Geodaysit 2023

Carla Cesaraccio


Investigating PLSR and RF for retrieving wheat crop traits in a field phenotyping experiment using full-range hyperspectral data: performance assessment and modelling interpretation
Ramin Heidarian Dehkordi, Mirco Boschetti, Gabriele Candiani, Federico Carotenuto, Carla Cesaraccio, Andrea Genangeli, Beniamino Gioli, Donato Cillis, Marina_Ranghetti

Crop traits monitoring is a fundamental step for controlling crop productivity in the context of precision agriculture and field phenotyping. At present, the usage of hyperspectral data in machine learning regression algorithms (MLRAs) has attracted increasing attention to alleviate the challenges associated with traditional crop trait measurements. However, the performance assessment of such hyperspectral-based MLRA models for crop trait retrievals with respect to the well-known natural variations in either structural or biochemical crop properties remains largely elusive. As such, this experiment was set up to assess whether full-range hyperspectral data, acquired by a handheld spectrometer (Spectral Evolution; 350 – 2500 nm), as inputs to partial least squares regression (PLSR) and random forest (RF) models are capable of modeling different wheat crop traits at the canopy level. The examined crop traits were leaf area index (LAI), canopy water content (CWC), canopy chlorophyll content (CCC), and canopy nitrogen content (CNC). This approach allowed us, as an overarching objective, to compare the performance of the two aforementioned MLRA models while also focusing on the physical interpretation of the modelling results for each particular crop trait.
Overall, PLSR provided remarkably higher accuracy, tested with a cross-validation strategy, as compared to RF for all the crop traits. More precisely, PLSR denoted R2 (resp. nRMSE%) values of 0.72 (11.97), 0.77 (10.89), 0.70 (14.61), and 0.74 (14.38) for LAI, CWC, CCC, and CNC, respectively. All PLSR models indicated robust prediction capability with RPD values greater than 1.4, and amongst them, CWC was found to have excellent prediction performance with an RPD higher than 2. However, RF yielded less predictive models with R2 (resp. nRMSE%) values of 0.59 (14.59), 0.42 (17.42), 0.50 (18.86), and 0.42 (21.41) for LAI, CWC, CCC, and CNC, respectively. RF models for LAI and CCC showed good prediction capabilities (RPD > 1.4), whilst RF models of neither CWC nor CNC were reliable (RPD < 1.4).
In general, RF band importance and PLSR regression coefficient results revealed physically- meaningful and consistent patterns for each specific crop trait. Specific wavelengths at SWIR (1716-1745 nm) and NIR (1057-1120 nm), Green, and the Red-Edge bands respectively showed the highest importance for LAI retrieval. Water absorption regions around 910 nm and 1200 nm as well as the Red-Edge and Visible parts were of higher importance for the retrieval of CWC. The best-performing bands were situated in Red-Edge and Green spectral channels for CCC retrieval. SWIR spectral regions between 1600-1800 nm and 2100-2300 nm appeared to be important (in particular with respect to the other traits) alongside the Red-Edge part of the spectrum to retrieve CNC.
We demonstrated that full-range hyperspectral data in combination with MLRA algorithms can provide accurate estimates of wheat crop traits at the canopy level. The success of utilizing hyperspectral data in MLRA algorithms was further highlighted by the physically-meaningful modelling performances in accordance with the subtle structural and biochemical crop properties. Our results suggest that such spectroscopic hyperspectral-based MLRA approaches could be a powerful tool to accurately monitor crop status throughout the cropping season to improve high-throughput phenotyping activities and to further aid precision agricultural practices.

AIT Contribution
Sala Videoconferenza @ PoliBa
Assessing transferability of Gaussian Process Regression for Canopy Chlorophyll Content and Leaf Area Index estimation from Sentinel-2 data exploiting a multi-site, year and crop dataset
Mirco Boschetti, Carla Cesaraccio, Beniamino Gioli

Authors: Margherita De Peppo, Francesco Nutini, Alberto Crema, Gabriele Candiani, Giovanni Antonio Re, Federico Sanna, Carla Cesaraccio, Beniamino Gioli, Mirco Boschetti

Spatio-temporal estimation of crop bio-parameters (BioPar) is required for agroecosystem management and monitoring. BioPar such as Canopy Chlorophyll Content (CCC) and Leaf Area Index (LAI) contribute to assess plant physiological status and health at leaf and canopy level. Remote sensing provides an effective way to spatially explicitly retrieve CCC and LAI at different spatial and temporal scales. Several studies demonstrated how Machine Learning (ML) techniques outperform traditional empirical approaches based on Vegetation Index in BioPar estimations from RS data. Among the different available algorithms Gaussian processes regression (GPR) is considered promising for LAI and CCC mapping. However, few of these studies have examined the performance of GPR in predicting crop parameters when applied to different site, season and crop typology (i.e. validation using independent dataset). The specific objectives of this study conducted in the framework of E-CROPS project were: (i) develop a transferable GPR algorithm for LAI and CCC estimation by exploiting a robust multi-crop, multi-year and site dataset; (ii) assess GPR BioPar retrieval performance against ground measurements acquired over independent dataset; (iii) compare result with other methods including empirically based VI models and operational product embedded in SNAP. In total, 209 (CCC) and 301 (LAI) observations were used to train GPR models. Then, over the unseen dataset (LAI n=820 and CCC n=305) the GPR was validated. The results showed that for both LAI and CCC GPR retrieval are reliable and comparable with SNAP estimates despite CCC show a consistent underestimation. LAI (CCC) estimation metrics ranges for the different data sets as follows: R2 0.2 to 0.75 (0.2 -0.7) and MAE 0.1 to 0.75 (0.5-3). Overall the results demonstrated the potentiality of GPR machine learning approach in LAI and CCC estimations when a robust training set is exploited, such condition guarantee a spatial-temporal transferability of the developed model. GPR BioPar estimation from Sentinel 2 can produce decametric quasi-weekly quantitative information for crop spatio-temporal monitoring. Such maps are a fundamental input for decision support systems devoted to smart crop management and early warning indication. Many precision agriculture techniques could thus benefit from information generated with ideal quality and frequency for site-specific practices aimed at reducing inputs and improving the use-efficiency of fertilizers.

AIT Contribution
Sala Biblioteca @ PoliBa