Precise and timely information on biomass yield and nitrogen uptake in intensively managed grasslands are essential for sustainable management decisions. Imaging sensors mounted on unmanned aerial vehicles (UAVs) along with photogrammetric structure-from-motion processing can provide timely data on crop traits rapidly and non-destructively with a high spatial resolution. The aim of this multi-temporal field study is to estimate aboveground dry matter yield (DMY), nitrogen concentration (N%) and uptake (Nup) of temperate grasslands from UAV-based image data using machine learning (ML) algorithms. The study is based on a two-year dataset from an experimental grassland trial. The experimental setup regarding climate conditions, N fertilizer treatments and slope yielded substantial variations in the dataset, covering a considerable amount of naturally occurring differences in the biomass and N status of grasslands in temperate regions with similar management strategies. Linear regression models and three ML algorithms, namely, random forest (RF), support vector machine (SVM), and partial least squares (PLS) regression were compared with and without a combination of both structural (sward height; SH) and spectral (vegetation indices and single bands) features. Prediction accuracy was quantified using a 10-fold 5-repeat cross-validation (CV) procedure. The results show a significant improvement of prediction accuracy when all structural and spectral features are combined, regardless of the algorithm. The PLS models were outperformed by their respective RF and SVM counterparts. At best, DMY was predicted with a median RMSECV of 197 kg ha−1, N% with a median RMSECV of 0.32%, and Nup with a median RMSECV of 7 kg ha−1. Furthermore, computationally less expensive models incorporating, e.g., only the single multispectral camera bands and SH metrics, or selected features based on variable importance achieved comparable results to the overall best models.

Questions Fertilizer application is a key driver affecting the diversity and conservation value grassland vegetation. Using a long-term fertilization experiment in mountain grassland, we addressed the following questions: (a) what is the effect of long-term fertilizer applications on species richness and plant species composition, and (b) is there any detectable trend in plant species composition during ten years of continuous observation? Location The Rengen Grassland Experiment, Eifel Mountains, Germany (established in 1941). Methods Five treatments including different fertilizers applied annually and one unfertilized control were analysed: Ca (718 kg Ca ha(-1)); CaN (752 Ca and 100 N kg ha(-1)); CaNP (752 Ca, 100 N and 35 P kg ha(-1)); CaNP-KCl (752 Ca, 100 N, 35 P and 133 K kg ha(-1)); CaNP-K2SO4 (752 Ca, 100 N, 35 P and 133 K kg ha(-1)). The experiment included five replicates per treatment in a fully randomized block design. All treatments were cut twice a year in late June or early July, and in mid-October. Percentage cover of individual plant species was estimated by visual observation in each plot in late June in the years 2005-2014. Results Despite inter-annual variability in the cover of the individual vascular plant species, the multivariate data analyses revealed a relatively similar response of the plant community to the different fertilizer applications throughout the ten years. With phosphorus application, no differences in botanical composition among treatments were found; however, they did differ from other treatments without phosphorus application. In the unfertilized control, there was a certain directed trend in plant species composition in response to ongoing nutrient impoverishment. Conclusion Species-rich grasslands of high nature conservation value were only maintained under P limitation in the control without fertilizer application (Violion caninae) and in the liming treatment (Polygono-Trisetion), but also in the treatment with liming and pure N addition. It seems that after 74 years, some stage of equilibrium of the grassland community was achieved in all treatments receiving any type of fertilizer application, but less so in the unfertilized control treatment.

Globally, drought constitutes a serious threat to food and water security. The complexity and multivariate nature of drought challenges its assessment, especially at local scales. The study aimed to assess spatiotemporal patterns of crop condition and drought impact at the spatial scale of field management units with a combined use of time-series from optical (Landsat, MODIS, Sentinel-2) and Synthetic Aperture Radar (SAR) (Sentinel 1) data. Several indicators were derived such as Normalized Difference Vegetation Index (NDVI), Normalized Difference Moisture Index (NDMI), Land Surface Temperature (LST), Tasseled cap indices and Sentinel-1 based backscattering intensity and relative surface moisture. We used logistic regression to evaluate the drought-induced variability of remotely sensed parameters estimated for different phases of crop growth. The parameters with the highest prediction rate were further used to estimate thresholds for drought/non-drought classification. The models were evaluated using the area under the receiver operating characteristic curve and validated with in-situ data. The results revealed that not all remotely sensed variables respond in the same manner to drought conditions. Growing season maximum NDVI and NDMI (70–75%) and SAR derived metrics (60%) reflect specifically the impact of agricultural drought. These metrics also depict stress affected areas with a larger spatial extent. LST was a useful indicator of crop condition especially for maize and sunflower with prediction rates of 86% and 71%, respectively. The developed approach can be further used to assess crop condition and to support decision-making in areas which are more susceptible and vulnerable to drought.

