Data Mining Practical Machine Learning Tools and Techniques

https://ris.utwente.nl/ws/files/29088356/Belgiu2017sentinel_2.pdf

Sentinel-2 cropland mapping using pixel-based and object-based time-weighted dynamic time warping analysis

Google Earth Engine (GEE) is a cloud-based geospatial processing platform for large-scale environmental monitoring and analysis. The free-to-use GEE platform provides access to (1) petabytes of publicly available remote sensing imagery and other ready-to-use products with an explorer web app; (2) high-speed parallel processing and machine learning algorithms using Google’s computational infrastructure; and (3) a library of Application Programming Interfaces (APIs) with development environments that support popular coding languages, such as JavaScript and Python. Together these core features enable users to discover, analyze and visualize geospatial big data in powerful ways without needing access to supercomputers or specialized coding expertise. The development of GEE has created much enthusiasm and engagement in the remote sensing and geospatial data science fields. Yet after a decade since GEE was launched, its impact on remote sensing and geospatial science has not been carefully explored. Thus, a systematic review of GEE that can provide readers with the “big picture” of the current status and general trends in GEE is needed. To this end, the decision was taken to perform a meta-analysis investigation of recent peer-reviewed GEE articles focusing on several features, including data, sensor type, study area, spatial resolution, application, strategy, and analytical methods. A total of 349 peer-reviewed articles published in 146 different journals between 2010 and October 2019 were reviewed. Publications and geographical distribution trends showed a broad spectrum of applications in environmental analyses at both regional and global scales. Remote sensing datasets were used in 90% of studies while 10% of the articles utilized ready-to-use products for analyses. Optical satellite imagery with medium spatial resolution, particularly Landsat data with an archive exceeding 40 years, has been used extensively. Linear regression and random forest were the most frequently used algorithms for satellite imagery processing. Among ready-to-use products, the normalized difference vegetation index (NDVI) was used in 27% of studies for vegetation, crop, land cover mapping and drought monitoring. The results of this study confirm that GEE has and continues to make substantive progress on global challenges involving process of geo-big data.

Google Earth Engine for geo-big data applications: A meta-analysis and systematic review

https://pure.mpg.de/pubman/item/item_3033361_3/component/file_3033369/1-s2.0-S0034425719300689-main.pdf

Historical background and current developments for mapping burned area from satellite Earth observation

Development of a Sentinel-2 burned area algorithm: Generation of a small fire database for sub-Saharan Africa

The aim of this paper is to investigate how optical and Synthetic Aperture Radar (SAR) data can be combined in an integrated multi-source framework to identify burned areas at the regional scale. The proposed approach is based on the use of fuzzy sets theory and a region-growing algorithm. Landsat TM and (C-band) ENVISAT Advanced Synthetic Aperture Radar (ASAR) images acquired for the year 2003 have been processed to extract burned area maps over Portugal. Pre-post fire SAR backscatter temporal difference has been integrated with optical spectral indices to the aim of reducing confusion between burned areas and low-albedo surfaces. The output fuzzy score maps have been compared with reference fire perimeters provided by the Fire Atlas of Portugal. Results show that commission and omission errors in the output burned area maps are a function of the threshold applied to the fuzzy score maps; between the two extremes of the greatest producer’s accuracy (omission error < 10%) and user’s accuracy (commission error < 5%), an intermediate threshold value provides errors of about 20% over the study area. The integration of SAR backscatter allowed reducing local commission errors from 65.4% (using optical data, only) to 11.4%, showing to significantly mitigate local errors due to the presence of cloud shadows and wetland areas. Overall, the proposed method is flexible and open to further developments; also in the perspective of the European Space Agency (ESA) Sentinel missions operationally providing SAR and optical datasets.

