Wireless sensor networks can revolutionise soil ecology by providing measurements at temporal and spatial granularities previously impossible. This paper presents our first steps towards fulfilling that goal by developing and deploying two experimental soil monitoring networks at urban forests in Baltimore, MD. The nodes of these networks periodically measure soil moisture and temperature and store the measurements in local memory. Raw measurements are incrementally retrieved by a sensor gateway and persistently stored in a database. The database also stores calibrated versions of the collected data. The measurement database is available to third-party applications through various Web Services interfaces. At a high level, the deployments were successful in exposing high-level variations of soil factors. However, we have encountered a number of challenging technical problems: need for low-level programming at multiple levels, calibration across space and time, and sensor faults. These problems must be addressed before sensor networks can fulfil their potential as high-quality instruments that can be deployed by scientists without major effort or cost.