Nominal 30-m Cropland Extent Map of Continental Africa by Integrating Pixel-Based and Object-Based Algorithms Using Sentinel-2 and Landsat-8 Data on Google Earth Engine

Xiong, Jian; Thenkabail, Prasad S.; Tilton, James C.; Gumma, Murali Krishna; Teluguntla, Pardhasaradhi; Oliphant, Adam; Congalton, Russell G.; Yadav, Kamini; Gorelick, Noel

doi:10.3390/rs9101065

Cited by 315 publications

(290 citation statements)

References 87 publications

(126 reference statements)

Supporting

Mentioning

256

Contrasting

Unclassified

Order By: Relevance

“…Each point represented a homogeneous cluster of at least 3 x 3 pixels. We calculated the overall accuracy (proportion of correctly classified reference points), producer's accuracy (PA), user's accuracy (UA) and F-score metrics for each classification (Congalton and Green 2009;Xiong et al 2017). The PA represents the proportion of reference data that is correctly classified (complement of the omission error).…”

Section: Methodsmentioning

confidence: 99%

“…forest cover, albedo, land surface temperature, etc), assessments of biodiversity decline for endemic species can benefit from the use of these fine grained multi-temporal data. It is key for conservation practitioners to understand the correct use (and limitations) of these products and/or platforms, because they form the basis for extracted/modeled ecosystem variables and in many cases are used for decision making and policy development (de Leeuw et al 2010, Congalton et al 2014, Perez-Hoyos et al 2017, Garcia-Alvarez et al 2019, Martinez-Fernandez et al 2019.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Habitat loss in the restricted range of the endemic Ghanaian cichlidLimbochromis robertsi

Lamboj

Lucanus

Darko

et al. 2019

Preprint

View full text Add to dashboard Cite

Remote sensing, through satellite image analysis has become an integral and invaluable tool to inform biodiversity conservation and monitoring of habitat degradation and restoration over time. Despite the disproportionately high levels of biodiversity loss in freshwater ecosystems worldwide, ichthyofauna are commonly overlooked in favor of other keystone species.Freshwater fish, as indicators of overall aquatic ecosystem health can also be indicators of larger scale problems within an ecosystem. If endemic and specialized fishes are at risk, the forest and landscape around their habitat is also undergoing change. As a case study demonstrating the utility of multi-temporal, multi-resolution satellite imagery, we examined deforestation and forest fragmentation around the Atewa Forest Reserve, south eastern Ghana. Within small creeks, Limbochromis robertsi, a unique freshwater cichlid with an extremely limited distribution range can be found. Historically, the land cover in the area has undergone substantial deforestation for agriculture and artisanal small-scale mining, primarily for gold. We found deforestation accelerated along with increased forest fragmentation in the 2014 -2017 period with the majority of the forest loss along the river and creek banks due to small-scale mining operations and increased agriculture. Field visits indicate a decrease in the total population by approximately 90% from the early 1990s to 2018. We illustrate the benefits of determining landscape metrics from local scale remote sensing studies as proxies to assess the decline of endemic species with restricted ranges, whose habitat characteristics and the subsequent pressures they face require detailed analysis at fine temporal and spatial scales not captured by global or continental scale datasets.

show abstract

Section: Methodsmentioning

confidence: 99%

mentioning

confidence: 99%

Habitat loss in the restricted range of the endemic Ghanaian cichlidLimbochromis robertsi

Lamboj

Lucanus

Darko

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…such as 30 m, are generally irregular because of low satellite revisit frequencies and cloud cover. One commonly used method used to reduce the “missing value” pixels is image composition (Xiong et al, 2017). However, the optimal composition period for cropland identification is not clear as short composition periods might describe dynamic land cover changes accurately, but missing pixels caused by cloud cover are also more likely to be included in the short composited image time series (Van Leeuwen, Huete & Laing, 1999), and density time series generated from short compositions also lead to information redundancy (Low et al, 2013; Wardlow & Egbert, 2010).…”

Section: Introductionmentioning

confidence: 99%

Estimation of different data compositions for early-season crop type classification

