A rule-based approach for crop identification using multi-temporal and multi-sensor phenological metrics

Abstract

Accurate classification and mapping of crops is essential for supporting sustainable land management. Such maps can be created based on satellite remote sensing; however, the selection of input data and optimal classifier algorithm still needs to be addressed especially for areas where field data is scarce. We exploited the intra-annual variation of temporal signatures of remotely sensed observations and used prior knowledge of crop calendars for the development of a two-step processing chain for crop classification. First, Landsat-based time-series metrics capturing within-season phenological variation were preprocessed and analyzed using Google Earth Engine cloud computing platform. The developmental stage of each crop was modeled by fitting harmonic function. The model's output was further used for the automatic generation of training samples. Second, several classification methods (support vector machines, random forest, decision fusion) were tested. As input data for crop classification, composites based on Sentinel-1 and Landsat images were used. Overall classification accuracies exceeded 80% when the seasonal composites were used. Winter cereals were the most accurately classified, while we observed misclassifications among summer crops. The proposed approach offers a potential to accurately map crops in the areas where in situ field data are scarce or unavailable.

A workflow for Sustainable Development Goals indicators assessment based on high-resolution satellite data

“…In order to avoid overfitting, an L2 regularization was used with regularization coefficient set to 0.1, and learning rate was set to 10 −3 . A committee of neural networks is used for providing crop classification and land cover maps for Ukraine using high spatial resolution Landsat 8, Sentinel-1 and Sentinel-2 imagery (Skakun et al 2015;Kussul et al 2016;Kussul et al 2018;Ghazaryan et al 2018) and appropriate in-situ data for 2000, 2010, 2016 and 2017 (Shelestov et al 2017b;Skakun et al 2016;Waldner et al 2016). The spatial resolution of the resulting maps is 30 m for 2000 and 2010, and 10 m for 2016 and 2017 ( Figure 3).…”

From ecology to remote sensing: using animals to map land cover

“…; Ghazaryan et al. ), describes important phenological traits that help discriminate different types of land management. Moreover, using NDVI data from MODIS, which has a stable temporal and spatial acquisition geometry, assured the consistency of our results across different study regions, making them comparable.…”

Decision Support on Monitoring and Disaster Management in Agriculture with Copernicus Sentinel Applications

“…The type of crops, their crop calendar, and the diversity of agronomic practices are details that are critical for the selection of the image-set and are quite familiar to the local inspectors of the PAs. In most studies, using all the available cloud-free images is recommended to account for all the varying agronomic practices which cause misclassifications [33,35,36]. Image pre-processing steps include the data subset to the area of interest, estimation of several radiometric indices, and image stack of the several satellite data and products.…”

“…The first step for the application of the DSS crop type mapping functionality is the Sentinel-2 image query. The selection of the optimal dates and number of images depends on the area of interest, the variety of crop types, and the spectral heterogeneity of the crop types [33,35,36]. Variation in the dates of planting or harvesting is a common source of error in the image classification processes [36].…”