Uwe Rascher’s research while affiliated with Forschungszentrum Jülich and other places

Cauliflower cultivation is subject to high-quality control criteria during sales, which underlines the importance of accurate harvest timing. Using time series data for plant phenotyping can provide insights into the dynamic development of cauliflower and allow more accurate predictions of when the crop is ready for harvest than single-time observations. However, data acquisition on a daily or weekly basis is resource-intensive, making selection of acquisition days highly important. We investigate which data acquisition days and development stages positively affect the model accuracy to get insights into prediction-relevant observation days and aid future data acquisition planning. We analyze harvest-readiness using the cauliflower image time series of the GrowliFlower dataset. We use an adjusted ResNet18 classification model, including positional encoding of the data acquisition dates to add implicit information about development. The explainable machine learning approach GroupSHAP analyzes time points' contributions. Time points with the lowest mean absolute contribution are excluded from the time series to determine their effect on model accuracy. Using image time series rather than single time points, we achieve an increase in accuracy of 4%. GroupSHAP allows the selection of time points that positively affect the model accuracy. By using seven selected time points instead of all 11 ones, the accuracy improves by an additional 4%, resulting in an overall accuracy of 89.3%. The selection of time points may therefore lead to a reduction in data collection in the future.

In this study, we address two relevant gaps when monitoring plant photosynthesis using remote sensing techniques; these are i) assess the seasonal trends and relationships observed between photosynthesis, optical vegetation indices, and chlorophyll fluorescence in crop systems and ii) evaluate the contribution of Sun-induced chlorophyll fluorescence (SIF) on linear and non-linear light-use efficiency-based (LUE) models for the remote estimation of plant photosynthesis. Coincident measurements of net plant photosynthesis (Anet), optical vegetation indices (i.e., Red edge index and photochemical reflectance index (PRI) among others), photochemical quantum yield (ΦPSII), and SIF were made at leaf level once a week in a wheat field under different nitrogen treatments. In LUE models, three key variables explain the seasonal variability of photosynthesis; these are the fraction of absorbed photosynthetically active radiation (fAPAR), LUE, and a correction factor related to meteorological conditions that limit LUE. In this study, the Red edge index was highly correlated with fAPAR (R2>0.70, p-value<0.05); however, neither PRI nor SIF were able to explain the seasonal changes of LUE (R2<0.10). ΦPSII seasonal values (0.10 – 0.40) measured during the experiment indicated strong downregulation of the photosynthetic machinery. This explained why, in this study, SIF did not present a linear relationship with LUE. Our results confirmed that under stress conditions the non-photochemical quenching mechanisms (NPQ) control the energy dissipation pathway, breaking the linear relationship between photochemistry and fluorescence. Additionally, our study proved that changes in Anet could be better explained when optical vegetation indices, chlorophyll fluorescence, and meteorological conditions are combined in non-linear LUE-based models (R2 increased from 0.10 for the linear model to 0.60 for the non-linear model). These results confirmed the need to build non-linear models for the remote quantification of photosynthesis.

Mapping crop variables at different growth stages is crucial to inform farmers and plant breeders about the crop status. For mapping purposes, inversion of canopy radiative transfer models (RTMs) is a viable alternative to parametric and non-parametric regression models, which often lack transferability in time and space. Due to the physical nature of RTMs, inversion outputs can be delivered in sound physical units that reflect the underlying processes in the canopy. In this study, we explored the capabilities of the coupled leaf–canopy RTM PROSAIL applied to high-spatial-resolution (0.015 m) multispectral unmanned aerial vehicle (UAV) data to retrieve the leaf chlorophyll content (LCC), leaf area index (LAI) and canopy chlorophyll content (CCC) of sweet and silage maize throughout one growing season. Two different retrieval methods were tested: (i) applying the RTM inversion scheme to mean reflectance data derived from single breeding plots (mean reflectance approach) and (ii) applying the same inversion scheme to an orthomosaic to separately retrieve the target variables for each pixel of the breeding plots (pixel-based approach). For LCC retrieval, soil and shaded pixels were removed by applying simple vegetation index thresholding. Retrieval of LCC from UAV data yielded promising results compared to ground measurements (sweet maize RMSE = 4.92g/2, silage maize RMSE = 3.74g/2) when using the mean reflectance approach. LAI retrieval was more challenging due to the blending of sunlit and shaded pixels present in the UAV data, but worked well at the early developmental stages (sweet maize RMSE = 0.70m2/m2, silage RMSE = 0.61m2/m2 across all dates). CCC retrieval significantly benefited from the pixel-based approach compared to the mean reflectance approach (RMSEs decreased from 45.6 to 33.1 g/m2). We argue that high-resolution UAV imagery is well suited for LCC retrieval, as shadows and background soil can be precisely removed, leaving only green plant pixels for the analysis. As for retrieving LAI, it proved to be challenging for two distinct varieties of maize that were characterized by contrasting canopy geometry.

