Synergistic Use of Radar Sentinel-1 and Optical Sentinel-2 Imagery for Crop Mapping: A Case Study for Belgium

Radar vegetation indices for monitoring surface vegetation: Developments, challenges, and trends

Monitoring surface vegetation is essential for environmental protection, disaster prevention, and carbon sequestration in forests. However, optical remote-sensing methods and their derivative technologies typically fail to fully meet this requirement due to constraints such as lighting and weather. Radar vegetation indices (RVIs), developed based on microwave remote-sensing data, describe the dielectric properties and morphological structure of vegetation and have been applied for vegetation monitoring at various scales. This technical review is the first to systematically summarize RVIs; it analyzes and discusses their principles, developments, categories and applications, and provides a comprehensive guide for their use. Additionally, the challenges faced by RVIs, as well as their applicability, were analyzed, and future improvements and development trends were carefully projected. The selection of RVIs must consider the type of data used, the terrain and location of the study area, and the major vegetation types. The effectiveness of RVIs applied to vegetation monitoring can be affected by various factors, including index performance, sensor type, study area, and data type and quality. These factors reduce the reliability and robustness of results, as well as guide the improvement direction of RVIs. The development of technologies, such as artificial intelligence, in remote sensing offers new possibilities for RVIs, enabling the removal of background scattering, improvement in interpretation accuracy, and reduction in application thresholds. Additionally, the development trends in high resolution, multi-polarization, multi-base, multi-dimensional, and networked synthetic aperture radar (SAR) and their satellite platforms offer data support for the next generation of RVIs. The rapid development of RVIs strongly supports the use of surface vegetation monitoring and terrestrial ecosystem research.

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.

Crop classification by using dual-pol SAR vegetation indices derived from Sentinel-1 SAR-C data

In the following study, an attempt is made for crop classification of rainy season through analyzing time-series Sentinel-1 SAR data of May 2020 to September 2020. The SVI_DP index derived from dual-pol (VV and VH) bands consisting of NRPB ([Formula: see text]), DPDD [Formula: see text]), IDPDD ([Formula: see text]), and VDDPI [Formula: see text] ratios are utilized for discriminating inter-vegetative boundaries of crop pixels. This study was conducted near Karnal city region, Karnal district, Haryana, India. The Sentinel-1 data has the capability to penetrate thick cloud cover and provide high revisit frequency data for rain-fed crops. Obtained classification achieved higher accuracy in both RF (93.77%) and SVM (93.50%) classifiers. Obtained linear regression statistics of mean raster imagery reveals that IDPDD index is much sensitive to other crop which has highest standard deviations in σ_vh° and σ_vv° bands throughout the period, and high R² with σ_vh° (0.70), VV (0.58), NRPB (0.693), and DPDD (0.697) indices. In contrast to this, IDPDD index has least correlation (< 0.289) with σ_vh°, σ_vv°, EVI 2, NRPB, and DPDD indices for water body which has smooth surface and lowest SAR backscattering with minimum standard deviations in the same period.

Investigating the effects of drought and lockdowns on smallholder and commercial agricultural production in KwaZulu-Natal using remotely sensed data

Not many efforts have been made so far to understand the effects of both the 2015-2016 drought and the 2020 lockdown measures on the agricultural production of smallholder vis-a-vis commercial farmers in Kwazulu-Natal. Google Earth Engine, and random forest algorithm, are used to generate a dataset that help to investigate this question. A regression is performed on double differenced data to investigate the effects of interest. A k-mean cluster analysis, is also used to determine whether the distribution patterns of crop production changed with drought and disruption of agricultural production input. Results show that: (1) droughts affected the agricultural production of both areas similarly. Crop cover declined in both areas for one season after droughts were broken. Then recovery was driven by greener, more productive crops rather than the expansion of crop area. (2) The response of both areas to the COVID-19 lockdown was also similar. Both smallholder and commercial areas' Normalised Difference Vegetation Index - a proxy for crop vitality - improved in response to regulations favourable to the sector and improved rainfall. No significant adjustments in crop cover were observed. Production therefore changed primarily at the intensive margin (improved productivity of existing croplands) rather than the extensive (changing the extent of land under cultivation). (3) Cluster analysis allows for a more granular view, showing that the positive impact of lockdowns on agriculture were concentrated in areas with high rainfall and close proximity to metropolitan markets. Both smallholder and commercial farmers therefore are reliant on market access together with favourable environmental conditions for improved production.

ELULC-10, a 10 m European Land Use and Land Cover Map Using Sentinel and Landsat Data in Google Earth Engine

Land Use/Land Cover (LULC) maps can be effectively produced by cost-effective and frequent satellite observations. Powerful cloud computing platforms are emerging as a growing trend in the high utilization of freely accessible remotely sensed data for LULC mapping over large-scale regions using big geodata. This study proposes a workflow to generate a 10 m LULC map of Europe with nine classes, ELULC-10, using European Sentinel-1/-2 and Landsat-8 images, as well as the LUCAS reference samples. More than 200 K and 300 K of in situ surveys and images, respectively, were employed as inputs in the Google Earth Engine (GEE) cloud computing platform to perform classification by an object-based segmentation algorithm and an Artificial Neural Network (ANN). A novel ANN-based data preparation was also presented to remove noisy reference samples from the LUCAS dataset. Additionally, the map was improved using several rule-based post-processing steps. The overall accuracy and kappa coefficient of 2021 ELULC-10 were 95.38% and 0.94, respectively. A detailed report of the classification accuracies was also provided, demonstrating an accurate classification of different classes, such as Woodland and Cropland. Furthermore, rule-based post processing improved LULC class identifications when compared with current studies. The workflow could also supply seasonal, yearly, and change maps considering the proposed integration of complex machine learning algorithms and large satellite and survey data.

Mapping Crop Types of Germany by Combining Temporal Statistical Metrics of Sentinel-1 and Sentinel-2 Time Series with LPIS Data

Nationwide and consistent information on agricultural land use forms an important basis for sustainable land management maintaining food security, (agro)biodiversity, and soil fertility, especially as German agriculture has shown high vulnerability to climate change. Sentinel-1 and Sentinel-2 satellite data of the Copernicus program offer time series with temporal, spatial, radiometric, and spectral characteristics that have great potential for mapping and monitoring agricultural crops. This paper presents an approach which synergistically uses these multispectral and Synthetic Aperture Radar (SAR) time series for the classification of 17 crop classes at 10 m spatial resolution for Germany in the year 2018. Input data for the Random Forest (RF) classification are monthly statistics of Sentinel-1 and Sentinel-2 time series. This approach reduces the amount of input data and pre-processing steps while retaining phenological information, which is crucial for crop type discrimination. For training and validation, Land Parcel Identification System (LPIS) data were available covering 15 of the 16 German Federal States. An overall map accuracy of 75.5% was achieved, with class-specific F1-scores above 80% for winter wheat, maize, sugar beet, and rapeseed. By combining optical and SAR data, overall accuracies could be increased by 6% and 9%, respectively, compared to single sensor approaches. While no increase in overall accuracy could be achieved by stratifying the classification in natural landscape regions, the class-wise accuracies for all but the cereal classes could be improved, on average, by 7%. In comparison to census data, the crop areas could be approximated well with, on average, only 1% of deviation in class-specific acreages. Using this streamlined approach, similar accuracies for the most widespread crop types as well as for smaller permanent crop classes were reached as in other Germany-wide crop type studies, indicating its potential for repeated nationwide crop type mapping.

Machine Learning Classification of Fused Sentinel-1 and Sentinel-2 Image Data towards Mapping Fruit Plantations in Highly Heterogenous Landscapes

Mapping smallholder fruit plantations using optical data is challenging due to morphological landscape heterogeneity and crop types having overlapping spectral signatures. Furthermore, cloud covers limit the use of optical sensing, especially in subtropical climates where they are persistent. This research assessed the effectiveness of Sentinel-1 (S1) and Sentinel-2 (S2) data for mapping fruit trees and co-existing land-use types by using support vector machine (SVM) and random forest (RF) classifiers independently. These classifiers were also applied to fused data from the two sensors. Feature ranks were extracted using the RF mean decrease accuracy (MDA) and forward variable selection (FVS) to identify optimal spectral windows to classify fruit trees. Based on RF MDA and FVS, the SVM classifier resulted in relatively high classification accuracy with overall accuracy (OA) = 0.91.6% and kappa coefficient = 0.91% when applied to the fused satellite data. Application of SVM to S1, S2, S2 selected variables and S1S2 fusion independently produced OA = 27.64, Kappa coefficient = 0.13%; OA= 87%, Kappa coefficient = 86.89%; OA = 69.33, Kappa coefficient = 69. %; OA = 87.01%, Kappa coefficient = 87%, respectively. Results also indicated that the optimal spectral bands for fruit trees mapping are green (B3) and SWIR_2 (B10) for S2, whereas for S1, the vertical-horizontal (VH) polarization band. Including the textural metrics from the VV channel improved crop discrimination and co-existing land use cover types. The fusion approach proved robust and well suited for accurate smallholder fruit plantation mapping.