Integration of Optical and SAR Data for Burned Area Mapping in Mediterranean Regions

The work focuses on developing a classification tree approach for in-season crop mapping during early summer, by integrating optical (Landsat 8 OLI) and X-band SAR (COSMO-SkyMed) data acquired over a test site in Northern Italy. The approach is based on a classification tree scheme fed with a set of synoptic seasonal features (minimum, maximum and average, computed over the multi-temporal datasets) derived from vegetation and soil condition proxies for optical (three spectral indices) and X-band SAR (backscatter) data. Best performing input features were selected based on crop type separability and preliminary classification tests. The final outputs are crop maps identifying seven crop types, delivered during the early growing season (mid-July). Validation was carried out for two seasons (2013 and 2014), achieving overall accuracy greater than 86%. Results highlighted the contribution of the X-band backscatter (σ°) in improving mapping accuracy and promoting the transferability of the algorithm over a different year, when compared to using only optical features.

/pdf/in-season-mapping-of-crop-type-with-optical-and-x-band-sar-2qlnkln6wt.pdf

In-Season Mapping of Crop Type with Optical and X-Band SAR Data: A Classification Tree Approach Using Synoptic Seasonal Features

This paper aims at investigating the potential of Sentinel-1 C-band synthetic aperture radar (SAR) observations for detecting fire scars in vegetated areas at regional scale. A comprehensive analysis of the backscattering coefficients is carried out. The experimental analysis is conducted by analyzing the scenario of the Sardinia Island, which is one of the Italian regions most affected by fire events over the summer season. The detection capability of fire scars in such an environment is demonstrated by exploiting information extracted from dual-polarized SAR data. Our results reveal a significant decrease of the VH response over the fire-disturbed forests, thus, highlighting the effectiveness of such cross-polarized observations. In order to prove the validity of the proposed approach for the detection of fire scars in the vegetation layer, the results of the conducted experiments have been suitably compared with burned areas identified by using an existing fuzzy-based algorithm, which has been applied to multispectral Landsat-8 operational land imager data. This investigation opens the way to systematic methods for monitoring fire scars in heterogeneous environments, and in particular in fire-prone Mediterranean ecosystems.

Effect of the Vegetation Fire on Backscattering: An Investigation Based on Sentinel-1 Observations

This study investigated the feasibility of delivering a crop type map early during the growing season. Landsat 8 OLI multi-temporal data acquired in 2013 season were used to classify seven crop types in Northern Italy. The accuracy achieved with four supervised algorithms, fed with multi-temporal spectral indices (EVI, NDFI, RGRI), was assessed as a function of the crop map delivery time during the season. Overall accuracy (Kappa) exceeds 85% (0.83) starting from mid-July, five months before the end of the season, when maximum accuracy is reached (OA=92%, Kappa=0.91). Among crop types, rice is the most accurately classified, followed by forages, maize and arboriculture, while soybean or double crops can be confused with other classes.

Assessing in-season crop classification performance using satellite data: a test case in Northern Italy

This paper describes a mapping project carried out using both optical and SAR data on an agricultural area in northern Italy where the main crops are corn, rice and wheat. Temporal trends of backscatter and reflectance, given by the variations in vegetation growth, soil conditions and agricultural practices were analyzed and interpreted thanks to the ground-measured data. Information extracted from both optical and SAR data (vegetation indices, backscatter and texture features) were used to create training sets for implementing three different classification approaches. The work aimed at comparing early crop maps with maps derived at the end of the season. Results show that the classification accuracy obtained using only multispectral optical data is higher than the one reached using only SAR as input. Integrating both optical and SAR multitemporal features provides some advantages in terms of a more reliable crop map, especially during an early temporal stage scenario. Among the supervised algorithms tested, Maximum Likelihood shows the best overall accuracy performances at each thematic level, time step and using both optical and SAR input data.

http://space4agri.irea.cnr.it/it/prodotti-della-ricerca/presentazione-igarss-2014-1

Ramin Azar

Papers

Integration of Optical and SAR Data for Burned Area Mapping in Mediterranean Regions

In-Season Mapping of Crop Type with Optical and X-Band SAR Data: A Classification Tree Approach Using Synoptic Seasonal Features

Effect of the Vegetation Fire on Backscattering: An Investigation Based on Sentinel-1 Observations

Assessing in-season crop classification performance using satellite data: a test case in Northern Italy

Agricultural crop mapping using optical and SAR multi-temporal seasonal data: A case study in Lombardy region, Italy