High-resolution weather network reveals a high spatial variability in air temperature in the Central valley of California with implications for crop and pest management.

Abstract

Weather is the most important driver of crop development. However, spatial variability in weather makes it hard to obtain reliable high resolution datasets across large areas. Most growers rely on data from a single station that can be up to 50km away to make decisions about irrigation, pest management and penology-associated cultural practices at the block level. In this regard, we hypothesize that kriging a large network of weather stations can improve thermal time data quality compared to using the closest station. This study aims to explore the spatial variability in California's Central Valley and what is the relationship between the density of weather stations used and the error in the measurement of temperature related metrics and derived models. For this purpose, we used temperature records from January 1st 2020 to March 1st 2021 collected by the California Irrigation Management Information System (CIMIS) and a system of 731 weather stations placed above the canopy of trees in commercial orchards (in-orchard). We observed large discrepancies (>300 GDDTb0) in thermal time accumulation between using an interpolation of all stations available and just using the closest CIMIS station. Our data suggests these differences are not systematic bias but true differences in mesoclimate. Similar results were observed for chill accumulation in areas especially prone to not meeting pistachio chill requirements where the discrepancies between using the site-specific in-orchard weather station network and not using them were up to 10 CP. The use of this high resolution network of weather stations revealed spatial patterns in grape, almond, pistachio and pests phenology not reported before. Whereas previous studies have been focused on predictions at the county or state or regional level, our data suggests that a finer resolution can result in major improvements in the quality of data crucial for crop decision making.