Performance and the Optimal Integration of Sentinel-1/2 Time-Series Features for Crop Classification in Northern Mongolia

Accurate and early crop-type maps are essential for agricultural policy development and food production assessment at regional and national levels. This study aims to produce a crop-type map with acceptable accuracy and spatial resolution in northern Mongolia by optimizing the combination of Sentinel-1 (S1) and Sentinel-2 (S2) images with the Google Earth Engine (GEE) environment. A total of three satellite data combination scenarios are set, including S1 alone, S2 alone, and the combination of S1 and S2. In order to avoid the impact of data gaps caused by clouds on crop classification, this study reconstructed the time series of S1 and S2 with a 10-day interval using the median composite method, linear moving interpolation, and Savitzky–Golay (SG) filter. Our results indicated that crop-type classification accuracy increased with the increase in data length to all three data combination scenarios. S2 alone has higher accuracy than S1 alone and the combination of S1 and S2. The crop-type map with the highest accuracy was generated using S2 data from 150 days of the year (DOY) (11 May) to 260 DOY (18 September). The OA and kappa were 0.93 and 0.78, respectively, and the F1-score for spring wheat and rapeseed were 0.96 and 0.80, respectively. The classification accuracy of the crop increased rapidly from 210 DOY (end of July) to 260 DOY (August to mid-September), and then it remained stable after 260 DOY. Based on our analysis, we filled the gap of the crop-type map with 10 m spatial resolution in northern Mongolia, revealing the best satellite combination and the best period for crop-type classification, which can benefit the achievement of sustainable development goals 2 (SDGs2).

Spatial Transferability of Random Forest Models for Crop Type Classification Using Sentinel-1 and Sentinel-2

Large-scale crop type mapping often requires prediction beyond the environmental settings of the training sites. Shifts in crop phenology, field characteristics, or ecological site conditions in the previously unseen area, may reduce the classification performance of machine learning classifiers that often overfit to the training sites. This study aims to assess the spatial transferability of Random Forest models for crop type classification across Germany. The effects of different input datasets, i.e., only optical, only Synthetic Aperture Radar (SAR), and optical-SAR data combination, and the impact of spatial feature selection were systematically tested to identify the optimal approach that shows the highest accuracy in the transfer region. The spatial feature selection, a feature selection approach combined with spatial cross-validation, should remove features that carry site-specific information in the training data, which in turn can reduce the accuracy of the classification model in previously unseen areas. Seven study sites distributed over Germany were analyzed using reference data for the major 11 crops grown in the year 2018. Sentinel-1 and Sentinel-2 data from October 2017 to October 2018 were used as input. The accuracy estimation was performed using the spatially independent sample sets. The results of the optical-SAR combination outperformed those of single sensors in the training sites (maximum F1-score–0.85), and likewise in the areas not covered by training data (maximum F1-score–0.79). Random forest models based on only SAR features showed the lowest accuracy losses when transferred to unseen regions (average F1loss–0.04). In contrast to using the entire feature set, spatial feature selection substantially reduces the number of input features while preserving good predictive performance on unseen sites. Altogether, applying spatial feature selection to a combination of optical-SAR features or using SAR-only features is beneficial for large-scale crop type classification where training data is not evenly distributed over the complete study region.

Integration of Sentinel-3 and MODIS Vegetation Indices with ERA-5 Agro-Meteorological Indicators for Operational Crop Yield Forecasting

Timely crop yield forecasts at a national level are substantial to support food policies, to assess agricultural production, and to subsidize regions affected by food shortage. This study presents an operational crop yield forecasting system for Poland that employs freely available satellite and agro-meteorological products provided by the Copernicus programme. The crop yield predictors consist of: (1) Vegetation condition indicators provided daily by Sentinel-3 OLCI (optical) and SLSTR (thermal) imagery, (2) a backward extension of Sentinel-3 data (before 2018) derived from cross-calibrated MODIS data, and (3) air temperature, total precipitation, surface radiation, and soil moisture derived from ERA-5 climate reanalysis generated by the European Centre for Medium-Range Weather Forecasts. The crop yield forecasting algorithm is based on thermal time (growing degree days derived from ERA-5 data) to better follow the crop development stage. The recursive feature elimination is used to derive an optimal set of predictors for each administrative unit, which are ultimately employed by the Extreme Gradient Boosting regressor to forecast yields using official yield statistics as a reference. According to intensive leave-one-year-out cross validation for the 2000–2019 period, the relative RMSE for voivodships (NUTS-2) are: 8% for winter wheat, and 13% for winter rapeseed and maize. Respectively, for municipalities (LAU) it equals 14% for winter wheat, 19% for winter rapeseed, and 27% for maize. The system is designed to be easily applicable in other regions and to be easily adaptable to cloud computing environments such as Data and Information Access Services (DIAS) or Amazon AWS, where data sets from the Copernicus programme are directly accessible.

An Interannual Transfer Learning Approach for Crop Classification in the Hetao Irrigation District, China

Crop type classification is critical for crop production estimation and optimal water allocation. Crop type data are challenging to generate if crop reference data are lacking, especially for target years with reference data missed in collection. Is it possible to transfer a trained crop type classification model to retrace the historical spatial distribution of crop types? Taking the Hetao Irrigation District (HID) in China as the study area, this study first designed a 10 m crop type classification framework based on the Google Earth Engine (GEE) for crop type mapping in the current season. Then, its interannual transferability to accurately retrace historical crop distributions was tested. The framework used Sentinel-1/2 data as the satellite data source, combined percentile, and monthly composite approaches to generate classification metrics and employed a random forest classifier with 300 trees for crop classification. Based on the proposed framework, this study first developed a 10 m crop type map of the HID for 2020 with an overall accuracy (OA) of 0.89 and then obtained a 10 m crop type map of the HID for 2019 with an OA of 0.92 by transferring the trained model for 2020 without crop reference samples. The results indicated that the designed framework could effectively identify HID crop types and have good transferability to obtain historical crop type data with acceptable accuracy. Our results found that SWIR1, Green, and Red Edge2 were the top three reflectance bands for crop classification. The land surface water index (LSWI), normalized difference water index (NDWI), and enhanced vegetation index (EVI) were the top three vegetation indices for crop classification. April to August was the most suitable time window for crop type classification in the HID. Sentinel-1 information played a positive role in the interannual transfer of the trained model, increasing the OA from 90.73% with Sentinel 2 alone to 91.58% with Sentinel-1 and Sentinel-2 together.

Monitoring Irrigation Events and Crop Dynamics Using Sentinel-1 and Sentinel-2 Time Series

Capturing and identifying field-based agricultural activities, such as the start, duration and end of irrigation, together with crop sowing/germination, growing period and time of harvest, offer informative metrics that can assist in precision agricultural activities in addition to broader water and food security monitoring efforts. While optically based band-ratios, such as the normalized difference vegetation index (NDVI) and normalized difference water index (NDWI), have been used as descriptors for monitoring crop dynamics, data are not always available due to the influence of clouds and other atmospheric effects on optical sensors. Satellite-based microwave systems, such as the synthetic aperture radar (SAR), offer an all-weather advantage in monitoring soil and crop conditions. In this paper, we leverage the relative strengths of both optical- and microwave-based approaches by combining high resolution Sentinel-1 SAR and Sentinel-2 optical imagery to monitor irrigation events and crop dynamics in a dryland agricultural landscape. A microwave backscatter model was used to analyze the responses of simulated backscatters to soil moisture, NDVI and NDWI (both are correlated with vegetation water content and can be regarded as vegetation descriptors), allowing an empirical relationship between these two platforms. A correlation analysis was also performed using Sentinel-1 SAR and Sentinel-2 optical data over crops of maize, alfalfa, carrot and Rhodes grass in Al Kharj farm of Saudi Arabia to identify an appropriate SAR-based vegetation descriptor. The results illustrate the relationship between SAR and both NDVI and NDWI and demonstrated the relationship between the cross-polarization ratio (VH/VV) and the two optical indices. We explore the capacity of this multi-platform and multi-sensor approach to inform on the spatio-temporal dynamics of a range of agricultural activities, which can be used to facilitate field-based management decisions.

Multi-Dimensional Dataset of Open Data and Satellite Images for Characterization of Food Security and Nutrition

Background

Nutrition is one of the main factors affecting the development and quality of life of a person. From a public health perspective, food security is an essential social determinant for promoting healthy nutrition. Food security embraces four dimensions: physical availability of food, economic and physical access to food, food utilization, and the sustainability of the dimensions above. Integrally addressing the four dimensions is vital. Surprisingly most of the works focused on a single dimension of food security: the physical availability of food.

Objective

The paper proposes a multi-dimensional dataset of open data and satellite images to characterize food security in the department of Cauca, Colombia.