Remote sensing systems based on unmanned aerial vehicles (UAVs) are well suited for airborne monitoring of small to medium-sized farmland in agricultural applications. An imaging system is often used in the form of a multispectral multi-camera system to derive well-established vegetation indices (VIs) efficiently. This study investigates the potential of such a multi-camera system with a novel approach to extend spectral sensitivity from visible-to-near-infrared (VNIR) to short-wave infrared (SWIR) (400–1700 nm) for estimating forage mass from an aerial carrier platform. The system test was performed in a grassland fertilizer trial in Germany near Cologne in late July 2019. Within 37 min, a spectral response in four different wavelength bands in the NIR and SWIR range was acquired during two consecutive flights. Spectral image data were calibrated to reflectance using two different methods. The resulting reflectance data sets were processed to orthomosaics for each wavelength band. From these orthomosaics for both calibration methods, the four-band NIR/SWIR GnyLi VI and the two-band NIR/SWIR Normalized Ratio Index (NRI), were calculated. During both UAV flights, spectral ground truth data were recorded with a spectroradiometer on 12 plots in total for validation of camera-based spectral data. The camera and spectroradiometer data sets were directly compared in resulting reflectance and further analyzed with simple linear regression (SLR) models to predict dry matter (DM) yield. In the camera-based SLRs, the NRI performed best with $$R^2$$ of 0.73 and 0.75 (RMSE: 0.18 and 0.17) before the GnyLi with $$R^{2}$$ of 0.71 and 0.73 (RMSE: 0.19 and 0.18). These results clearly indicate the potential of the camera system for applications in forage mass monitoring.

Monitoring and predicting above ground biomass yield of grasslands are of key importance for grassland management. Established manual methods such as clipping or rising plate meter measurements provide accurate estimates of forage yield, but are time consuming and labor intensive, and do not provide spatially continuous data as required for precision agriculture applications. Therefore, the main objective of this study is to investigate the potential of sward height metrics derived from low-cost unmanned aerial vehicle-based image data to predict forage yield. The study was conducted over a period of 3 consecutive years (2014–2016) at the Rengen Grassland Experiment (RGE) in Germany. The RGE was established in 1941 and is since then under the same management regime of five treatments in a random block design and two harvest cuts per year. For UAV-based image acquisition, a DJI Phantom 2 with a mounted Canon Powershot S110 was used as a low-cost aerial imaging system. The data were investigated at different levels (e.g., harvest date-specific, year-specific, and plant community-specific). A pooled data model resulted in an R 2 of 0.65 with a RMSE of 956.57 kg ha −1 , although cut-specific or date-specific models yielded better results. In general, the UAV-based metrics outperformed the traditional rising plate meter measurements, but was affected by the timing of the harvest cut and plant community.

Grass growth and utilisation on grassland farms across the UK and Ireland are below their potential, and nutrient management is sub-optimal. Excessive soil phosphorus concentrations are an issue, contributing to eutrophication of waterways, while sub-optimal soil pH due to a reduction in lime use is seen in many areas. This paper discusses the potential for introducing technology onto grassland farms, and implementing sensible farm-specific management decisions as a means of enhancing productivity and minimising the environmental impact of livestock fanning. The adoption rate of Precision Agriculture (PA) technology by grassland farmers in the UK and Ireland is still relatively low, and significantly lagging behind the advances that have taken place within the arable farming community. With rapidly advancing technology and investment in research, there are now a range of soil sensors available along with technology for grass growth monitoring, yield mapping, remote sensing, satellite imagery, and devices for tracking animal movement, behaviour and feed intake. In this review paper we examine the latest technology developments specific to grassland farming and we question their suitability and 'readiness' for adoption by farmers. The review identifies the challenges that are specific to grassland such as complex soil, vegetation and climatic variables along with economic barriers that may be preventing change within the farming community in these areas.

