Sowing date detection at the field scale using CubeSats remote sensing

Yuval Sadeh, Xuan Zhu, Karine Chenu, David Dunkerley

Research output: Contribution to journalArticleResearchpeer-review

32 Citations (Scopus)


Sowing dates have a great influence on crop yields as they affect what environmental conditions the plants will experience. Therefore, sowing dates are important to many individuals and organizations on the food production chain, including food manufacturers and traders. Despite their importance, large-scale comprehensive data on sowing dates are currently rare and often available only as broad estimates at the state, county or district level. Continuously obtaining sowing dates at the field scale is expensive, time-consuming and prone to human errors. Remote sensing on the other hand has the potential of conducting rapid, cost-effective and continuous surveys of farm management practices over large scales. Over the last decade, a new era in Earth observation satellites began with the production and deployment of smaller, lighter and cheaper nano-satellites known as CubeSats. Images from these satellites can potentially be used to capture sowing dates over time. In this study, we developed an innovative semi-automated sowing date detection methodology, which uses high spatio-temporal resolution CubeSat images to detect sowing dates at the field scale. We identified the sowing dates by using Planet's PlanetScope data to detect changes on the fields’ surface caused by no-tillage sowing. Our approach overcame the signal inconsistences existing among the numerous sensors in the constellation, and detected 85% of the sown fields with R2 = 0.99. We succeeded to identify the actual sowing dates of individual fields with a median gap of 0 days within an unparalleled RMSE of 0.9 and 1.9 days in a set of national trials and in fields of a commercial farm, respectively. The methodology presented in this study can be used to produce and update field level crop statistics on a near daily-basis at a low cost. It is robust and simple and can be applied over a wide range of soil types, atmospheric conditions, crop types and sensors, to detect sowing dates in regions where no-tillage sowing is practiced.

Original languageEnglish
Pages (from-to)568-580
Number of pages13
JournalComputers and Electronics in Agriculture
Publication statusPublished - 1 Feb 2019


  • Change detection
  • CubeSats
  • Farming management
  • Planting date
  • Satellite image
  • Sowing dates

Cite this