Methods

The food security dataset integrates multiple open data sources; therefore, the Cross-Industry Standard Process for Data Mining methodology was used to guide the construction of the dataset. It includes sources such as population and agricultural census, nutrition surveys, and satellite images.

Results

An open multidimensional dataset for the Department of Cauca with 926 attributes and 9 rows (each row representing a Municipality) from multiple sources in Colombia, is configured. Then, machine learning models were used to characterize food security and nutrition in the Cauca Department. As a result, The Food security index calculated for Cauca using a linear regression model (Mean Absolute Error of 0.391) is 57.444 in a range between 0 and 100, with 100 the best score. Also, an approach for extracting four features (Agriculture, Habitation, Road, Water) of satellite images were tested with the ResNet50 model trained from scratch, having the best performance with a macro-accuracy, macro-precision, macro-recall, and macro-F1-score of 91.7, 86.2, 66.91, and 74.92%, respectively.

Conclusion

It shows how the CRISP-DM methodology can be used to create an open public health data repository. Furthermore, this methodology could be generalized to other types of problems requiring the creation of a dataset. In addition, the use of satellite images presents an alternative for places where data collection is challenging. The model and methodology proposed based on open data become a low-cost and effective solution that could be used by decision-makers, especially in developing countries, to support food security planning.

Density Estimates as Representations of Agricultural Fields for Remote Sensing-Based Monitoring of Tillage and Vegetation Cover

We consider the use of remote sensing for large-scale monitoring of agricultural land use, focusing on classification of tillage and vegetation cover for individual field parcels across large spatial areas. From the perspective of remote sensing and modelling, field parcels are challenging as objects of interest due to highly varying shape and size but relatively uniform pixel content and texture. To model such areas we need representations that can be reliably estimated already for small parcels and that are invariant to the size of the parcel. We propose representing the parcels using density estimates of remote imaging pixels and provide a computational pipeline that combines the representation with arbitrary supervised learning algorithms, while allowing easy integration of multiple imaging sources. We demonstrate the method in the task of the automatic monitoring of autumn tillage method and vegetation cover of Finnish crop fields, based on the integrated analysis of intensity of Synthetic Aperture Radar (SAR) polarity bands of the Sentinel-1 satellite and spectral indices calculated from Sentinel-2 multispectral image data. We use a collection of 127,757 field parcels monitored in April 2018 and annotated to six tillage method and vegetation cover classes, reaching 70% classification accuracy for test parcels when using both SAR and multispectral data. Besides this task, the method could also directly be applied for other agricultural monitoring tasks, such as crop yield prediction.

Mapping Winter Wheat with Optical and SAR Images Based on Google Earth Engine in Henan Province, China

The timely and accurate acquisition of winter wheat acreage is crucial for food security. This study investigated the feasibility of extracting the spatial distribution map of winter wheat in Henan Province by using synthetic aperture radar (SAR, Sentinel-1A) and optical (Sentinel-2) images. Firstly, the SAR images were aggregated based on the growth period of winter wheat, and the optical images were aggregated based on the moderate resolution imaging spectroradiometer normalized difference vegetation index (MODIS-NDVI) curve. Then, five spectral features, two polarization features, and four texture features were selected as feature variables. Finally, the Google Earth Engine (GEE) cloud platform was employed to extract winter wheat acreage through the random forest (RF) algorithm. The results show that: (1) aggregated images based on the growth period of winter wheat and sensor characteristics can improve the mapping accuracy and efficiency; (2) the extraction accuracy of using only SAR images was improved with the accumulation of growth period. The extraction accuracy of using the SAR images in the full growth period reached 80.1%; and (3) the identification effect of integrated images was relatively good, which makes up for the shortcomings of SAR and optical images and improves the extraction accuracy of winter wheat.

Land Cover Changes after the Massive Rohingya Refugee Influx in Bangladesh: Neo-Classic Unsupervised Approach

The Rohingya refugee influx to Bangladesh in 2017 was a historical incident; the number of refugees was so massive that significant impacts to local communities was inevitable. The Bangladesh government provided land in a preserved area for constructing makeshift camps for the refugees. Previous studies have revealed the land cover changes and impacts of the refugee influx around campsites, especially with regard to local forest resources. Our aim is to establish a convenient approach of providing up-to-date information to monitor holistic local situations. We employed a classic unsupervised technique—a combination of k-means clustering and maximum likelihood estimation—with the latest rich time-series satellite images of Sentinal-1 and Sentinal-2. A combination of VV and normalized difference water index (NDWI) images was successful in identifying built-up/disturbed areas, and a combination of VH and NDWI images was successful in differentiating wetland/saltpan, agriculture /open field, degraded forest/bush, and forest areas. By doing this, we provided annual land cover classification maps for the entire Teknaf peninsula for the pre- and post-influx periods with both fair quality and without prior training data. Our analyses revealed that on-going impacts were still observed by May 2021. As a simple estimation of the intervention consequence, the built-up/disturbed areas increased 6825 ha (compared with the 2015–17 period). However, while the impacts on the original forest were not found to be significant, the degraded forest/bush areas were largely degraded by 4606 ha. These cultivated lands would be used for agricultural activities. This is in line with the reported farmers’ increased income, despite local people with other occupations that are all equally facing the decreases in income. The convenience of our unsupervised classification approach would help keep accumulating a time-series land cover classification, which is important in monitoring impacts on local communities.

A Convolutional Neural Network Algorithm for Soil Moisture Prediction from Sentinel-1 SAR Images

Achieving the rational, optimal, and sustainable use of resources (water and soil) is vital to drink and feed 9.725 billion by 2050. Agriculture is the first source of food production, the biggest consumer of freshwater, and the natural filter of air purification. Hence, smart agriculture is a “ray of hope” in regard to food, water, and environmental security. Satellites and artificial intelligence have the potential to help agriculture flourish. This research is an essential step towards achieving smart agriculture. Prediction of soil moisture is important for determining when to irrigate and how much water to apply, to avoid problems associated with over- and under-watering. This also contributes to an increase in the number of areas being cultivated and, hence, agricultural productivity and air purification. Soil moisture measurement techniques, in situ, are point measurements, tedious, time-consuming, expensive, and labor-intensive. Therefore, we aim to provide a new approach to detect moisture content in soil without actually being in contact with it. In this paper, we propose a convolutional neural network (CNN) architecture that can predict soil moisture content over agricultural areas from Sentinel-1 images. The dual-pol (VV–VH) Sentinel-1 SAR data have being utilized (V = vertical, H = horizontal). The CNN model is composed of six convolutional layers, one max-pooling layer, one flatten layer, and one fully connected layer. The total number of Sentinel-1 images used for running CNN is 17,325 images. The best values of the performance metrics (coefficient of determination (R2=0.8664), mean absolute error (MAE=0.0144), and root mean square error (RMSE=0.0274)) have been achieved due to the use of Sigma naught VH and Sigma naught VV as input data to the CNN architecture (C). Results show that VV polarization is better than VH polarization for soil moisture retrieval, and that Sigma naught, Gamma naught, and Beta naught have the same influence on soil moisture estimation.

Synergy of Sentinel-1 and Sentinel-2 Imagery for Early Seasonal Agricultural Crop Mapping

The exploitation of the unprecedented capacity of Sentinel-1 (S1) and Sentinel-2 (S2) data offers new opportunities for crop mapping. In the framework of the SenSAgri project, this work studies the synergy of very high-resolution Sentinel time series to produce accurate early seasonal binary cropland mask and crop type map products. A crop classification processing chain is proposed to address the following: (1) high dimensionality challenges arising from the explosive growth in available satellite observations and (2) the scarcity of training data. The two-fold methodology is based on an S1-S2 classification system combining the so-called soft output predictions of two individually trained classifiers. The performances of the SenSAgri processing chain were assessed over three European test sites characterized by different agricultural systems. A large number of highly diverse and independent data sets were used for validation experiments. The agreement between independent classification algorithms of the Sentinel data was confirmed through different experiments. The presented results assess the interest of decision-level fusion strategies, such as the product of experts. Accurate crop map products were obtained over different countries in the early season with limited training data. The results highlight the benefit of fusion for early crop mapping and the interest of detecting cropland areas before the identification of crop types.

The impact of conflict-driven cropland abandonment on food insecurity in South Sudan revealed using satellite remote sensing

Armed conflicts often hinder food security through cropland abandonment and restrict the collection of on-the-ground information required for targeted relief distribution. Satellite remote sensing provides a means for gathering information about disruptions during armed conflicts and assessing the food security status in conflict zones. Using ~7,500 multisource satellite images, we implemented a data-driven approach that showed a reduction in cultivated croplands in war-ravaged South Sudan by 16% from 2016 to 2018. Propensity score matching revealed a statistical relationship between cropland abandonment and armed conflicts that contributed to drastic decreases in food supply. Our analysis shows that the abandoned croplands could have supported at least a quarter of the population in the southern states of South Sudan and demonstrates that remote sensing can play a crucial role in the assessment of cropland abandonment in food-insecure regions, thereby improving the basis for timely aid provision.

