National-Scale Cropland Mapping Based on Phenological Metrics, Environmental Covariates, and Machine Learning on Google Earth Engine

Use of Optical and Radar Imagery for Crop Type Classification in Africa: A Review

Multi-source remote sensing-derived information on crops contributes significantly to agricultural monitoring, assessment, and management. In Africa, some challenges (i.e., small-scale farming practices associated with diverse crop types and agricultural system complexity, and cloud coverage during the growing season) can imped agricultural monitoring using multi-source remote sensing. The combination of optical remote sensing and synthetic aperture radar (SAR) data has emerged as an opportune strategy for improving the precision and reliability of crop type mapping and monitoring. This work aims to conduct an extensive review of the challenges of agricultural monitoring and mapping in Africa in great detail as well as the current research progress of agricultural monitoring based on optical and Radar satellites. In this context optical data may provide high spatial resolution and detailed spectral information, which allows for the differentiation of different crop types based on their spectral signatures. However, synthetic aperture radar (SAR) satellites can provide important contributions given the ability of this technology to penetrate cloud cover, particularly in African tropical regions, as opposed to optical data. This review explores various combination techniques employed to integrate optical and SAR data for crop type classification and their applicability and limitations in the context of African countries. Furthermore, challenges are discussed in this review as well as and the limitations associated with optical and SAR data combination, such as the data availability, sensor compatibility, and the need for accurate ground truth data for model training and validation. This study also highlights the potential of advanced modelling (i.e., machine learning algorithms, such as support vector machines, random forests, and convolutional neural networks) in improving the accuracy and automation of crop type classification using combined data. Finally, this review concludes with future research directions and recommendations for utilizing optical and SAR data combination techniques in crop type classification for African agricultural systems. Furthermore, it emphasizes the importance of developing robust and scalable classification models that can accommodate the diversity of crop types, farming practices, and environmental conditions prevalent in Africa. Through the utilization of combined remote sensing technologies, informed decisions can be made to support sustainable agricultural practices, strengthen nutritional security, and contribute to the socioeconomic development of the continent.

Land cover and crop types mapping using different spatial resolution imagery in a Mediterranean irrigated area

Crop type identification is critical for agricultural sustainability policy development and environmental assessments. Therefore, it is important to obtain their spatial distribution via different approaches. Medium-, high- and very high-resolution optical satellite sensors are efficient tools for acquiring this information, particularly for challenging studies such as those conducted in heterogeneous agricultural fields. This research examined the ability of four multitemporal datasets (Sentinel-1-SAR (S1), Sentinel-2-MSI (S2), RapidEye (RE), and PlanetScope (PS)) to identify land cover and crop types (LCCT) in a Mediterranean irrigated area. To map LCCT distribution, a supervised pixel-based classification is adopted using Support Vector Machine with a radial basis function kernel (SVMRB) and Random Forest (RF). Thus, LCCT maps were generated into three levels, including six (Level I), ten (Level II), and fourteen (Level III) classes. Overall, the findings revealed high overall accuracies of >92%, >83%, and > 81% for Level I, Level II, and Level III, respectively, except for Sentinel-1. It was found that accuracy improves considerably when the number of classes decreases, especially when cropland or non-cropland classes are grouped into one. Furthermore, there was a similarity in performance between S2 alone and S1S2. PlanetScope LCCT classifications outperform other sensors. In addition, the present study demonstrated that SVM achieved better performances against RF and can thereby effectively extract LCCT information from high-resolution imagery as PlanetScope.

Extraction and spatiotemporal changes of open-pit mines during 1985-2020 using Google Earth Engine: A case study of Qingzhou City, Shandong Province, China

