In a BARD-funded research project, researchers from ARO, Dr. Tarin Paz-Kagan, and Dr. Or Sperling, together with Prof. Maciej Zwieniecki from UC Davis, are using big data to build decision-support applications to help growers cope with climate shifts. The diagnostic approach is based on an analysis of seasonal changes that will give growers a snapshot of environmental conditions and orchard performance. Big data analysis based on remote sensing imageries (ground cameras, UAVs, and satellites), environmental conditions, and orchard performance will help create models to predict tree-phenology. Large spatial scale predictions will consider the effects of future climate on deciduous tree-crops and derive best farming practices to ensure yields.
Researchers are working to develop a diagnostic approach based on seasonal changes in the carbohydrate content in trees. They are targeting the temperature-kinetics (C-T model) of specific enzymatic groups that regulate carbohydrate composition at the cellular level. In order to accelerate the investigation, they wish to establish the detection of carbohydrates in woody tissues by near-infrared spectroscopy (NIRS). They will calibrate it to their research conditions by enzymatic analyses. Other factors, including management practices, climate shifts, genetics, grafting, age, and farming practices, will be included in the analysis. Finally, they will monitor tree phenology and temperature conditions at a high spatial and temporal resolution based on earth-observations of satellites to calibrate models and test the feasibility of this large-scale approach for orchard management.
Prior to their BARD research project, the researchers generated a big data-set of carbohydrate concentrations in three tree-crops (almond, pistachio, and walnut) that span some 400 farms across the entire California Central Valley (i.e., the carbohydrate observatory). This survey represents the entire geographical distribution of profitable plantations in California. Researchers are now working to scale up these observations by the development of high-throughput tools for carbohydrate analysis. They have also developed a novel mechanistic approach that explains carbohydrate-dynamics in dormant trees by variable temperature conditions. It is a novel physiological framework that projects inter-annual variability in bloom time. They will validate the model and parameterize it to new conditions by earth-observations from satellites with high revisit time over orchards in California and Israel. Once completed, this research could breach the gap between fundamental science and farming applications by developing decision-support applications for mitigating climate shifts.
Dr. Tarin Paz-Kagan elaborated on the teams work:
What big data is being collected?
The data that is being collected include, Carbohydrate observatory data that will also be detected by NIRS spectroscopy, Phenology observation by aerial images using time-lapse cameras, high-resolution UAV data, satellite data to study tree phenology on orchards in Israel and California and hourly temperatures and daily min and max data in Israel and California.
How will it be analyzed?
The Carbohydrate Observatory data includes temporal-density of carbohydrate concentrations in orchard trees across an extensive geographic reference with additional data on age, cultivar, and orchard management of the trees. These will be analyzed using high-throughput NSC detection to pave the way for our applications into precise orchard management in commercial scales. We will establish a cloud-based repository from all of our former results to evaluate the high throughput NIRS application for NSC analysis.
The C-T model is based on a mechanistic link between bloom time and the thermal conditions during dormancy. This mechanistic link will be tested by manipulating carbohydrate distribution by temperature. Large-scale and high-resolution temperature and phenology monitoring by aerial imagery will be developed by integrating data from time-leap cameras, high-resolution UAV data, satellite data (i.e., Planet and Sentinel 2) and will be analyzed by object-based recognition using a convolution neural network.
In the long term, how do the researchers imagine this could work for farmers?
We designed research to mitigate climate shifts’ challenges for the Californian and Israeli deciduous tree-crops industry. We recognized a critical knowledge gap concerning the restraints of tree-crops to produce in abnormal winter conditions. Hence, the research will establish three new tools: (1) High-throughput and large-scale phenology recordings (based on satellite imagery), (2) Diagnoses of trees’ dormant-physiology chemometrics), and (3) A physiological framework to link physiological changes to environmental conditions (based on eco-physiological modeling). Initially, we will determine the limiting factors to productivity in multiple climate conditions. Accordingly, our new tools will guide commercial farmers to maximize pollination, select cultivars, and adjust irrigation and fertilization to optimize dormancy and bloom in various climates. Second, we will associate novel phenotypes of winter metabolism and phenology to yield. Then, our new approach could integrate into deciduous tree crops’ breeding programs to promote climate resilience.