Georgi Jelev

Single- and Multi-Date Crop Identification Using PROBA-V 100 and 300 m S1 Products on Zlatia Test Site, Bulgaria

The monitoring of crops is of vital importance for food and environmental security in a global and European context. The main goal of this study was to assess the crop mapping performance provided by the 100 m spatial resolution of PROBA-V compared to coarser resolution data (e.g., PROBA-V at 300 m) for a 2250 km2 test site in Bulgaria. The focus was on winter and summer crop mapping with three to five classes. For classification, single- and multi-date spectral data were used as well as NDVI time series. Our results demonstrate that crop identification using 100 m PROBA-V data performed significantly better in all experiments compared to the PROBA-V 300 m data. PROBA-V multispectral imagery, acquired in spring (March) was the most appropriate for winter crop identification, while satellite data acquired in summer (July) was superior for summer crop identification. The classification accuracy from PROBA-V 100 m compared to PROBA-V 300 m was improved by 5.8% to 14.8% depending on crop type. Stacked multi-date satellite images with three to four images gave overall classification accuracies of 74%–77% (PROBA-V 100 m data) and 66%–70% (PROBA-V 300 m data) with four classes (wheat, rapeseed, maize, and sunflower). This demonstrates that three to four image acquisitions, well distributed over the growing season, capture most of the spectral and temporal variability in our test site. Regarding the PROBA-V NDVI time series, useful results were only obtained if crops were grouped into two broader crop type classes (summer and winter crops). Mapping accuracies decreased significantly when mapping more classes. Again, a positive impact of the increased spatial resolution was noted. Together, the findings demonstrate the positive effect of the 100 m resolution PROBA-V data compared to the 300 m for crop mapping. This has important implications for future data provision and strengthens the arguments for a second generation of this mission originally designed solely as a “gap-filler mission”.

Clustering of spectral images using Echo state networks

In the present work we applied a recently developed procedure for multidimensional data clustering to processing of spectral satellite images. The core of our approach lays in projection of multidimensional image to a two dimensional one. The main aim is to discover points with similar characteristics. This was done by clustering of the resulting image. The processing technique exploits equilibrium states of a kind of recurrent neural network - Echo state network (ESN) - that are obtained after intrinsic plasticity (IP) tuning of the ESN using multidimensional data as inputs. The proposed in our previous work automated procedure for multidimensional data clustering is further refined and tested on the satellite image data. The obtained number and position of clusters of a multi-spectral image of a mountain region in Bulgaria is compared with the classification of the region landscape given by the Ministry of Regional Development and Public Works.

Recurrent neural networks for automatic clustering of multispectral satellite images

In the present work we applied a recently developed procedure for multidimensional data clustering to multispectral satellite images. The core of our approach lays in projection of the multidimensional image to a two dimensional space. For this purpose we used extensively investigated family of recurrent artificial neural networks (RNN) called “Echo state network” (ESN). ESN incorporates a randomly generated recurrent reservoir with sigmoid nonlinearities of neurons outputs. The procedure called Intrinsic Plasticity (IP) that is aimed at reservoir output entropy maximization was applied for adapting of reservoir steady states to the multidimensional input data. Next we consider all possible combinations between steady states of each two neurons in the reservoir as two-dimensional projections of the original multidimensional data. These low dimensional projections were subjected to subtractive clustering in order to determine number and position of data clusters. Two approaches to choose a proper projection among the all possible combinations between neurons were investigated. The first one is based on the calculation of two-dimensional density distributions of each projection, determination of number of their local maxima and choice of the projections with biggest number of these maxima. The second one applies clustering to all projections and chooses those with maximum number of clusters. Multispectral data from Landsat 7 Enhanced Thematic Mapper Plus (ETM+) instrument are used in this work. The obtained number and position of clusters of a multi-spectral image of a mountain region in Bulgaria is compared with the regional landscape classification.