Crop Type Mapping from Optical and Radar Time Series Using Attention-Based Deep Learning

Crop maps are key inputs for crop inventory production and yield estimation and can inform the implementation of effective farm management practices. Producing these maps at detailed scales requires exhaustive field surveys that can be laborious, time-consuming, and expensive to replicate. With a growing archive of remote sensing data, there are enormous opportunities to exploit dense satellite image time series (SITS), temporal sequences of images over the same area. Generally, crop type mapping relies on single-sensor inputs and is solved with the help of traditional learning algorithms such as random forests or support vector machines. Nowadays, deep learning techniques have brought significant improvements by leveraging information in both spatial and temporal dimensions, which are relevant in crop studies. The concurrent availability of Sentinel-1 (synthetic aperture radar) and Sentinel-2 (optical) data offers a great opportunity to utilize them jointly; however, optimizing their synergy has been understudied with deep learning techniques. In this work, we analyze and compare three fusion strategies (input, layer, and decision levels) to identify the best strategy that optimizes optical-radar classification performance. They are applied to a recent architecture, notably, the pixel-set encoder–temporal attention encoder (PSE-TAE) developed specifically for object-based classification of SITS and based on self-attention mechanisms. Experiments are carried out in Brittany, in the northwest of France, with Sentinel-1 and Sentinel-2 time series. Input and layer-level fusion competitively achieved the best overall F-score surpassing decision-level fusion by 2%. On a per-class basis, decision-level fusion increased the accuracy of dominant classes, whereas layer-level fusion improves up to 13% for minority classes. Against single-sensor baseline, multi-sensor fusion strategies identified crop types more accurately: for example, input-level outperformed Sentinel-2 and Sentinel-1 by 3% and 9% in F-score, respectively. We have also conducted experiments that showed the importance of fusion for early time series classification and under high cloud cover condition.

Deep/Transfer Learning with Feature Space Ensemble Networks (FeatSpaceEnsNets) and Average Ensemble Networks (AvgEnsNets) for Change Detection Using DInSAR Sentinel-1 and Optical Sentinel-2 Satellite Data Fusion

Differential interferometric synthetic aperture radar (DInSAR), coherence, phase, and displacement are derived from processing SAR images to monitor geological phenomena and urban change. Previously, Sentinel-1 SAR data combined with Sentinel-2 optical imagery has improved classification accuracy in various domains. However, the fusing of Sentinel-1 DInSAR processed imagery with Sentinel-2 optical imagery has not been thoroughly investigated. Thus, we explored this fusion in urban change detection by creating a verified balanced binary classification dataset comprising 1440 blobs. Machine learning models using feature descriptors and non-deep learning classifiers, including a two-layer convolutional neural network (ConvNet2), were used as baselines. Transfer learning by feature extraction (TLFE) using various pre-trained models, deep learning from random initialization, and transfer learning by fine-tuning (TLFT) were all evaluated. We introduce a feature space ensemble family (FeatSpaceEnsNet), an average ensemble family (AvgEnsNet), and a hybrid ensemble family (HybridEnsNet) of TLFE neural networks. The FeatSpaceEnsNets combine TLFE features directly in the feature space using logistic regression. AvgEnsNets combine TLFEs at the decision level by aggregation. HybridEnsNets are a combination of FeatSpaceEnsNets and AvgEnsNets. Several FeatSpaceEnsNets, AvgEnsNets, and HybridEnsNets, comprising a heterogeneous mixture of different depth and architecture models, are defined and evaluated. We show that, in general, TLFE outperforms both TLFT and classic deep learning for the small dataset used and that larger ensembles of TLFE models do not always improve accuracy. The best performing ensemble is an AvgEnsNet (84.862%) comprised of a ResNet50, ResNeXt50, and EfficientNet B4. This was matched by a similarly composed FeatSpaceEnsNet with an F1 score of 0.001 and variance of 0.266 less. The best performing HybridEnsNet had an accuracy of 84.775%. All of the ensembles evaluated outperform the best performing single model, ResNet50 with TLFE (83.751%), except for AvgEnsNet 3, AvgEnsNet 6, and FeatSpaceEnsNet 5. Five of the seven similarly composed FeatSpaceEnsNets outperform the corresponding AvgEnsNet.

Integration of Sentinel 1 and Sentinel 2 Satellite Images for Crop Mapping

Crop identification is key to global food security. Due to the large scale of crop estimation, the science of remote sensing was able to do well in this field. The purpose of this study is to study the shortcomings and strengths of combined radar data and optical images to identify the type of crops in Tarom region (Iran). For this purpose, Sentinel 1 and Sentinel 2 images were used to create a map in the study area. The Sentinel 1 data came from Google Earth Engine’s (GEE) Level-1 Ground Range Detected (GRD) Interferometric Wide Swath (IW) product. Sentinel 1 radar observations were projected onto a standard 10-m grid in GRD output. The Sen2Cor method was used to mask for clouds and cloud shadows, and the Sentinel 2 Level-1C data was sourced from the Copernicus Open Access Hub. To estimate the purpose of classification, stochastic forest classification method was used to predict classification accuracy. Using seven types of crops, the classification map of the 2020 growth season in Tarom was prepared using 10-day Sentinel 2 smooth mosaic NDVI and 12-day Sentinel 1 back mosaic. Kappa coefficient of 0.75 and a maximum accuracy of 85% were reported in this study. To achieve maximum classification accuracy, it is recommended to use a combination of radar and optical data, as this combination increases the chances of examining the details compared to the single-sensor classification method and achieves more reliable information.

Sentinel-2 Recognition of Uncovered and Plastic Covered Agricultural Soil

Medium resolution satellite data, such as Sentinel-2 of the Copernicus programme, offer great new opportunities for the agricultural sector, and provide insights on soil surface characteristics and their management. Soil monitoring requires a high-quality dataset of uncovered and plastic covered agricultural soil. We developed a methodology to identify uncovered soil pixels in agricultural parcels during seedbed preparation and considered the impacts of clouds and shadows, vegetation cover, and artificial covers, such as those of greenhouses and plastic mulch films. We preserved the spatial and temporal integrity of parcels in the process and analysed spectral anomalies and their sources. The approach is based on freely available tools, namely Google Earth Engine and R Programming packages. We tested the methodology on the northern region of Belgium, which is characterised by small, fragmented parcels. We selected a period between mid-April to end-May, when active agricultural management practices leave the soil bare in preparation for the main cropping season. The spectral angle mapper was used to identify soil covered by non-plastic greenhouses or temporary soil covers, such as plastic mulch films. The effect of underlying soil on temporary covers was considered. The retrogressive plastic greenhouse index was used for detecting plastic greenhouses. The result was a high quality dataset of potential bare uncovered agricultural soil that allows further soil surface characterisation. This offered an improved understanding of the use of artificial covers, their spatial distribution, and their corresponding crops during the considered period. Artificial covers occurred most frequently in maize parcels. The approach resulted in precision values exceeding 0.9 for the detection of temporary covers and non-plastic greenhouses and a sensitivity value exceeding 0.95 for non-plastic and plastic greenhouses.

Land Use Land Cover Classification with U-Net: Advantages of Combining Sentinel-1 and Sentinel-2 Imagery

The U-net is nowadays among the most popular deep learning algorithms for land use/land cover (LULC) mapping; nevertheless, it has rarely been used with synthetic aperture radar (SAR) and multispectral (MS) imagery. On the other hand, the discrimination between plantations and forests in LULC maps has been emphasized, especially for tropical areas, due to their differences in biodiversity and ecosystem services provision. In this study, we trained a U-net using different imagery inputs from Sentinel-1 and Sentinel-2 satellites, MS, SAR and a combination of both (MS + SAR); while a random forests algorithm (RF) with the MS + SAR input was also trained to evaluate the difference in algorithm selection. The classification system included ten classes, including old-growth and secondary forests, as well as old-growth and young plantations. The most accurate results were obtained with the MS + SAR U-net, where the highest overall accuracy (0.76) and average F1-score (0.58) were achieved. Although MS + SAR and MS U-nets gave similar results for almost all of the classes, for old-growth plantations and secondary forest, the addition of the SAR band caused an F1-score increment of 0.08–0.11 (0.62 vs. 0.54 and 0.45 vs. 0.34, respectively). Consecutively, in comparison with the MS + SAR RF, the MS + SAR U-net obtained higher F1-scores for almost all the classes. Our results show that using the U-net with a combined input of SAR and MS images enabled a higher F1-score and accuracy for a detailed LULC map, in comparison with other evaluated methods.

