Yuqi Cheng

A new research paradigm for global land cover mapping

ABSTRACT In this paper, we introduced major challenges in mapping croplands, settlements, water and wetlands, and discussed challenges in the use of multi-temporal and multi-sensor data. We then summarized some of the on-going efforts in improving qualities of global land cover maps. Existing technologies provide sufficient data for better map making if extra efforts can be made instead of harmonizing and integrating various global land cover products. Developing and selecting better algorithms, including more input variables (new types of data or features) for classification, having representative training samples are among conventional measures generally believed effective in improving mapping accuracies at local scales. We pointed out that data were more important in improving mapping accuracies than algorithms. Finally, we proposed a new paradigm for global land cover mapping, which included a view of vegetation classes based on their types and form, canopy cover and height. The new paradigm suggests that a universally applicable training sample set is not only possible but also effective in improving land cover classification at the continental and global scales. To ensure an easy transition from traditional land cover mapping to the new paradigm, we recommended that an all-in-one data management and analysis system be constructed.

Oil palm mapping using Landsat and PALSAR: a case study in Malaysia

ABSTRACT Irrespective of the positive economic benefit or negative environmental impact of the rapid expansion of oil palm plantations in tropical regions, it is important to be able to create accurate land-cover maps for such areas. Optical remote sensing is vulnerable to the effects of clouds, which can limit data availability for the oil palm plantation areas in the humid tropics. The satellite-flown PALSAR (Phased Array type L-band Synthetic Aperture Radar) instrument, which provides all-day/all-weather Earth observations, offers the opportunity to identify and map oil palm plantations in cloudy regions. This study used a Support Vector Machine (SVM) classifier and a Mahalanobis distance (MD) classifier to undertake supervised classifications of Landsat, PALSAR, and combined Landsat and PALSAR data (Landsat+PALSAR) for two locations in peninsular Malaysia. Results indicate that accuracies from Landsat+PALSAR are better than accuracies from Landsat and PALSAR along for both study areas using both classifiers. The extents of the oil palm areas estimated from these maps were compared with values obtained through human photointerpretation of Google Earth™ images in previous studies. Based on the R2 statistics, it was established that the Landsat+PALSAR combination performed best for both study areas and demonstrated good potential for oil palm plantation mapping.

Monitoring cropland changes along the Nile River in Egypt over past three decades 1984–2015 using remote sensing

ABSTRACT The Nile River basin is the main agricultural area in Egypt. In recent decades, human activities and climate change have remarkably influenced the ecological environment there. Those changes have caused land degradation, sea level rise, and conflicts between land and population, threatening the agricultural system and food security of Egypt. In this study, cropland mapping along the Nile in Egypt over the past three decades (1984–2015) was conducted at annual frequency, using 961 Landsat TM/ETM+/OLI images. Spectral features of selected growing season images and band ratio-based indices were used in supervised classification. Thereafter, terrain and time series information were used to filter possible classification errors on the basis of logical judgment and statistical analysis. The average overall classification accuracy of cropland was greater than 90%. Furthermore, temporal and spatial characteristics of cropland expansion were analysed. The results highlight the annual geographical distribution of cropland dynamics from the Nile Valley to desert. In total, cropland areas had increased by 33.7% from 2848.1 kha in 1984 to 3807.8 kha in 2015, with an annual average increase of 31.0 kha in these 32 years.

Using a global reference sample set and a cropland map for area estimation in China

A technically transparent and freely available reference sample set for validation of global land cover mapping was recently established to assess the accuracies of land cover maps with multiple resolutions. This sample set can be used to estimate areas because of its equal-area hexagon-based sampling design. The capabilities of these sample set-based area estimates for cropland were investigated in this paper. A 30-m cropland map for China was consolidated using three thematic maps (cropland, forest and wetland maps) to reduce confusion between cropland and forest/wetland. We compared three area estimation methods using the sample set and the 30 m cropland map. The methods investigated were: (1) pixel counting from a complete coverage map, (2) direct estimation from reference samples, and (3) model-assisted estimation combining the map with samples. Our results indicated that all three methods produced generally consistent estimates which agreed with cropland area measured from an independent national land use dataset. Areas estimated from the reference sample set were less biased by comparing with a National Land Use Dataset of China (NLUD-C). This study indicates that the reference sample set can be used as an alternative source to estimate areas over large regions.

Towards a global oil palm sample database: design and implications

ABSTRACT Global oil palm plantations have expanded in the last few decades, resulting in negative impacts on the environment. Satellite remote sensing plays an important role in monitoring the expansion of oil palm plantations, but requires high-quality ground samples for training and validation. To facilitate the monitoring of oil palm plantations on a large scale, we propose an oil palm sample database that includes the five countries with the largest areas of oil palm plantations: Indonesia, Malaysia, Nigeria, Thailand, and Ghana. In total, 45,896 samples were collected using a hexagonal sampling design. High-resolution images from Google Earth, the Advanced Land Observing Satellite (ALOS) Phased Array type L-band Synthetic Aperture Radar (PALSAR) images, and Landsat optical images were used to identify oil palm plantations and other types of land cover (croplands, forests, grasslands, shrublands, water, hard surfaces, and bare land). The characteristics of oil palm cover and its environment, including PALSAR backscattering coefficients, terrain, and climate recorded in this database are also discussed. The results indicate that using the PALSAR band algebra threshold alone is not recommended to distinguish oil palm from other land-cover/use types.