The global use of mineral resources has increased exponentially for decades and will continue to grow for the foreseeable future, resulting in increasingly negative impacts on the surrounding environment. However, to date, there are a lack of historical and current spatial extent datasets with high accuracy for mining areas in many parts of the world, which has hindered a more comprehensive understanding of the environmental impacts of mining. Using the Google Earth Engine cloud platform and the Landsat normalized difference vegetation index (NDVI) datasets, the spatial extent data of open-pit mining areas for eight years (1985, 1990, 1995, 2000, 2005, 2010, 2015, and 2020) was extracted by the Otsu algorithm. The limestone mining areas in Qingzhou, Shandong Province, China, was selected as a case study. The annual maximum NDVI was first derived from the Landsat NDVI datasets, and then the Otsu algorithm was used to segment the annual maximum NDVI images to obtain the extent of the mining areas. Finally, the spatiotemporal characteristics of the mining areas in the study region were analyzed in reference to previous survey data. The results showed that the mining areas were primarily located in Shaozhuang Town, Wangfu Street and the northern part of Miaozi Town, and the proportion of mining areas within these three administrative areas has increased annually from 88% in 1985 to more than 98% in 2010. Moreover, the open-pit mining areas in ​​Qingzhou gradually expanded from a scattered, point-like distribution to a large, contiguous distribution. From 1985 to 2020, the open-pit mining area expanded to more than 10 times its original size at a rate of 0.5 km²/year. In 2015, this area reached its maximum size of 19.7 km² and slightly decreased in 2020. Furthermore, the expansion of the mining areas in Qingzhou went through three stages: a slow growth period before 1995, a rapid expansion period from 1995 to 2005, and a shutdown and remediation period after 2005. A quantitative accuracy assessment was performed by calculating the Intersection over Union (IoU) of the extraction results and the visual interpretation results from Gaofen-2 images with 1-m spatial resolution. The IoU reached 72%. The results showed that it was feasible to threshold the Landsat annual maximum NDVI data by the Otsu algorithm to extract the annual spatial extent of the open-pit mining areas. Our method will be easily transferable to other regions worldwide, enabling the monitoring of mine environments.

Quantifying the Impacts of the 2020 Flood on Crop Production and Food Security in the Middle Reaches of the Yangtze River, China

This study uses satellite imagery and geospatial data to examine the impact of floods over the main planting areas for double-cropping rice and grain crops in the middle reaches of the Yangtze River. During summer 2020, a long-lasting 62-day heavy rainfall caused record-breaking floods over large areas of China, especially the Yangtze basin. Through close examination of Sentinel-1/2 satellite imagery and Copernicus Global Land Cover, between July and August 2020, the inundation area reached 21,941 and 23,063 km2, and the crop-affected area reached 11,649 and 11,346 km2, respectively. We estimated that approximately 4.66 million metric tons of grain crops were seriously affected in these two months. While the PRC government denied that food security existed, the number of Grains and Feeds imported from the U.S. between January to July 2021 increased by 316%. This study shows that with modern remote sensing techniques, stakeholders can obtain critical estimates of large-scale disaster events much earlier than other indicators, such as disaster field surveys or crop price statistics. Potential use could include but is not limited to monitoring floods and land use coverage changes.

Land Use and Land Cover Mapping Using Sentinel-2, Landsat-8 Satellite Images, and Google Earth Engine: A Comparison of Two Composition Methods

Accurate and real-time land use/land cover (LULC) maps are important to provide precise information for dynamic monitoring, planning, and management of the Earth. With the advent of cloud computing platforms, time series feature extraction techniques, and machine learning classifiers, new opportunities are arising in more accurate and large-scale LULC mapping. In this study, we aimed at finding out how two composition methods and spectral–temporal metrics extracted from satellite time series can affect the ability of a machine learning classifier to produce accurate LULC maps. We used the Google Earth Engine (GEE) cloud computing platform to create cloud-free Sentinel-2 (S-2) and Landsat-8 (L-8) time series over the Tehran Province (Iran) as of 2020. Two composition methods, namely, seasonal composites and percentiles metrics, were used to define four datasets based on satellite time series, vegetation indices, and topographic layers. The random forest classifier was used in LULC classification and for identifying the most important variables. Accuracy assessment results showed that the S-2 outperformed the L-8 spectral–temporal metrics at the overall and class level. Moreover, the comparison of composition methods indicated that seasonal composites outperformed percentile metrics in both S-2 and L-8 time series. At the class level, the improved performance of seasonal composites was related to their ability to provide better information about the phenological variation of different LULC classes. Finally, we conclude that this methodology can produce LULC maps based on cloud computing GEE in an accurate and fast way and can be used in large-scale LULC mapping.