Hao

Niu

et al. 2018

PeerJ

View full text Add to dashboard Cite

Timely and accurate crop type distribution maps are an important inputs for crop yield estimation and production forecasting as multi-temporal images can observe phenological differences among crops. Therefore, time series remote sensing data are essential for crop type mapping, and image composition has commonly been used to improve the quality of the image time series. However, the optimal composition period is unclear as long composition periods (such as compositions lasting half a year) are less informative and short composition periods lead to information redundancy and missing pixels. In this study, we initially acquired daily 30 m Normalized Difference Vegetation Index (NDVI) time series by fusing MODIS, Landsat, Gaofen and Huanjing (HJ) NDVI, and then composited the NDVI time series using four strategies (daily, 8-day, 16-day, and 32-day). We used Random Forest to identify crop types and evaluated the classification performances of the NDVI time series generated from four composition strategies in two studies regions from Xinjiang, China. Results indicated that crop classification performance improved as crop separabilities and classification accuracies increased, and classification uncertainties dropped in the green-up stage of the crops. When using daily NDVI time series, overall accuracies saturated at 113-day and 116-day in Bole and Luntai, and the saturated overall accuracies (OAs) were 86.13% and 91.89%, respectively. Cotton could be identified 40∼60 days and 35∼45 days earlier than the harvest in Bole and Luntai when using daily, 8-day and 16-day composition NDVI time series since both producer’s accuracies (PAs) and user’s accuracies (UAs) were higher than 85%. Among the four compositions, the daily NDVI time series generated the highest classification accuracies. Although the 8-day, 16-day and 32-day compositions had similar saturated overall accuracies (around 85% in Bole and 83% in Luntai), the 8-day and 16-day compositions achieved these accuracies around 155-day in Bole and 133-day in Luntai, which were earlier than the 32-day composition (170-day in both Bole and Luntai). Therefore, when the daily NDVI time series cannot be acquired, the 16-day composition is recommended in this study.

show abstract

“…For many field-scale applications, such as precision farming and management, it would be ideal to obtain crop growth information on the basis of a more precise and timely distribution map [15]. Although the cropland extent map at 30-m resolution or better is an urgent need for food and water security analysis [16], such a province-wide map of rice distribution for China is still unavailable in the literature. Among current medium resolution satellites, Landsat 8 OLI is still a preference for classification research due to its steady and high-quality imaging capability.…”

Section: Introductionmentioning

confidence: 99%

“…Among current medium resolution satellites, Landsat 8 OLI is still a preference for classification research due to its steady and high-quality imaging capability. Sentinel-2 data recently become routinely available at 10 m resolution with a relatively short revisit cycle, which could be an ideal complement to the 30 m Landsat 8 optical imagery [16].…”

Section: Introductionmentioning

confidence: 99%

Evaluation of One-Class Support Vector Classification for Mapping the Paddy Rice Planting Area in Jiangsu Province of China from Landsat 8 OLI Imagery

Jiang

et al. 2018

Remote Sensing

View full text Add to dashboard Cite

Identification of paddy fields is essential for monitoring the rice cultivated area and predicting rice productivity. Timely and accurate extraction of rice distribution can bring vital information for national food security, agricultural policy formulation, and regional environmental sustainability. Conventional classification methods usually suffered from low accuracy, multi-class training samples, or demanding imagery requirements. This paper proposes to use one-class support vector classification (OCSVC) to extract rice cultivated area with Landsat Optical Land Imager (OLI) imagery. Instead of sampling and training all land cover types as performed by multi-class classification methods, OCSVC only used the training samples of target class (rice) for rice mapping. The performance of OCSVC was evaluated in terms of the classification accuracy of rice mapping and rice acreage estimation based on high-resolution imagery, field survey data and rice acreage data from government reports for Jiangsu Province, China. At the county-level, OCSVC was also compared with the commonly used multi-class support vector classification (MCSVC), decision tree classification (DTC), and vegetation index-based thresholding (VIT). Our results demonstrated that OCSVC produced a comparable overall accuracy to DTC and outperformed MCSVC and VIT. The computational efficiency of OCSVC increased approximately ten times as compared to MCSVC. The OCSVC produced the best correlation between its classified area and reported area among the four classification methods evaluated. When applied to the provincial level, the classification overall accuracy for OCSVC was 88.54%. The detected rice planting area for Jiangsu Province was 22,602 km 2 , which was consistent with the statistics from the National Bureau of Statistics (22,948 km 2 ). This OCSVC-based mapping strategy provides a practical and efficient way to detect the rice planting extent with Landsat imagery at a large scale.

show abstract

Nominal 30-m Cropland Extent Map of Continental Africa by Integrating Pixel-Based and Object-Based Algorithms Using Sentinel-2 and Landsat-8 Data on Google Earth Engine

Cited by 315 publications

References 87 publications

Habitat loss in the restricted range of the endemic Ghanaian cichlidLimbochromis robertsi

Habitat loss in the restricted range of the endemic Ghanaian cichlidLimbochromis robertsi

Estimation of different data compositions for early-season crop type classification

Evaluation of One-Class Support Vector Classification for Mapping the Paddy Rice Planting Area in Jiangsu Province of China from Landsat 8 OLI Imagery

Contact Info

Product

Resources

About