Mapping oil palm extent in Malaysia using ALOS-2 PALSAR-2 data

ABSTRACT The extent of oil palm plantations has increased rapidly in Malaysia over the past few decades. To evaluate ecological effects and economic values, it is important to produce an accurate oil palm map for Malaysia. The Phased Array Type L-band Synthetic Aperture Radar (PALSAR) on the Advance Land Observing Satellite (ALOS) is useful in land-cover mapping in tropical regions under all-weather conditions. In this study, PALSAR-2 images from 2015 were used for oil palm mapping with maximum likelihood classifier (MLC)-based supervised classification. The processed PALSAR-2 data were resampled to multiple coarser resolutions (50, 100, 250, 500, and 1000 m), and then used to investigate the effect of speckle in oil palm mapping. Both independent testing samples and inventories from the Malaysia Palm Oil Board (MPOB) were used to evaluate the mapping accuracy. The oil palm mapping result indicates 50–500 m to be a good resolution for either retaining spatial details or reducing speckle noise of PALSAR-2 images. Among which, the best overall mapping accuracies and average oil palm accuracies reached 94.50% and 89.78%, respectively. Moreover, the oil palm area derived from the 100-m resolution map is 6.14 million hectares (Mha), which is the closest to the official MPOB inventories (~8.87% overestimation).

A multiple dataset approach for 30-m resolution land cover mapping: a case study of continental Africa

ABSTRACT Recent developments in global land-cover mapping have focused on spatial resolution improvement with more heterogeneous features to integrate spatial, spectral and temporal information. In this study, hundreds of features derived from four seasonal Landsat 8 OLI (Operational Land Imager) spectral bands, Moderate Resolution Imaging Spectroradiometer (MODIS) time series vegetation index (VI) data, night-time light (NTL), digital elevation models (DEM) and climatic variables were used for land cover mapping with a target 30-m resolution for the whole African continent. In total, 49,007 training samples (from 11,231 locations) and 23,803 validation samples (from 5,414 locations) interpreted from seasonal Landsat, MODIS Normalized Difference Vegetation Index (NDVI) time series and high-resolution images in Google Earth were used for classifier training (Random Forest) and map validation. Overall accuracy was 76% at 30-m spatial resolution, which is better than previous land cover mapping for the African continent. Besides, accuracies for cropland were improved dramatically by more than 10%. Our method also addressed many remaining issues for 30-m mapping (e.g. boundary effects and declines in resolution). This framework is promising for automatic and efficient global land cover mapping resulting in better visual effects and classification accuracy.

Towards global oil palm plantation mapping using remote-sensing data

ABSTRACT In recent decades, palm oil, which forms one of the world’s major bulk feedstock and oil crops, has been cultivated at an increasing scale to meet new demand. Oil palm expansion has driven not only socio-economic development but also serious ecological problems and environmental pollution through deforestation and fires to clear the forests. Uneconomic oil palm plantations can influence the balance of regional ecosystems and the carbon cycle. Many countries report national statistics on the area of oil palm, but few document the extent and locations of oil palm plantations. In this study, we produce and make available oil palm maps that include 15 countries with more than 10,000 ha of planted oil palms. Phased Array Type L-band Synthetic Aperture (PALSAR-2) images and high-resolution (<2.5 m) images in Google Earth were used to produce oil palm maps by supervised classification and visual interpretation. Two independent verification systems were used to evaluate map accuracy. The characteristics of oil palm plantations distribution and their environment suitability including terrain and climate conditions of the global oil palm planted regions are also discussed. The results indicate that the total area of oil palm in global in 2016 was estimated to be 29.49 million hectares (Mha) although the mapping result showed a good correlation with other records, but relatively large uncertainty in Africa. Most oil palm trees grow in warm (24–29.5°C), wet conditions (1000–4000 mm p.a. of precipitation), flat terrain (slope less than 8°), and low elevation (0–800 m); however, these growing conditions are slightly different in different continents.

Exploring the temporal density of Landsat observations for cropland mapping: experiments from Egypt, Ethiopia, and South Africa

ABSTRACT Accurate land-use/land-cover mapping based on remote-sensing images depends on clear and frequent observations. This study aimed to explore how many Landsat images were needed within a year and when they should be acquired, for cropland mapping in Africa. Three Landsat footprints in Egypt (Path/Row: 177/039, 127 images), Ethiopia (Path/Row: 168/054, 98 images), and South Africa (Path/Row: 170/078, 207 images) from 1984 to 2016 were used together with spectral indices and a 30-m digital elevation model in a random forest-based supervised classification. Detailed exploration was conducted into the number and temporal distribution of Landsat images required. Our results indicated that average cropland mapping accuracies for these three sites ranged from 81.17% to 87.59% (Egypt), 54.43% to 79.72% (Ethiopia), and 28.11% to 59.35% (South Africa) using different numbers of images within a year. The overall cropland accuracies were improved with an increase in available Landsat images within a year and reached a relatively stable stage when more than five images were acquired in all three sites. Growing season images played a key role in identifying cropland, accounting for a 13.22% average accuracy improvement compared with non-growing season images. Therefore, at least five images are recommended from a computational efficiency perspective, although fewer images, as low as two growing season images, can also achieve good results in specific regions.