This paper describes the results of the work, which was divided into three parts: 1) the proof-of-concept of lightweight low cost UAV-borne multispectral NIR (shortwave infrared) sensor, 2) the evaluation of performances of airborne tailored NIR sensor vs. consumer-grade RGB digital camera, and 3) the feasibility study and the comparison of NIR multispectral data vs. Sentinel-2A high-resolution satellite multispectral imagery. The moderated cost-efficient UAV-borne imaging remote sensing solution was requested by the agricultural sector in Ukraine. The existing solutions did not meet the requirements of the end user neither suitable high-resolution satellite multispectral imagery was available too. The designed system integrated consumer-grade RGB digital camera and NIR sensor. Multi-temporary data was collected eight times during one growth season. Eight aerial missions were conducted in midsummer 2015 over the same area, which consisted of 50 plots with various cultivars of wheat, barley and rye. The vegetation indices were calculated for both datasets. Vegetation indices calculated from the NIR sensor highly correlated with plant chlorophyll and plant LAI, and revealed satisfactory correlation with plant fresh mass. Vegetation indices derived from onboard RGB camera did not reveal the significant correlation with any of plant growth parameter. The NDVI values calculated from NIR data demonstrated high correlation with the ones, which were derived from the Sentinel-2 high-resolution satellite multispectral images. The values of UAV-derived NDVI were lower than Sentinel-2-derived NDVI, and the regression slope of this relationship varied in different plant species. Reasons of such variation are discussed in the paper. After the numerous field-tests the customer accepted the developed tailored COTS UAV-borne multispectral solution cost-efficient and sufficient.
Vegetation is a sensitive indicator suitable for testing of ecological stresses and natural anomalies of the technogenic character. First, it is determined by the prompt response of photosynthetic apparatus to changes of environmental conditions, mainly by change of green pigment (chlorophyll) content in leaves. Second, the specific kind of a reflectance spectrum of leaves is due to chlorophyll presence in them, and the area in the range of 500-80 nm is extremely sensitive to variations of its pigment content. Thirdly, there are interesting results now concerning spectral properties of leaves and crops canopies obtaining with high-resolution spectroscopy. The data are high informative in relation to content of chlorophyll and some other biochemical constituents of a cell. The high resistance to various types of noises is inherent to methods developed on the basis of such spectral data. We have developed a method for chlorophyll estimation using the 1-st derivative plots of reflectance spectral curves. The method gives good results for plant-soil systems with both for 100% and
incomplete projective covering as our simulation models show. Field measurements of chlorophyll content in closed and open canopies crops confirm the results. A hardware-software complex has been produced by us for chlorophyll determining under field conditions. It consists of spectral and computing blocks. First of them is a two-beam spectrometer of high resolution supplied by a system to visualize of measured object. The irradiance and temperature sensors are included to the spectral block as well as GPS-receiver. The following technical characteristics are inherent to the block: spectral range 500-800 nm, band-pass 1.5 nm, field of view 16x16o, scanning time 0.1-1.0 s, dynamic range of signal 1:1024 (10 bit), signal/noise ratio 400, amount of pixels in image 1240, range of estimated chlorophyll concentrations 1.5-8.0 mg/dm2, supply voltage 12 V, weight 8 kg. Computing block is intended for spectral date processing to obtain chlorophyll estimations using our algorithm. The block is supplied by our original software WINCHL, which includes spectrum and algorithm
libraries and various mathematical tools. Accumulation of reflectance spectra of various plants together with data of environmental conditions at measurements gives a good possibility to use all of them for future scientific researches and developing other important parameters of canopy status.