It remains challenging to assess the goods and ecosystem services (ES) that grasslands provide to agriculture and society. Their characterization is stakeholder dependent. Some grassland ES depend mostly on the spatiotemporal distribution of grasslands, whereas others depend strongly on grassland plant composition. Grassland management strongly determines the level of ES but also of environmental impacts. In this chapter, we present a framework to analyze ecosystem goods and services and environmental impacts related to grasslands. Based on a socioecologic approach, it helps to clarify the relative role of the spatiotemporal distribution and composition of grasslands and of key levels of analysis. Based on a functional ecology approach, it analyzes relations between grassland environmental factors, management, functional diversity and goods, services, and impacts from grasslands. To address the functional diversity issue, grassland plants are characterized by their growth strategies and their functional and temporal complementarities. This framework can also analyze trade-offs and synergies. To demonstrate the utility of the framework, we apply it to a multilevel case study: 186 seminatural grasslands and short-term (≤5 years) and long-term (>5 years) temporary grasslands of six farms located in the “Segala” region (southern Massif Central, France). At the field level, some of the short-term temporary grasslands are spring centered with low legume content, while others are season centered with high legume content. When a single farm contains a wide range of within- or between-field functional plant diversity due to contrasting land use, including the combination of temporary and seminatural grasslands, this promotes ES and limits environmental impacts. At the landscape level, the great spatiotemporal diversity of grasslands is likely to promote specific regulating services in addition to the effect of their composition. Finally, we investigate the need for cross-scale analysis to identify trade-offs between ES and the need to define what design, assessment, and management approaches should be developed to enhance ES.

Monitoring grassland biomass throughout the growing season is of key importance in sustainable, site-specific management decisions. Precision agriculture applications can support these decisions. However, precision agriculture relies on timely and accurate information on plant parameters with a high spatial and temporal resolution. The use of structural and spectral features derived from unmanned aerial vehicle (UAV)-based image data from low-cost sensors is a promising nondestructive approach to assess plant traits such as above-ground biomass or plant height. Therefore, the main objectives were (1) to evaluate the potential of low-cost UAV-based canopy surface models to monitor sward height as an indicator of grassland biomass, (2) to evaluate the potential of vegetation indices from low-cost UAV-based red-greenblue (RGB) digital image data, and (3) to compare the mentioned methods with established methods for biomass monitoring such as rising plate meters and spectroradiometer-based narrowband vegetation indices over the growing season in 2017, including three cuts. We compared the accuracy of each single UAV-based height feature and vegetation index using a combined multivariate approach to estimate fresh and dry biomass. The heterogeneous sward structure with high spatiotemporal variability led to varying performance in biomass estimation depending on the growths (time between two cuts) and choice of predictor variable. The results showed that biomass prediction by height features provided moderate-to-good results (cross-validation R-2 = 0.57 to 0.73 for dry biomass and 0.43 to 0.79 for fresh biomass), but reference measurements based on rising plate meters were more robust when estimating biomass. The spectral features (RGB-based vegetation indices and spectroradiometer-based vegetation indices) yielded varying accuracy and suitability for biomass prediction. Despite the variability, our findings indicate a promising approach for grassland biomass monitoring. (C) The Authors. Published by SPIE under a Creative Commons Attribution 4.0 Unported License.

Forage monitoring in grassland is an important task to support management decisions. Spatial data on (i) yield,(ii) quality, and (iii) floristic composition are of interest. The spatio-temporal variability in grasslands is significant and requires fast and low-cost methods for data delivery. Therefore, the overarching aim of this contribution is the investigation of low-cost and non-calibrated UAV-derived RGB imagery for forage monitoring. Study area is the Rengen Grassland Experiment (RGE) in Germany which is a long-term field experiment since 1941. Due to the experiment layout, destructive biomass sampling during the growing period was not possible. Hence, non-destructive Rising Plate Meter (RPM) measurements, which are a common method to estimate biomass in grasslands, were carried out. UAV campaigns with a Canon Powershot 110 mounted on a DJI Phantom 2 were conducted in the first growing season in 2014. From the RGB imagery, the RGB vegetation index (RGBVI) and the Grassland Index (GrassI) introduced by Bendig et al. (2015) and Bareth et al. (2015), respectively, were computed. The RGBVI and the GrassI perform very well against the RPM measurements resulting in R2 of 0.84 and 0.9, respectively. These results indicate the potential of low-cost UAV methods for grassland monitoring and correspond well to the studies of Viljanen et al. (2018) and Näsi et al. (2018).

Crop growth and condition monitoring is important for estimating crop production and supporting agricultural management. We used several remotely sensed datasets for mapping phenometrics and assessing maize condition. For the derivation of phenometrics, MODIS data were integrated with Landsat-8 and Sentinel-2 to create synthetic time series in order to fill the missing values in original data. For crop condition monitoring, indicators were derived based on vegetation index time series and Land Surface Temperature (LST). The synthetic data could estimate the start and the end of the season with 3-12 days deviation. Vegetation indices were sensitive to crop condition during growing seasons with unfavorable hydroclimatic conditions. LST was a useful indicator of drought-induced crop stress with prediction rates over 80%. The developed approach can be further used to assess crop condition on field level in areas vulnerable to extreme events such as droughts.

