Reconstruction of cloud-contaminated satellite remote sensing images using kernel PCA-based image modelling

Greeshma, N K; Baburaj, M; George, Sudhish N.

doi:10.1007/s12517-015-2199-3

Cited by 10 publications

(1 citation statement)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A number of methods have been developed to reconstruct image pixels under cloudy areas, categorized into three groups: spatial, spectral, and temporal. The spatial-based methods using interpolation techniques are only effective when applied to restore small cloudy areas (Greeshma, Baburaj, and Sudhish 2016;Chang, Bai, and Chen 2015). The spectral-based methods developed based on the relationship between spectral band data from one or several sensors have several advantages owing to using other available spectral information around the target-contaminated pixels Shen et al 2014).…”

Section: Introductionmentioning

confidence: 99%

Satellite characterization of drought over cultivated areas in Central America

Son

Chen

et al. 2018

International Journal of Remote Sensing

View full text Add to dashboard Cite

Effects of drought on crop production have been a prevailing challenge confronting farmers in Central America. Information on drought associated with cropping areas is deemed vital for officials to evaluate issues of crop production losses and food insecurity in the region. This study aimed to develop an approach for agricultural drought assessment in Central America, using Moderate Resolution Imaging Spectroradiometer (MODIS) landsurface temperature (LST) and enhanced vegetation index (EVI) data during December 2000 to December 2016. The data processing comprised three steps: (1) LST and EVI reconstruction using random forest, (2) drought delineation using the temperature vegetation dryness index (derived from LST and EVI), and (3) error verification. The reconstructed results of LST and EVI data verified with the cloud-free reference data indicated coefficient of determination (R 2 ) values higher than 0.84, while the root mean square error values achieved for LST and EVI were smaller than 1.5°C and 0.07, respectively. The spatially aggregated results of drought delineation compared with the soil moisture active passive data showed that the correlation coefficient (r) values obtained for the apante season (December−March) were between −0.4 and −0.8, while the values for the primera (April−August) and postrera (September−November) seasons were in ranges of 0 to −0.7 and −0.1 to −0.6, respectively. The larger area of very dry moisture condition was especially observed for the 2002 and 2014 primera seasons. Eventually, the probability of drought occurrence calculated every 8 days over a 16 years period was spatially aggregated with the cropping areas at the municipality level for crop management purposes in the region. ARTICLE HISTORY

show abstract

Section: Introductionmentioning

confidence: 99%