Validation of MERIS LAI and FAPAR products for winter wheat-sown test fields in North-East Bulgaria

Progress in deriving land surface biophysical parameters in a spatially explicit manner using remotely sensed data has greatly enhanced our ability to model ecosystem processes and monitor crop development. A multitude of satellite sensors and algorithms have been used to generate ready-to-use maps of various biophysical parameters. Validation of these products for different vegetation types is needed to assess their reliability and consistency. While most of the current satellite biophysical products have spatial resolution of one kilometre, a recent effort utilizing data from the Medium Resolution Imaging Spectrometer (MERIS) provided leaf area index (LAI), Fraction of Absorbed Photosynthetically Active Radiation (FAPAR), and other canopy parameters in a resolution as fine as 300 m over the European continent. This resolution would be more appropriate for application at the regional scale, particularly for crop monitoring. This higher-resolution MERIS product has been evaluated in a limited number of studies to date. This article aims to validate LAI and FAPAR from the MERIS 10-day composite BioPar BP-10 product over winter wheat fields in northeast Bulgaria. The ground measurements of LAI and FAPAR were up-scaled and 30 m resolution reference raster layers were created using empirical relationships with Landsat TM (RMSE = 0.06 and RMSE = 0.22 for FAPAR and LAI, respectively). MERIS FAPAR and LAI were found to have significant correlation with FAPAR and LAI from the reference raster layers (R2 = 0.84 and R2 = 0.78, respectively). When MERIS Green LAI was calculated (incorporating the fraction of vegetation and brown vegetation cover from the BioPar BP-10 product), better correspondence with LAI values from the reference raster layer was achieved, with RMSE and bias reduced by 30–35%. The results from this study confirm the findings of previous validations showing that MERIS Green LAI tends to overestimate LAI values lower than 1. As a conclusion of the study, the BioPar BP-10 product was found to provide reliable estimates of FAPAR and acceptably accurate estimates of LAI for winter wheat crops in North-East Bulgaria.

Monitoring of winter crop status in Bulgaria using a series of NOAA AVHRR NDVI images

The study is an application of satellite remote sensing methods for monitoring winter crops in Bulgaria over the time period 1997–2008 based on 7-day NOAA AVHRR NDVI level 3 satellite imagery. A satellite image geodatabase was created in ArcGIS/ArcInfo 9.2 and was supplemented with ground-based phenological data. The monitoring is focused on vegetation status and stress situations, making use of vegetation condition index (VCI) images obtained from the NDVI time series. The VCI images covered the four main phenophases of winter crops for the agricultural years 2006–2007 and 2007–2008. By analyzing the average values of these VCI images, the status of each phenophase for the considered years was assessed. The results pointed to drought conditions for 2006–2007 and favourable to optimal conditions for 2007–2008.

Algorithms for lineaments detection in processing of multispectral images

Satellite remote sensing is a universal tool to investigate the different areas of Earth and environmental sciences. The advancement of the implementation capabilities of the optoelectronic devices which are long-term-tested in the laboratory and the field and are mounted on-board of the remote sensing platforms further improves the capability of instruments to acquire information about the Earth and its resources in global, regional and local scales. With the start of new high-spatial and spectral resolution satellite and aircraft imagery new applications for large-scale mapping and monitoring becomes possible. The integration with Geographic Information Systems (GIS) allows a synergistic processing of the multi-source spatial and spectral data. Here we present the results of a joint project DFNI I01/8 funded by the Bulgarian Science Fund focused on the algorithms of the preprocessing and the processing spectral data by using the methods of the corrections and of the visual and automatic interpretation. The objects of this study are lineaments. The lineaments are basically the line features on the earths surface which are a sign of the geological structures. The geological lineaments usually appear on the multispectral images like lines or edges or linear shapes which is the result of the color variations of the surface structures. The basic geometry of a line is orientation, length and curve. The detection of the geological lineaments is an important operation in the exploration for mineral deposits, in the investigation of active fault patterns, in the prospecting of water resources, in the protecting people, etc. In this study the integrated approach for the detecting of the lineaments is applied. It combines together the methods of the visual interpretation of various geological and geographical indications in the multispectral satellite images, the application of the spatial analysis in GIS and the automatic processing of the multispectral images by Canny algorithm, Directional Filter and Neural Network. Landsat multispectral images of the Eastern Rhodopes in Bulgaria for carrying out the procedure are used. Canny algorithm for extracting edges represents series of filters (Gaussian, Sobel, etc.) applied to all bands of the image using the free IDL source. Directional Filter is applied to sharpen the image in a specific preferred direction. Another method is the Neural Network algorithm for recognizing lineaments. The lineaments are effectively extracted using different methods of automatic. The results from the above mentioned methods are compared to the results derived from the visual interpretation of satellite images and from the geological map. In conclusion, the rose diagrams of the distribution of the geological lineaments and the maps of their density are completed.