Non-destructive methods to derive spatial information on the development of forage mass are of key importance in managed grasslands. Established methods are rising plate meter (RPM) and rapid pasture meter, which both require in-field work, are rather time consuming, and do not provide spatially continuous data. Therefore, the overall objective of this study is to investigate low-cost unmanned aerial vehicle (UAV)-based RGB image acquisition for grassland monitoring. The idea of this paper is to transfer the successfully introduced approach of crop surface models (CSMs) for ultrahigh resolution analysis of plant height to managed grasslands. The study area is the Rengen Long-term Grassland Experiment, Germany, which is a two-cut experiment and was established in 1941. We conducted RPM and UAV-based data acquisition over six growth periods in 2014, 2015, and 2016. In 3 years, 26 RPM and 46 UAV campaigns were conducted under varying weather conditions (cloudy/sunny). The UAV-based RGB imagery was photogrammetrically processed with Structure from Motion and Multi-view Stereopsis techniques, producing multi-temporal CSMs for grassland sward height analysis. The regression analysis of UAV-derived sward height (CSM-SH) against RPM-measured sward height (RPM-SH) resulted in R 2 of 0.91, 0.87, and 0.83 for 2014, 2015, and 2016, respectively. The pooling of the data for all 3 years resulted in an R 2 of 0.86. These findings prove the successful transfer of the CSM approach for grassland monitoring and the potential of UAV-based monitoring to replace manual or in-field measurements with RPM or rapid pasture meter.

Climate change, food insecurity and limited land and water resources strengthen the need for operational and spatially explicit information on vegetation condition and dynamics. The detection of vegetation condition as well as multiannual and seasonal changes using satellite remote sensing, however, depends on the choice of data including length and frequency of time series. Thus, this contribution focuses on the derivation of the optimal remotely sensed data for vegetation monitoring and extraction of relevant metrics. Time series of satellite data from Landsat-8, Sentinel-1/2, and MODIS were used to identify characteristics of vegetation at different spatiotemporal scales. We derived parameters, such as: maximum and amplitude based on vegetation index time series, as well as Land Surface Temperature (LST). Along with optical data, we used backscattering intensity over consecutive vegetation growing seasons. The analysis was carried out using Google Earth Engine, a cloud computing platform which allows to access various data archives and conduct data-intensive analysis. Taking advantage of this platform, we developed a web-based application named GreenLeaf. The application is computing metrics and plotting time series, based on parameters defined by the user. The derived vegetation condition parameters provide sufficient information to detect vegetation change. In addition, the images acquired from near-coincident dates provide similar information over continuous surfaces. The developed application contributes to the use of satellite data and the simplification of data access for users with limited remote sensing experience and/or restricted processing power. Aiming at providing this knowledge to stakeholders can further support decision making on multiple scales.

Radiation Use Efficiency (RUE) defines the productivity with which absorbed photosynthetically active radiation (APAR) is converted to plant biomass. Readily used in crop growth models to predict dry matter accumulation, RUE is commonly determined by elaborate static sensor measurements in the field. Different definitions are used, based on total absorbed PAR (RUEtotal) or PAR absorbed by the photosynthetically active leaf tissue only (RUEgreen). Previous studies have shown that the fraction of PAR absorbed (fAPAR), which supports the assessment of RUE, can be reliably estimated via remote sensing (RS), but unfortunately at spatial resolutions too coarse for experimental agriculture. UAV-based RS offers the possibility to cover plant reflectance at very high spatial and temporal resolution, possibly covering several experimental plots in little time. We investigated if (a) UAV-based low-cost camera imagery allowed estimating RUEs in different experimental plots where maize was cultivated in the growing season of 2016, (b) those values were different from the ones previously reported in literature and (c) there was a difference between RUEtotal and RUEgreen. We determined fractional cover and canopy reflectance based on the RS imagery. Our study found that RUEtotal ranges between 4.05 and 4.59, and RUEgreen between 4.11 and 4.65. These values are higher than those published in other research articles, but not outside the range of plausibility. The difference between RUEtotal and RUEgreen was minimal, possibly due to prolonged canopy greenness induced by the stay-green trait of the cultivar grown. The procedure presented here makes time-consuming APAR measurements for determining RUE especially in large experiments superfluous.

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.

