Conference Agenda

.

Plant stress, a persistent natural phenomenon, presents a significant threat to global agriculture and food security. As changing climate patterns lead to more frequent and severe drought, pest, or disease occurrences, the demand for innovative and precise approaches to evaluate and monitor plant stress in agriculture grows. In this context, a new generation of thermal remote sensing data emerges as a valuable tool. It has the potential to provide the necessary data not only to analyze and understand the impact of stress caused by different factors on crops but also to detect it promptly, allowing for timely mitigation.

constellr develops a constellation of state-of-the-art high-resolution thermal infrared (TIR) as well as visible and near-infrared (VNIR) sensors, planned for launch by the end of 2024, to monitor land surface temperature (LST). The HiVE (High-precision Versatile Ecosphere monitoring mission) constellation comprises micro-satellites in the 100 kg class, with orbits in a sun-synchronous plane at an altitude of 550 kilometers. With a remarkable 1-day temporal resolution reached starting 2026 with 5 satellites in orbit, 30 meters spatial resolution for the TIR bands, and up to 10m for the VNIR bands, HiVE is uniquely equipped to provide accurate and timely data optimized for agricultural needs.

We will present the technical specification, status of the HiVE mission, and the added value of these data for agricultural practice. Thus, constellr is currently performing different proof of concept studies to quantify the added value of thermal data for identifying plant stress. Firstly, LST time-series are analyzed to detect trends and anomalies as a proxy for crop stress in time and space. Secondly, LST data are exploited to derive actual evapotranspiration, which in turn is used to quantify the drought stress. Besides that, an outlook on further LST-based products relevant to agriculture based on LST will be shown.

In the coming years, additional hyperspectral missions, such as the Copernicus CHIME, will increment operationally the data stream already provided by the ASI PRISMA and DLR EnMap missions. This will enable new research possibilities within the “agriculture and food security” domain. In the agri-food sector, hyperspectral data, characterised by narrow bands covering the full range from VIS to SWIR, can provide a unique contribution to better i) estimate within-season crop traits and ii) quantity crop residue presence after harvesting. The retrieval of within-season crop traits allows early warning indications of potential stress, supports smart agriculture practices within a Precision Farming framework, and improves yield estimates. The identification and quantification of Non-Photosynthetic Vegetation (NPV) are fundamental to track sustainable agro-practices for soil conservation (e.g. minimum tillage) and to provide information for the carbon budget in agriculture. In this framework the ASI-PRISMASCIENZA project “PRIS4VEG” exploited a comprehensive multiyear PRISMA dataset (2020 - 2023) together with field bio-parameter measurements and ancillary farm data (e.g., crop sequence and agro-practices) acquired in Jolanda di Savoia site (North of Italy).

A hybrid approach, fine-tuned with an active learning procedure (HAL), was successfully tested on PRISMA hyperspectral data to estimate crop traits, such as leaf area index (LAI), chlorophyll and nitrogen content at both leaf (LCC, LNC) and canopy (CCC, CNC) levels. A machine learning regression algorithm (MLRA), based on enhanced hyperspectral input identified by spectroscopic modelling of diagnostic NPV cellulose-lignin specific absorption features, was used to assess the presence and cover of crop residues. The MLRA was trained using an extensive and well-documented spectral library and tested on independent ground, airborne and spaceborne (i.e., PRISMA) data. PRISMA maps provided interesting spatio-temporal patterns related to Genetics-Environment-Management interactions, demonstrating the contribution of hyperspectral data in generating spatially explicit information for the agro-monitoring sector.

The NISAR mission, a joint effort between NASA and the Indian Space Research Organization (ISRO) will be launching in the spring of 2024. Once through its commissioning and calibration/validation periods, the mission will be collecting dual-polarized L-band SAR data at a 20 m ground-projected resolution and 240 km swath, two times every 12 days over most land surfaces. In addition to these background observations the mission will be collecting S-band data inside of India, along California's west coast, and over Ecosystems cal/val sites located worldwide, many of which are derived from the Group on Earth Observation's Joint Experiment for Crop Assessment and Monitoring (JECAM).

Among the agriculture applications that NISAR will be addressing through its systematic observing strategy will be in the determination of active crop area that will be produced every quarter. Early studies using L-band SAR through the Japanese Aerospace Exploration Agency's ALOS sensor, or NASA's UAVSAR, and ESA's Sentinel-1 SARs, has show that time series of SAR data can be used for crop classification as well.

In this talk, we will discuss the status of the NISAR mission and highlight its development for agriculture applications and the larger remote sensing community.

Managing dwindling water resources responsibly and protecting crops from adverse growing conditions in the face of climate change are key challenges for sustainable and robust agriculture everywhere in the world. Thermal infrared remote sensing to measure canopy and soil temperatures in the field has a tremendous potential to directly track the water cycle in crops and soil, and directly monitor crop stress. In contrast to thermal infrared, visible and near-infrared spectral bands only detect crop stress too late, once irreversible damage is already done.

Hydrosat is launching a thermal infrared satellite constellation to provide daily, global, and high-resolution land surface temperature measurements targeting a spatial resolution of 50 m. We currently employ multi-sensor data fusion techniques and sophisticated numerical water balance models to bridge the gap until Hydrosat’s full constellation is fully operational.

IrriWatch is Hydrosat’s irrigation management decision support system, combining empirical and biophysics-based models with daily satellite data. IrriWatch allows growers to track the water demand and growth progress of their crops down to individual pixels in near-real-time, enabling cost savings and yield increases by optimizing irrigation and fertilization operations.

We present recent validation studies comparing output from the IrriWatch algorithms to ground truth data obtained from in-field sensors and drone flights, showcasing the capabilities and limitations of monitoring agricultural fields from space. For example, while we find that thermal sharpening works well under suitable conditions, it cannot replace native high-resolution data. From space imagery we observe excellent agreement between predicted and actual dry matter accumulation, even compared to high-resolution yield maps obtained from combine harvesters. IrriWatch also typically provides realistic estimates of root zone soil moisture from space, including forecasts of irrigation water demand and monitoring of total applied water. We find that water balance calculations are mostly limited when soil parameters and rainfall data are inaccurate, and we present recent developments to overcome these limitations.

.