of accuracy and improvement step (de Souza et al 2013). The use of spectral colors and field checks confirmed more accurately than land cover in Landsat imagery.…”

Section: Land Cover Reviewmentioning

confidence: 99%

Distribution and habitat preferences of Sumatran elephant (Elephas maximus sumatranus) in Riau, Indonesia

et al. 2018

View full text Add to dashboard Cite

Sukmantoro YW, Alikodra HS, Kartono AP, Efransjah. 2019. Distribution and habitat preferences of Sumatran elephant (Elephas maximus sumatranus) in Riau, Indonesia. Biodiversitas 20: 226-235. The research on elephant distribution and habitat preference was carried out in two locations Balai Raja and Tesso Nilo, Riau Province, Indonesia on 2012-2016. Three GPS Collars were installed on one individual adult female at Balai Raja (Desma) and two individuals at Tesso Nilo (Angelina and Butet). For the elephants’ habitat, the forest cover identification was performed by spectral color analysis on the 8 ETM+ Landsat images and resulted in land cover detail on MCP (Minimum Convex Polygon), and 50% and 95% fix kernel density of movement of the elephant including in the morning, the day and the night. Based on MCP estimation, Desma habitat area was 328.86 km2, while Angelina and Butet were 632.57 km2 and 701.60 km2, respectively. Land cover in MCP indicated that the oil palm, mixed agriculture, and rubber were dominantly areas used by Desma, while acacia, forest and oil palm plantation were dominantly area used by Angelina and Butet. In habitat preferences, Desma preferred forests and swamp areas to day activity. However, Angelina and Butet selected acacia plantation, forest, infrastructures such as the mills and land opening as the location of movement. Water bodies and shrub were not widely used for the elephants’ activities. This study indicated that the pattern of habitat selection is a strategy for elephants to avoid conflict with the human. Although resources are limited, the habitat selection can reduce conflict incidence.

show abstract

“…Then, available parametric algorithms were carried out in ERDAS IMAGINE software to classify the images into three categories: forests, residential areas and other features. For this purpose, maximum likelihood (Patil1 et al, 2012), fully-fuzzy supervised (Zhang & Foody, 2001), minimum distance (Yang et al, 2015) and Mahalonobis classifications (Souza et al, 2013) were run in ERDAS IMAGINE (www.erdas.com). The validation of classified images was based on the collected GCPs data through the sampling of different references: aerial photos from 1972 and 1987, digital topographic maps for 2000 and field observation samples and Google Earth images of 2010.…”

Section: Forests and Residential Areas Detectionmentioning

confidence: 99%

Analysing Spatial and Statistical Dependencies of Deforestation Affected by Residential Growth: Gorganrood Basin, Northeast Iran

et al. 2017

View full text Add to dashboard Cite

This study aimed to examine deforestation and residential growth trends and their spatial dependencies from 1972 to 2010 in Northeast of Iran. First, change rates of forests and residential areas were mapped using Landsat satellite images in 1972–1987, 1987–2000 and 2000–2010. Then, the forest change patterns were interpreted using univariate local Moran's I (local univariate spatial autocorrelation), and the spatial autocorrelation between deforestation and residential growth was tested through bivariate local Moran's I (bivariate local spatial autocorrelation). Furthermore, the spatial relationships between deforestation and residential growth rates were quantified by ordinary least squares, spatial lag (SL) and geographically weighted regression. Results indicated that approximately 25% of forests have been converted to other land‐use types in the span of 38 years, since 1972. Local univariate spatial autocorrelation maps showed that significant values of high–high cluster scattered in all locations in the first span, in the east and south aspects in the second duration, and in the eastern part in the third span. Bivariate local spatial autocorrelation indicated a meaningful Moran's I values of −0·12, −0·26 and −0·20 between deforestation and residential growth, chronologically. Analyses of spatial regression models showed that geographically weighted regression performed better than SL and ordinary least squares in the first (R2 = 0·315, AIC = 6,160) and third periods (R2 = 0·27, AIC = 6,351), whereas, the validity of SL was the highest in the second period (R2 = 0·36, AIC = 6,288). However, the overall trends of deforestation and residential growth have decreased, but the rate of deforestation induced by residential growth is still significant. Spatial exploration of residential growth in deforestation leads to determine its influences in local scale for better conservation of these valuable natural resources. Copyright © 2017 John Wiley & Sons, Ltd.

show abstract

Methods of performance evaluation for the supervised classification of satellite imagery in determining land cover classes

Cited by 9 publications

References 16 publications

Accuracy Assessment of Cloud Mask Detection Algorithms for Cbers-4 Wfi Imagery

Accuracy Assessment of Cloud Mask Detection Algorithms for Cbers-4 Wfi Imagery

Distribution and habitat preferences of Sumatran elephant (Elephas maximus sumatranus) in Riau, Indonesia

Analysing Spatial and Statistical Dependencies of Deforestation Affected by Residential Growth: Gorganrood Basin, Northeast Iran

Contact Info

Product

Resources

About