Impervious Surfaces Mapping at City Scale by Fusion of Radar and Optical Data through a Random Forest Classifier

Urbanization increases the amount of impervious surfaces, making accurate information on spatial and temporal expansion trends essential; the challenge is to develop a cost- and labor-effective technique that is compatible with the assessment of multiple geographical locations in developing countries. Several studies have identified the potential of remote sensing and multiple source information in impervious surface quantification. Therefore, this study aims to fuse datasets from the Sentinel 1 and 2 Satellites to map the impervious surfaces of nine Pakistani cities and estimate their growth rates from 2016 to 2020 utilizing the random forest algorithm. All bands in the optical and radar images were resampled to 10 m resolution, projected to same coordinate system and geometrically aligned to stack into a single product. The models were then trained, and classifications were validated with land cover samples from Google Earth’s high-resolution images. Overall accuracies of classified maps ranged from 85% to 98% with the resultant quantities showing a strong linear relationship (R-squared value of 0.998) with the Copernicus Global Land Services data. There was up to 9% increase in accuracy and up to 12 % increase in kappa coefficient from the fused data with respect to optical alone. A McNemar test confirmed the superiority of fused data. Finally, the cities had growth rates ranging from 0.5% to 2.5%, with an average of 1.8%. The information obtained can alert urban planners and environmentalists to assess impervious surface impacts in the cities.

Sentinel-1 Time Series for Crop Identification in the Framework of the Future CAP Monitoring

In this upcoming Common Agricultural Policy (CAP) reform, the use of satellite imagery is taking an increasing role for improving the Integrated Administration and Control System (IACS). Considering the operational aspect of the CAP monitoring process, the use of Sentinel-1 SAR (Synthetic Aperture Radar) images is highly relevant, especially in regions with a frequent cloud cover, such as Belgium. Indeed, SAR imagery does not depend on sunlight and is barely affected by the presence of clouds. Moreover, the SAR signal is particularly sensitive to the geometry and the water content of the target. Crop identification is often a pre-requisite to monitor agriculture at parcel level (ploughing, harvest, grassland mowing, intercropping, etc.) The main goal of this study is to assess the performances and constraints of a SAR-based crop classification in an operational large-scale application. The Random Forest object-oriented classification model is built on Sentinel-1 time series from January to August 2020 only. It can identify crops in the Walloon Region (south part of Belgium) with high performance: 93.4% of well-classified area, representing 88.4% of the parcels. Among the 48 crop groups, the six most represented ones get a F1-score higher or equal to 84%. Additionally, this research documents how the classification performance is affected by different parameters: the SAR orbit, the size of the training dataset, the use of different internal buffers on parcel polygons before signal extraction, the set of explanatory variables, and the period of the time series. In an operational context, this allows to choose the right balance between classification accuracy and model complexity. A key result is that using a training dataset containing only 3.2% of the total number of parcels allows to correctly classify 91.7% of the agricultural area. The impact of rain and snow is also discussed. Finally, this research analyses how the classification accuracy depends on some characteristics of the parcels like their shape or size. This allows to assess the relevance of the classification depending on those characteristics, as well as to identify a subset of parcels for which the global accuracy is higher.

Spatial Pattern Analysis of the Ecosystem Services in the Guangdong-Hong Kong-Macao Greater Bay Area Using Sentinel-1 and Sentinel-2 Imagery Based on Deep Learning Method

Assessment of ecosystem services supply, demand, and budgets can help to achieve sustainable urban development. The Guangdong-Hong Kong-Macao Greater Bay Area, as one of the most developed megacities in China, sets up a goal of high-quality development while fostering ecosystem services. Therefore, assessing the ecosystem services in this study area is very important to guide further development. However, the spatial pattern of ecosystem services, especially at local scales, is not well understood. Using the available 2017 land cover product, Sentinel-1 SAR and Sentinel-2 optical images, a deep learning land cover mapping framework integrating deep change vector analysis and the ResUnet model was proposed. Based on the produced 10 m land cover map for the year 2020, recent spatial patterns of the ecosystem services at different scales (i.e., the GBA, 11 cities, urban–rural gradient, and pixel) were analyzed. The results showed that: (1) Forest was the primary land cover in Guangzhou, Huizhou, Shenzhen, Zhuhai, Jiangmen, Zhaoqing, and Hong Kong, and an impervious surface was the main land cover in the other four cities. (2) Although ecosystem services in the GBA were sufficient to meet their demand, there was undersupply for all the three general services in Macao and for the provision services in Zhongshan, Dongguan, Shenzhen, and Foshan. (3) Along the urban–rural gradient in the GBA, supply and demand capacity showed an increasing and decreasing trend, respectively. As for the city-level analysis, Huizhou and Zhuhai showed a fluctuation pattern while Jiangmen, Zhaoqing, and Hong Kong presented a decreasing pattern along the gradient. (4) Inclusion of neighborhood landscape led to increased demand scores in a small proportion of impervious areas and oversupply for a very large percent of bare land.

Remote Sensing-Based Analysis of Urban Landscape Change in the City of Bucharest, Romania

The paper investigates the urban landscape changes for the last 50 years in Bucharest, the capital city of Romania. Bucharest shows a complex structural transformation driven by the socialist urban policy, followed by an intensive real-estate market development. Our analysis is based on a diachronic set of high-resolution satellite imagery: declassified CORONA KH-4B from 1968, SPOT-1 from 1989, and multisensor stacked layers from Sentinel-1 SAR together with Sentinel-2MSI from 2018. Three different datasets of land cover/use are extracted for the reference years. Each dataset reveals its own urban structure pattern. The first one illustrates a radiography of the city in the second part of the 20th century, where rural patterns meet the modern ones, while the second one reveals the frame of a city in a full process of transformation with multiple constructions sites, based on the socialist model. The third one presents an image of a cosmopolitan city during an expansion process, with a high degree of landscape heterogeneity. All the datasets are included in a built-up change analysis in order to map and assess the spatial transformations of the city pattern over 5 decades. In order to quantify and map the changes, the Built-up Change Index (BCI) is introduced. The results highlight a particular situation linked to the policy development visions for each decade, with major changes of about 50% for different built-up classes. The GIS analysis illustrates two major landscape transformations: from the old semirural structures with houses surrounded by gardens from 1968, to a compact pattern with large districts of blocks of flats in 1989, and a contemporary city defined by an uncontrolled urban sprawl process in 2018.

Sentinel-1 Backscatter Analysis and Radiative Transfer Modeling of Dense Winter Wheat Time Series

This study evaluates a temporally dense VV-polarized Sentinel-1 C-band backscatter time series (revisit time of 1.5 days) for wheat fields near Munich (Germany). A dense time series consisting of images from different orbits (varying acquisition) is analyzed, and Radiative Transfer (RT)-based model combinations are adapted and evaluated with the use of radar backscatter. The model shortcomings are related to scattering mechanism changes throughout the growth period with the use of polarimetric decomposition. Furthermore, changes in the RT modeled backscatter results with spatial aggregation from the pixel to field scales are quantified and related to the sensitivity of the RT models, and their soil moisture output are quantified and related to changes in backscatter. Therefore, various (sub)sets of the dense Sentinel-1 time series are analyzed to relate and quantify the impact of the abovementioned points on the modeling results. The results indicate that the incidence angle is the main driver for backscatter differences between consecutive acquisitions with various recording scenarios. The influence of changing azimuth angles was found to be negligible. Further analyses of polarimetric entropy and scattering alpha angle using a dual polarimetric eigen-based decomposition show that scattering mechanisms change over time. The patterns analyzed in the entropy-alpha space indicate that scattering mechanism changes are mainly driven by the incidence angle and not by the azimuth angle. Besides the analysis of differences within the Sentinel-1 data, we analyze the capability of RT model approaches to capture the observed Sentinel-1 backscatter changes due to various acquisition geometries. For this, the surface models “Oh92” or “IEM_B” (Baghdadi’s version of the Integral Equation Method) are coupled with the canopy model “SSRT” (Single Scattering Radiative Transfer). To resolve the shortcomings of the RT model setup in handling varying incidence angles and therefore the backscatter changes observed between consecutive time steps of a dense winter wheat time series, an empirical calibration parameter (coef) influencing the transmissivity (T) is introduced. The results show that shortcomings of simplified RT model architectures caused by handling time series consisting of images with varied incidence angles can be at least partially compensated by including a calibration coefficient to parameterize the modeled transmissivity for the varying incidence angle scenarios individually.

