The CSU Chico Geographical Information Center, Valley Vision, USDA Rural Development and the San Francisco Bay Area Chapter of the Internet Society collaborated on a research project focused on mobile broadband availability in farm fields to examine two questions:

• What is the existing service to support precision agriculture needs in farm fields themselves as opposed to nearby public roads where testing typically occurs?
• How does increasing granularity of testing mobile broadband speeds affect eligibility under the 4 Mbps download/1 Mbps upload threshold for the USDA Rural Utilities Service Community Connect Grant program?

Using the CalSPEED mobile broadband measurement tool developed by CSU Monterey Bay for the California Public Utilities Commission, the project incorporated two phases of testing. Phase 1 focused on a more granular identification of performance characteristics across the rural areas of Yolo County, as compared to the testing grid for the CPUC testing program. In March of 2017, the GIC conducted tests at 155 sites in the areas of Yolo County. The sampling locations were all located on public roads and were intended to identify areas for further testing at the farm field level. Phase 1 identified three areas of low broadband performance, along Highway 16 in western Yolo County (the Capay Valley), along the southern border of Yolo County between Winters and Davis and in the southeastern part of the county in the Sacramento-San Joaquin Delta near Courtland. Phase 2 implemented testing on seven farms within the fields themselves, with five farms located in areas predicted to be low performing by Phase 1 and two farms located in areas predicted to be high performing by Phase 1.

The results indicate a number of important aspects of mobile broadband service in the farm fields. First, for two of the low performance area farms, no single provider offers suitable mobile coverage under California’s standard. These farmers cannot access mobile-based precision agriculture technologies for all of their property unless they establish multiple subscriptions. Given that the adoption rate for sensor-based technologies that can reduce water and chemical input is already so low, these types of barriers only further limit adoption. Second, locations without LTE coverage are more likely to be unserved.

Statistical analysis of the results show that interpolation used for the statewide and Phase 1 testing overstates the availability of mobile speeds meeting the 4 Mbps threshold for the USDA Community Connect Grant program. For the two providers that are typically considered to provide the most robust rural service (AT&T and Verizon), the statewide testing predicted much higher speeds than are actually available for the 0 – 4 Mbps tier. The Phase 1 countywide testing reduced the error for the 0 – 4 Mbps tier in all cases, but the error was still biased to over- predicting speeds. The variation between statewide results, countywide Phase 1 results and farm field Phase 2 results indicate that the USDA eligibility threshold for Community Connects grant funding is much more likely to be accurate at the density used for the Phase 1 testing than at the CPUC statewide testing density.