The PRISMA satellite mission launched on March 22nd, 2019 is one of the latest spaceborne imaging spectroscopy mission for Earth Observation. The PRISMA satellite comprises a high-spectral resolution VNIR-SWIR imaging spectrometer and a panchromatic camera. In summer 2019, first operations during the commissioning phase were mainly devoted to acquisitions in specific areas for evaluating instrument functioning, in-flight performance, and mission data product accuracy. A field and airborne campaign was carried out over an agriculture area in Italy to collect in-situ multi-source spectroscopy measurements at different scales simultaneously with PRISMA. The spectral, radiometric and spatial performance of PRISMA Level 1 Top-Of-Atmosphere radiance (LTOA) product were analyzed. The in-situ surface reflectance measurements over different landcovers were propagated to LTOA using MODTRAN5 radiative transfer simulations and compared with satellite observations. Overall, this work offers a first quantitative evaluation about the PRISMA mission performance and imaging spectroscopy LTOA data product consistency. Our results show that the spectral smile is less than 5 nm, the average spectral resolution is 13 nm and 11 nm (VNIR and SWIR respectively) and it varies ±2 nm across track. The radiometric comparison between PRISMA and field/airborne spectroscopy shows a difference lower than 5% for NIR and SWIR, whereas it is included in the 2–7% range in the VIS. The estimated instrument signal to noise ratio (SNR) is ≈400–500 in the NIR and part of the SWIR (<1300 nm), lower SNR values were found at shorter (<700 nm) and longer wavelengths (>1600 nm). The VNIR-to-SWIR spatial co-registration error is below 8 m and the spatial resolution is 37.11 m and 38.38 m for VNIR and SWIR respectively. The results are in-line with the expectations and mission requirements and indicate that acquired images are suitable for further scientific applications. However, this first assessment is based on data from a rural area and this cannot be fully exhaustive. Further studies are needed to confirm the performance for other land cover types like snow, inland and coastal waters, deserts or urban areas.

The Copernicus Hyperspectral Imaging Mission for the Environment (CHIME) is in preparation to carry a unique visible to shortwave infrared spectrometer. CHIME will globally provide routine hyperspectral observations to support new and enhanced services for, among others, sustainable agricultural and biodiversity management. The mission shall provide Level 1B, 1C and 2A products, as well a set of downstream products related to the different environmental applications, such as the quantification of vegetation traits. In this context, this work presents the first hybrid retrieval models for the operational delivery of vegetation properties products. Within ESA's CHIME end-to-end (E2E) simulator study, 13 leaf and canopy trait models were developed as part of the L2B vegetation (L2BV) module. The E2E framework functions as a simulated reality that enabled to test and improve the algorithms. The models were further tuned and validated against campaign data using active learning methods. As a proof of concept, the prototype retrieval models were applied to both hyperspectral airborne (HyPlant) and spaceborne (PRISMA) imagery that were first resampled to CHIME band settings. Among the provided vegetation products, it led to a first space-based canopy nitrogen content map over a heterogeneous landscape. The obtained CHIME-like L2BV traits maps demonstrate the feasibility to routinely deliver a collection of next-generation vegetation products across the globe.