Sentinel-1 and 2 Time-Series for Vegetation Mapping Using Random Forest Classification: A Case Study of Northern Croatia

Land-cover (LC) mapping in a morphologically heterogeneous landscape area is a challenging task since various LC classes (e.g., crop types in agricultural areas) are spectrally similar. Most research is still mostly relying on optical satellite imagery for these tasks, whereas synthetic aperture radar (SAR) imagery is often neglected. Therefore, this research assessed the classification accuracy using the recent Sentinel-1 (S1) SAR and Sentinel-2 (S2) time-series data for LC mapping, especially vegetation classes. Additionally, ancillary data, such as texture features, spectral indices from S1 and S2, respectively, as well as digital elevation model (DEM), were used in different classification scenarios. Random Forest (RF) was used for classification tasks using a proposed hybrid reference dataset derived from European Land Use and Coverage Area Frame Survey (LUCAS), CORINE, and Land Parcel Identification Systems (LPIS) LC database. Based on the RF variable selection using Mean Decrease Accuracy (MDA), the combination of S1 and S2 data yielded the highest overall accuracy (OA) of 91.78%, with a total disagreement of 8.22%. The most pertinent features for vegetation mapping were GLCM Mean and Variance for S1, NDVI, along with Red and SWIR band for S2, whereas the digital elevation model produced major classification enhancement as an input feature. The results of this study demonstrated that the aforementioned approach (i.e., RF using a hybrid reference dataset) is well-suited for vegetation mapping using Sentinel imagery, which can be applied for large-scale LC classifications.

Using Sentinel-1, Sentinel-2, and Planet Imagery to Map Crop Type of Smallholder Farms

Remote sensing offers a way to map crop types across large spatio-temporal scales at low costs. However, mapping crop types is challenging in heterogeneous, smallholder farming systems, such as those in India, where field sizes are often smaller than the resolution of historically available imagery. In this study, we examined the potential of relatively new, high-resolution imagery (Sentinel-1, Sentinel-2, and PlanetScope) to identify four major crop types (maize, mustard, tobacco, and wheat) in eastern India using support vector machine (SVM). We found that a trained SVM model that included all three sensors led to the highest classification accuracy (85%), and the inclusion of Planet data was particularly helpful for classifying crop types for the smallest farms (<600 m2). This was likely because its higher spatial resolution (3 m) could better account for field-level variations in smallholder systems. We also examined the impact of image timing on the classification accuracy, and we found that early-season images did little to improve our models. Overall, we found that readily available Sentinel-1, Sentinel-2, and Planet imagery were able to map crop types at the field-scale with high accuracy in Indian smallholder systems. The findings from this study have important implications for the identification of the most effective ways to map crop types in smallholder systems.

Detection and Quantification of Irrigation Water Amounts at 500 m Using Sentinel-1 Surface Soil Moisture

Detailed information about irrigation timing and water use at a high spatial resolution is critical for monitoring and improving agricultural water use efficiency. However, neither statistical surveys nor remote sensing-based approaches can currently accommodate this need. To address this gap, we propose a novel approach based on the TU Wien Sentinel-1 Surface Soil Moisture product, characterized by a spatial sampling of 500 m and a revisit time of 1.5–4 days over Europe. Spatiotemporal patterns of soil moisture are used to identify individual irrigation events and estimate irrigation water amounts. To retrieve the latter, we include formulations of evapotranspiration and drainage losses to account for vertical fluxes, which may significantly influence sub-daily soil moisture variations. The proposed approach was evaluated against field-scale irrigation data reported by farmers at three sites in Germany with heterogeneous field sizes, crop patterns, irrigation systems and management. Our results show that most field-scale irrigation events can be detected using soil moisture information (mean F-score = 0.77). Irrigation estimates, in terms of temporal dynamics as well as spatial patterns, were in agreement with reference data (mean Pearson correlation = 0.64) regardless of field-specific characteristics (e.g., crop type). Hence, the proposed approach has the potential to be applied over large regions with varying cropping systems.

Delineating Smallholder Maize Farms from Sentinel-1 Coupled with Sentinel-2 Data Using Machine Learning

Rural communities rely on smallholder maize farms for subsistence agriculture, the main driver of local economic activity and food security. However, their planted area estimates are unknown in most developing countries. This study explores the use of Sentinel-1 and Sentinel-2 data to map smallholder maize farms. The random forest (RF), support vector (SVM) machine learning algorithms and model stacking (ST) were applied. Results show that the classification of combined Sentinel-1 and Sentinel-2 data improved the RF, SVM and ST algorithms by 24.2%, 8.7%, and 9.1%, respectively, compared to the classification of Sentinel-1 data individually. Similarities in the estimated areas (7001.35 ± 1.2 ha for RF, 7926.03 ± 0.7 ha for SVM and 7099.59 ± 0.8 ha for ST) show that machine learning can estimate smallholder maize areas with high accuracies. The study concludes that the single-date Sentinel-1 data were insufficient to map smallholder maize farms. However, single-date Sentinel-1 combined with Sentinel-2 data were sufficient in mapping smallholder farms. These results can be used to support the generation and validation of national crop statistics, thus contributing to food security.

Machine Learning-Based Classification for Crop-Type Mapping Using the Fusion of High-Resolution Satellite Imagery in a Semiarid Area

The monitoring of cultivated crops and the types of different land covers is a relevant environmental and economic issue for agricultural lands management and crop yield prediction. In this context, this paper aims to use and evaluate the contribution of multisensors classification based on machine learning classifiers to crop-type identification in a semiarid area of Morocco. It is a very heterogeneous zone characterized by mixed crops (tree crops with annual crops, same crop with different phenological states during the same agricultural season, crop rotation, etc.). Therefore, such heterogeneity made the crop-type discrimination more complicated. To overcome these challenges, the present work is the first study in this area which used the fusion of high spatiotemporal resolution Sentinel-1 and Sentinel-2 satellite images for land use and land cover mapping. Three machine learning classifier algorithms, artificial neural network (ANN), support vector machine (SVM), and maximum likelihood (ML), were applied to identify and map crop types in irrigated perimeter. In situ observations of the year 2018, for the R3 perimeter of Haouz plain in central Morocco, were used with satellite data of the same year to perform this work. The results showed that combined images acquired in C-band and the optical range improved clearly the crop-type classification performance (overall accuracy = 89%; Kappa = 0.85) compared to the classification results of optical or SAR data alone.

National Crop Mapping Using Sentinel-1 Time Series: A Knowledge-Based Descriptive Algorithm

National-level mapping of crop types is important to monitor food security, understand environmental conditions, inform optimal use of the landscape, and contribute to agricultural policy. Countries or economic regions currently and increasingly use satellite sensor data for classifying crops over large areas. However, most methods have been based on machine learning algorithms, with these often requiring large training datasets that are not always available and may be costly to produce or collect. Focusing on Wales (United Kingdom), the research demonstrates how the knowledge that the agricultural community has gathered together over past decades can be used to develop algorithms for mapping different crop types. Specifically, we aimed to develop an alternative method for consistent and accurate crop type mapping where cloud cover is quite persistent and without the need for extensive in situ/ground datasets. The classification approach is parcel-based and informed by concomitant analysis of knowledge-based crop growth stages and Sentinel-1 C-band SAR time series. For 2018, crop type classifications were generated nationally for Wales, with regional overall accuracies ranging between 85.8% and 90.6%. The method was particularly successful in distinguishing barley from wheat, which is a major source of error in other crop products available for Wales. This study demonstrates that crops can be accurately identified and mapped across a large area (i.e., Wales) using Sentinel-1 C-band data and by capitalizing on knowledge of crop growth stages. The developed algorithm is flexible and, compared to the other methods that allow crop mapping in Wales, the approach provided more consistent discrimination and lower variability in accuracies between classes and regions.

Improving the Accuracy of Multiple Algorithms for Crop Classification by Integrating Sentinel-1 Observations with Sentinel-2 Data

The availability of an unprecedented amount of open remote sensing data, such as Sentinel-1 and -2 data within the Copernicus program, has boosted the idea of combining the use of optical and radar data to improve the accuracy of agricultural applications such as crop classification. Sentinel-1’s Synthetic Aperture Radar (SAR) provides co- and cross-polarized backscatter, which offers the opportunity to monitor agricultural crops using radar at high spatial and temporal resolution. In this study, we assessed the potential of integrating Sentinel-1 information (VV and VH backscatter and their ratio VH/VV with Sentinel-2A data (NDVI) to perform crop classification and to define which are the most important input data that provide the most accurate classification results. Further, we examined the temporal dynamics of remote sensing data for cereal, horticultural, and industrial crops, perennials, deciduous trees, and legumes. To select the best SAR input feature, we tried two approaches, one based on classification with only SAR features and one based on integrating SAR with optical data. In total, nine scenarios were tested. Furthermore, we evaluated the performance of 22 nonparametric classifiers on which most of these algorithms had not been tested before with SAR data. The results revealed that the best performing scenario was the one integrating VH and VV with normalized difference vegetation index (NDVI) and cubic support vector machine (SVM) (the kernel function of the classifier is cubic) as the classifier with the highest accuracy among all those tested.