Accurate agricultural land use (LU) map is essential for many agro-environmental applications. With advances in technology, object-based image classification and non-parametric machine learning algorithms evolved. Still, no particular method has universal applicability. This paper compares robust non-parametric machine learning algorithms, random forest (RF) and support vector machine (SVM), and a common parametric algorithm maximum likelihood (MLC) based on multiple Landsat 8 images. We have also assessed the classifier performance relative to the choice either pixel-based (PB) or field-based (FB) approach. The study area, a semi-desert irrigated region, lies in Khorezm province and Republic of Karakalpakstan in Uzbekistan. Accuracy assessment showed higher overall accuracy (OA) and kappa index (KI) of the nonparametric machine learning FB-RF and FB-SVM algorithms over the PB-RF, PB-SVM and PB-MLC algorithms. The lowest OA and KI occurred with the parametric FB-MLC. Based on the results, the FB machine learning non-parametric algorithms are recommended for mapping irrigated croplands.

Long-term in vitro fermentation (240 h) evaluating amylase-treated, ash-corrected neutral detergent fiber, with addition of sodium sulfite (aNDFom) digestibility is required to quantify the indigestible fiber fraction. It is commonly accepted to inoculate rumen fluid more than one time during such fermentations, every 96 h or at 120 h. However, no studies have been conducted to verify if the reinoculation is actually required to properly carry out the fermentation process. The current study aims to evaluate the effects of these procedures on aNDFom digestibility at 240 h. The study was conducted on a total of 24 forage samples (8 alfalfa hays, 8 grass hays, and 8 corn silages). Samples were digested in triplicate at 240 h in vitro. Rumen fluid was added twice (at 96 and 192 h) in treatment 1, after 120 h in treatment 2, whereas no addition was made in treatment 3. At the end of the fermentations, residual aNDFom was quantified to calculate digestibility. Among treatments, no difference was found in digestibility of aNDFom. Moreover, treatment 1 resulted in higher variability compared with other treatments. Results obtained in the current study show that subsequent addition of rumen fluid is not necessary for a proper estimation of aNDFom digestibility, and can be avoided.

Prof. Dr. Gunter Menz passed away on 9 August 2016 following a dramatic accident.[...]

Proper parameterisation and quantification of model uncertainty are two essential tasks in improvement and assessment of model performance. Bayesian calibration is a method that combines both tasks by quantifying probability distributions for model parameters and outputs. However, the method is rarely applied to complex models because of its high computational demand when used with high-dimensional parameter spaces. We therefore combined Bayesian calibration with sensitivity analysis, using the screening method by Morris (1991), in order to reduce model complexity by fixing parameters to which model output was only weakly sensitive to a nominal value. Further, the robustness of the model with respect to reduction in the number of free parameters were examined according to model discrepancy and output uncertainty. The process-based grassland model BASGRA was examined in the present study on two sites in Norway and in Germany, for two grass species (Phleum pratense and Arrhenatherum elatius). According to this study, a reduction of free model parameters from 66 to 45 was possible. The sensitivity analysis showed that the parameters to be fixed were consistent across sites (which differed in climate and soil conditions), while model calibration had to be performed separately for each combination of site and species. The output uncertainty decreased slightly, but still covered the field observations of aboveground biomass. Considering the training data, the mean square error for both the 66 and the 45 parameter model was dominated by errors in timing (phase shift), whereas no general pattern was found in errors when using the validation data. Stronger model reduction should be avoided, as the error term increased and output uncertainty was underestimated. (C) 2017 Elsevier B.V. All rights reserved.

Monitoring the spectral response of intensively managed grassland throughout the growing season allows optimizing fertilizer inputs by monitoring plant growth. For example, site-specific fertilizer application as part of precision agriculture (PA) management requires information within short time. But, this requires field-based measurements with hyper- or multispectral sensors, which may not be feasible on a day to day farming practice. Exploiting the information of RGB images from consumer grade cameras mounted on unmanned aerial vehicles (UAV) can offer cost-efficient as well as near-real time analysis of grasslands with high temporal and spatial resolution. The potential of RGB imagery-based vegetation indices (VI) from consumer grade cameras mounted on UAVs has been explored recently in several. However, for multitemporal analyses it is desirable to calibrate the digital numbers (DN) of RGB-images to physical units. In this study, we explored the comparability of the RGBVI from a consumer grade camera mounted on a low-cost UAV to well established vegetation indices from hyperspectral field measurements for applications in grassland. The study was conducted in 2014 on the Rengen Grassland Experiment (RGE) in Germany. Image DN values were calibrated into reflectance by using the Empirical Line Method (Smith & Milton 1999). Depending on sampling date and VI the correlation between the UAV-based RGBVI and VIs such as the NDVI resulted in varying R2 values from no correlation to up to 0.9. These results indicate, that calibrated RGB-based VIs have the potential to support or substitute hyperspectral field measurements to facilitate management decisions on grasslands.