Search Results (3)

Gaps often occur in eddy covariance flux measurements, leading to data loss and necessitating accurate gap-filling. Furthermore, gaps in evapotranspiration (ET) measurements of annual field crops are particularly challenging to fill because crops undergo rapid change over a short season. In this study, an innovative deep learning (DL) gap-filling method was tested on a database comprising six datasets from different crops (cotton, tomato, and wheat). For various gap scenarios, the performance of the method was compared with the common gap-filling technique, marginal distribution sampling (MDS), which is based on lookup tables. Furthermore, a predictor importance analysis was performed to evaluate the importance of the different meteorological inputs in estimating ET. On the half-hourly time scale, the deep learning method showed a significant 13.5% decrease in nRMSE (normalized root mean square error) throughout all datasets and gap durations. A substantially smaller standard deviation of mean nRMSE, compared with marginal distribution sampling, was also observed. On the whole-gap time scale (half a day to six days), average nMBE (normalized mean bias error) was similar to that of MDS, whereas standard deviation was improved. Using only air temperature and relative humidity as input variables provided an RMSE that was significantly smaller than that resulting from the MDS method. These results suggest that the deep learning method developed here is reliable and more consistent than the standard gap-filling method and thereby demonstrates the potential of advanced deep learning techniques for improving dynamic time series modeling. Full article

Public domain synthetic-aperture radar (SAR) imagery, particularly from Sentinel-1, has widened the scope of day and night vegetation monitoring, even when cloud cover limits optical Earth observation. Yet, it is challenging to combine SAR images acquired at different incidence angles and from ascending and descending orbits because of the backscatter dependence on the incidence angle. This study demonstrates two transformations that facilitate collective use of Sentinel-1 imagery, regardless of the acquisition geometry, for agricultural monitoring of several crops in Israel (wheat, processing tomatoes, and cotton). First, the radar backscattering coefficient (σ⁰) was multiplied by the local incidence angle (θ) of every pixel. This transformation improved the empirical prediction of the crop coefficient (K_c), leaf area index (LAI), and crop height in all three crops. The second method, which is based on the radar brightness coefficient (β⁰), proved useful for estimating K_c, LAI, and crop height in processing tomatoes and cotton. Following the suggested transformations, R² increased by 0.0172 to 0.668, and RMSE improved by 5 to 52%. Additionally, the models based on the suggested transformations were found to be superior to the models based on the dual-polarization radar vegetation index (RVI). Consequently, vegetation monitoring using SAR imagery acquired at different viewing geometries became more effective. Full article

Crop monitoring throughout the growing season is key for optimized agricultural production. Satellite remote sensing is a useful tool for estimating crop variables, yet continuous high spatial resolution earth observations are often interrupted by clouds. This paper demonstrates overcoming this limitation by combining observations from two public-domain spaceborne optical sensors. Ground measurements were conducted in the Hula Valley, Israel, over four growing seasons to monitor the development of processing tomato. These measurements included continuous water consumption measurements using an eddy-covariance tower from which the crop coefficient (K_c) was calculated and measurements of Leaf Area Index (LAI) and crop height. Satellite imagery acquired by Sentinel-2 and VENµS was used to derive vegetation indices and model K_c, LAI, and crop height. The conjoint use of Sentinel-2 and VENµS imagery facilitated accurate estimation of K_c (R² = 0.82, RMSE = 0.09), LAI (R² = 0.79, RMSE = 1.2), and crop height (R² = 0.81, RMSE = 7 cm). Additionally, our empirical models for LAI estimation were found to perform better than the SNAP biophysical processor (R² = 0.53, RMSE = 2.3). Accordingly, Sentinel-2 and VENµS imagery was demonstrated to be a viable tool for agricultural monitoring. Full article