Application of Google Earth Engine Cloud Computing Platform, Sentinel Imagery, and Neural Networks for Crop Mapping in Canada

The ability of the Canadian agriculture sector to make better decisions and manage its operations more competitively in the long term is only as good as the information available to inform decision-making. At all levels of Government, a reliable flow of information between scientists, practitioners, policy-makers, and commodity groups is critical for developing and supporting agricultural policies and programs. Given the vastness and complexity of Canada’s agricultural regions, space-based remote sensing is one of the most reliable approaches to get detailed information describing the evolving state of the country’s environment. Agriculture and Agri-Food Canada (AAFC)—the Canadian federal department responsible for agriculture—produces the Annual Space-Based Crop Inventory (ACI) maps for Canada. These maps are valuable operational space-based remote sensing products which cover the agricultural land use and non-agricultural land cover found within Canada’s agricultural extent. Developing and implementing novel methods for improving these products are an ongoing priority of AAFC. Consequently, it is beneficial to implement advanced machine learning and big data processing methods along with open-access satellite imagery to effectively produce accurate ACI maps. In this study, for the first time, the Google Earth Engine (GEE) cloud computing platform was used along with an Artificial Neural Networks (ANN) algorithm and Sentinel-1, -2 images to produce an object-based ACI map for 2018. Furthermore, different limitations of the proposed method were discussed, and several suggestions were provided for future studies. The Overall Accuracy (OA) and Kappa Coefficient (KC) of the final 2018 ACI map using the proposed GEE cloud method were 77% and 0.74, respectively. Moreover, the average Producer Accuracy (PA) and User Accuracy (UA) for the 17 cropland classes were 79% and 77%, respectively. Although these levels of accuracies were slightly lower than those of the AAFC’s ACI map, this study demonstrated that the proposed cloud computing method should be investigated further because it was more efficient in terms of cost, time, computation, and automation.

Radar Satellite Image Time Series Analysis for High-Resolution Mapping of Man-Made Forest Change in Chongming Eco-Island

Forests are important oxygen sources and carbon sinks. They provide a series of ecosystem services that are crucial to eco-island sustainability, such as the protection of soils, conservation of biodiversity, and development of the eco-tourism, etc. Interestingly, Chongming eco-island is located in the borderlands between fresh- and saltwater environments, where the Yangtze River meets the East China Sea. Most forests in Chongming island are therefore man-made and very vulnerable to the ecological environment mixing of freshwater streams and rivers with salty ocean tides, and are affected by climate and human activity. Estimating and monitoring forest change within an estuary is required for the sustainable management of forest resources and forest-based eco-tourism. Most optical satellites are unsuitable for continuous forest mapping due to cloud cover and their relatively low spatial and temporal resolution. Here, using Sentinel-1 satellite carrying an imaging C-band synthetic aperture radar, radar vegetation index was derived to detect the forest dynamics on the island. Furthermore we quantified forest area change in three well known and the most strictly protected and representative areas, namely Dongping National Forest Park, Dongtan National Wetland Park, and Xisha National Wetland Park, in the Chongming eco-island over the past five years at 10-metre resolution. We recorded the early and mid summers when the forest canopies grew to the peak in the study areas. The planted forest in Dongping National Forest Park grew an area of 7.38 hectares from 2015 to 2019, and disappeared from an area of almost 2.59 hectares in 2018. The man-made forest of Xisha National Wetland Park increased at an area of almost 20.19 hectares over the past five years. The forest in Dongtan National Wetland Park increased to an area of almost 2.12 hectares in the period of 2015–2017 and 2018–2019. However, from 2017 to 2018, the area of planted forests in Dongtan National Wetland Park decreased by 1.35 hectares. This study shows man-made forest change can be measured and that evidence can be collected to show how the forest is altered by human activities, and informs forest management decision-making for Chongming eco-island.

Lowland Rice Mapping in Sédhiou Region (Senegal) Using Sentinel 1 and Sentinel 2 Data and Random Forest

In developing countries, information on the area and spatial distribution of paddy rice fields is an essential requirement for ensuring food security and facilitating targeted actions of both technical assistance and restoration of degraded production areas. In this study, Sentinel 1 (S1) and Sentinel 2 (S2) imagery was used to map lowland rice crop areas in the Sédhiou region (Senegal) for the 2017, 2018, and 2019 growing seasons using the Random Forest (RF) algorithm. Ground sample datasets were annually collected (416, 455, and 400 samples) for training and testing yearly RF classification. A procedure was preliminarily applied to process S2 scenes and yield a normalized difference vegetation index (NDVI) time series less affected by clouds. A total of 93 predictors were calculated from S2 NDVI time series and S1 vertical transmit–horizontal receive (VH) and vertical transmit–vertical receive (VV) backscatters. Guided regularized random forest (GRRF) was used to deal with the arising multicollinearity and identify the most important predictors. The RF classifier was then applied to the selected predictors. The algorithm predicted the five land cover types present in the test areas, with a maximum accuracy of 87% and kappa coefficient of 0.8 in 2019. The broad land cover maps identified around 12,500 (2017), 13,800 (2018), and 12,800 (2019) ha of lowland rice crops. The study highlighted a partial difficulty of the classifier to distinguish rice from natural herbaceous vegetation (NHV) due to similar temporal patterns and high intra-class variability. Moreover, the results of this investigation indicated that S2-derived predictors provided more valuable information compared to VV and VH backscatter-derived predictors, but a combination of radar and optical imagery always outperformed a classification based on single-sensor inputs. An example is finally provided that illustrates how the maps obtained can be combined with ground observations through a ratio estimator in order to yield a statistically sound prediction of rice area all over the study region.

National Scale Land Cover Classification for Ecosystem Services Mapping and Assessment, Using Multitemporal Copernicus EO Data and Google Earth Engine

Land-Use/Land-Cover (LULC) products are a common source of information and a key input for spatially explicit models of ecosystem service (ES) supply and demand. Global, continental, and regional, readily available, and free land-cover products generated through Earth Observation (EO) data, can be potentially used as relevant to ES mapping and assessment processes from regional to national scales. However, several limitations exist in these products, highlighting the need for timely land-cover extraction on demand, that could replace or complement existing products. This study focuses on the development of a classification workflow for fine-scale, object-based land cover mapping, employed on terrestrial ES mapping, within the Greek terrestrial territory. The processing was implemented in the Google Earth Engine cloud computing environment using 10 m spatial resolution Sentinel-1 and Sentinel-2 data. Furthermore, the relevance of different training data extraction strategies and temporal EO information for increasing the classification accuracy was also evaluated. The different classification schemes demonstrated differences in overall accuracy ranging from 0.88% to 4.94% with the most accurate classification scheme being the manual sampling/monthly feature classification achieving a 79.55% overall accuracy. The classification results suggest that existing LULC data must be cautiously considered for automated extraction of training samples, in the case of new supervised land cover classifications aiming also to discern complex vegetation classes. The code used in this study is available on GitHub and runs on the Google Earth Engine web platform.

Comparing Sentinel-1 and -2 Data and Indices for Agricultural Land Use Monitoring

Agricultural vegetation development and harvest date monitoring over large areas requires frequent remote sensing observations. In regions with persistent cloud coverage during the vegetation season this is only feasible with active systems, such as SAR, and is limited for optical data. To date, optical remote sensing vegetation indices are more frequently used to monitor agricultural vegetation status because they are easily processed, and the characteristics are widely known. This study evaluated the correlations of three Sentinel-2 optical indices with Sentinel-1 SAR indices over agricultural areas to gain knowledge about their relationship. We compared Sentinel-2 Normalized Difference Vegetation Index, Normalized Difference Water Index, and Plant Senescence Radiation Index with Sentinel-1 SAR VV and VH backscatter, VH/VV ratio, and Sentinel-1 Radar Vegetation Index. The study was conducted on 22 test sites covering approximately 35,000 ha of four different main European agricultural land use types, namely grassland, maize, spring barley, and winter wheat, in Lower Saxony, Germany, in 2018. We investigated the relationship between Sentinel-1 and Sentinel-2 indices for each land use type considering three phenophases (growing, green, senescence). The strength of the correlations of optical and SAR indices differed among land use type and phenophase. There was no generic correlation between optical and SAR indices in our study. However, when the data were split by land use types and phenophases, the correlations increased remarkably. Overall, the highest correlations were found for the Radar Vegetation Index and VH backscatter. Correlations for grassland were lower than for the other land use types. Adding auxiliary data to a multiple linear regression analysis revealed that, in addition to land use type and phenophase information, the lower quartile and median SAR values per field, and a spatial variable, improved the models. Other auxiliary data retrieved from a digital elevation model, Sentinel-1 orbit direction, soil type information, and other SAR values had minor impacts on the model performance. In conclusion, despite the different nature of the signal generation, there were distinct relationships between optical and SAR indices which were independent of environmental variables but could be stratified by land use type and phenophase. These relationships showed similar patterns across different test sites. However, a regional clustering of landscapes would significantly improve the relationships.

Object-Based Multi-Temporal and Multi-Source Land Cover Mapping Leveraging Hierarchical Class Relationships

European satellite missions Sentinel-1 (S1) and Sentinel-2 (S2) provide at high spatial resolution and high revisit time, respectively, radar and optical images that support a wide range of Earth surface monitoring tasks, such as Land Use/Land Cover mapping. A long-standing challenge in the remote sensing community is about how to efficiently exploit multiple sources of information and leverage their complementarity, in order to obtain the most out of radar and optical data. In this work, we propose to deal with land cover mapping in an object-based image analysis (OBIA) setting via a deep learning framework designed to leverage the multi-source complementarity provided by radar and optical satellite image time series (SITS). The proposed architecture is based on an extension of Recurrent Neural Network (RNN) enriched via a modified attention mechanism capable to fit the specificity of SITS data. Our framework also integrates a pretraining strategy that allows to exploit specific domain knowledge, shaped as hierarchy over the set of land cover classes, to guide the model training. Thorough experimental evaluations, involving several competitive approaches were conducted on two study sites, namely the Reunion island and a part of the Senegalese groundnut basin. Classification results, 79% of global accuracy on the Reunion island and 90% on the Senegalese site, respectively, have demonstrated the suitability of the proposal.

Crop Type Classification Using Fusion of Sentinel-1 and Sentinel-2 Data: Assessing the Impact of Feature Selection, Optical Data Availability, and Parcel Sizes on the Accuracies

Crop type classification using Earth Observation (EO) data is challenging, particularly for crop types with similar phenological growth stages. In this regard, the synergy of optical and Synthetic-Aperture Radar (SAR) data enables a broad representation of biophysical and structural information on target objects, enhancing crop type mapping. However, the fusion of multi-sensor dense time-series data often comes with the challenge of high dimensional feature space. In this study, we (1) evaluate how the usage of only optical, only SAR, and their fusion affect the classification accuracy; (2) identify the combination of which time-steps and feature-sets lead to peak accuracy; (3) analyze misclassifications based on the parcel size, optical data availability, and crops’ temporal profiles. Two fusion approaches were considered and compared in this study: feature stacking and decision fusion. To distinguish the most relevant feature subsets time- and variable-wise, grouped forward feature selection (gFFS) was used. gFFS allows focusing analysis and interpretation on feature sets of interest like spectral bands, vegetation indices (VIs), or data sensing time rather than on single features. This feature selection strategy leads to better interpretability of results while substantially reducing computational expenses. The results showed that, in contrast to most other studies, SAR datasets outperform optical datasets. Similar to most other studies, the optical-SAR combination outperformed single sensor predictions. No significant difference was recorded between feature stacking and decision fusion. Random Forest (RF) appears to be robust to high feature space dimensionality. The feature selection did not improve the accuracies even for the optical-SAR feature stack with 320 features. Nevertheless, the combination of RF feature importance and time- and variable-wise gFFS rankings in one visualization enhances interpretability and understanding of the features’ relevance for specific classification tasks. For example, by enabling the identification of features that have high RF feature importance values but are, in their information content, correlated with other features. This study contributes to the growing domain of interpretable machine learning.

Deriving Wheat Crop Productivity Indicators Using Sentinel-1 Time Series

High-frequency Earth observation (EO) data have been shown to be effective in identifying crops and monitoring their development. The purpose of this paper is to derive quantitative indicators of crop productivity using synthetic aperture radar (SAR). This study shows that the field-specific SAR time series can be used to characterise growth and maturation periods and to estimate the performance of cereals. Winter wheat fields on the Rothamsted Research farm in Harpenden (UK) were selected for the analysis during three cropping seasons (2017 to 2019). Average SAR backscatter from Sentinel-1 satellites was extracted for each field and temporal analysis was applied to the backscatter cross-polarisation ratio (VH/VV). The calculation of the different curve parameters during the growing period involves (i) fitting of two logistic curves to the dynamics of the SAR time series, which describe timing and intensity of growth and maturation, respectively; (ii) plotting the associated first and second derivative in order to assist the determination of key stages in the crop development; and (iii) exploring the correlation matrix for the derived indicators and their predictive power for yield. The results show that the day of the year of the maximum VH/VV value was negatively correlated with yield (r = −0.56), and the duration of “full” vegetation was positively correlated with yield (r = 0.61). Significant seasonal variation in the timing of peak vegetation (p = 0.042), the midpoint of growth (p = 0.037), the duration of the growing season (p = 0.039) and yield (p = 0.016) were observed and were consistent with observations of crop phenology. Further research is required to obtain a more detailed picture of the uncertainty of the presented novel methodology, as well as its validity across a wider range of agroecosystems.

Assessing Spatial Limits of Sentinel-2 Data on Arable Crops in the Context of Checks by Monitoring

The availability of large amounts of Sentinel-2 data has been a trigger for its increasing exploitation in various types of applications. It is, therefore, of importance to understand the limits above which these data still guarantee a meaningful outcome. This paper proposes a new method to quantify and specify restrictions of the Sentinel-2 imagery in the context of checks by monitoring, a newly introduced control approach within the European Common Agriculture Policy framework. The method consists of a comparison of normalized difference vegetation index (NDVI) time series constructed from data of different spatial resolution to estimate the performance and limits of the coarser one. Using similarity assessment of Sentinel-2 (10 m pixel size) and PlanetScope (3 m pixel size) NDVI time series, it was estimated that for 10% out of 867 fields less than 0.5 ha in size, Sentinel-2 data did not provide reliable evidence of the activity or state of the agriculture field over a given timeframe. Statistical analysis revealed that the number of clean or full pixels and the proportion of pixels lost after an application of a 5-m (1/2 pixel) negative buffer are the geospatial parameters of the field that have the highest influence on the ability of the Sentinel-2 data to qualify the field’s state in time. We specified the following limiting criteria: at least 8 full pixels inside a border and less than 60% of pixels lost. It was concluded that compliance with the criteria still assures a high level of extracted information reliability. Our research proved the promising potential, which was higher than anticipated, of Sentinel-2 data for the continuous state assessment of small fields. The method could be applied to other sensors and indicators.

Modelling Crop Biomass from Synthetic Remote Sensing Time Series: Example for the DEMMIN Test Site, Germany

This study compares the performance of the five widely used crop growth models (CGMs): World Food Studies (WOFOST), Coalition for Environmentally Responsible Economies (CERES)-Wheat, AquaCrop, cropping systems simulation model (CropSyst), and the semi-empiric light use efficiency approach (LUE) for the prediction of winter wheat biomass on the Durable Environmental Multidisciplinary Monitoring Information Network (DEMMIN) test site, Germany. The study focuses on the use of remote sensing (RS) data, acquired in 2015, in CGMs, as they offer spatial information on the actual conditions of the vegetation. Along with this, the study investigates the data fusion of Landsat (30 m) and Moderate Resolution Imaging Spectroradiometer (MODIS) (500 m) data using the spatial and temporal reflectance adaptive reflectance fusion model (STARFM) fusion algorithm. These synthetic RS data offer a 30-m spatial and one-day temporal resolution. The dataset therefore provides the necessary information to run CGMs and it is possible to examine the fine-scale spatial and temporal changes in crop phenology for specific fields, or sub sections of them, and to monitor crop growth daily, considering the impact of daily climate variability. The analysis includes a detailed comparison of the simulated and measured crop biomass. The modelled crop biomass using synthetic RS data is compared to the model outputs using the original MODIS time series as well. On comparison with the MODIS product, the study finds the performance of CGMs more reliable, precise, and significant with synthetic time series. Using synthetic RS data, the models AquaCrop and LUE, in contrast to other models, simulate the winter wheat biomass best, with an output of high R2 (>0.82), low RMSE (<600 g/m2) and significant p-value (<0.05) during the study period. However, inputting MODIS data makes the models underperform, with low R2 (<0.68) and high RMSE (>600 g/m2). The study shows that the models requiring fewer input parameters (AquaCrop and LUE) to simulate crop biomass are highly applicable and precise. At the same time, they are easier to implement than models, which need more input parameters (WOFOST and CERES-Wheat).

Assessing the Accuracy of Multiple Classification Algorithms for Crop Classification Using Landsat-8 and Sentinel-2 Data

The launch of Sentinel-2A and B satellites has boosted the development of many applications that could benefit from the fine resolution of the supplied information, both in time and in space. Crop classification is a necessary task for efficient land management. We evaluated the benefits of combining Landsat-8 and Sentinel-2A information for irrigated crop classification. We also assessed the robustness and efficiency of 22 nonparametric classification algorithms for classifying irrigated crops in a semiarid region in the southeast of Spain. A parcel-based approach was proposed calculating the mean normalized difference vegetation index (NDVI) of each plot and the standard deviation to generate a calibration-testing set of data. More than 2000 visited plots for 12 different crops along the study site were utilized as ground truth. Ensemble classifiers were the most robust algorithms but not the most efficient because of their low prediction rate. Nearest neighbor methods and support vector machines have the best balance between robustness and efficiency as methods for classification. Although the F1 score is close to 90%, some misclassifications were found for spring crops (e.g., barley, wheat and peas). However, crops with quite similar cycles could be differentiated, such as purple garlic and white garlic, showing the powerfulness of the developed tool.