Automatic mapping of planting year for tree crops with Landsat satellite time series stacks

摘要

California's Central Valley faces serious challenges of water scarcity and degraded groundwater quality due to nitrogen leaching. Orchard age is one of the key determinants for fruit and nut production and directly affects consumptive water use and fertilizer demand. However, regional and statewide spatially explicit information on orchard planting years in California is still lacking, despite some attempts to estimate tree ages using multi temporal satellite imagery in other regions. Here we developed a robust detection method to track crop cover dynamics and identify the planting year through time series of Landsat imagery within the Google Earth Engine (GEE) platform. We used the full archive of Landsat data (Landsat-5 TM, Landsat-7 ETM +, and Landsat-8 OLI) from 1984 to 2017 as inputs and automated the GEE workflow for the on-fly-mapping. Preprocessing was initially performed using JavaScript to obtain high quality reflectance and Normalized Difference Vegetation Index (NDVI) time series for each Landsat pixel. Annual maximum NDVI was then aggregated to the orchard level based on the field boundary. Our change detection algorithm incorporated a set of decision rules, including adaptive identification of potential years with robust Z-score thresholds, elimination of false detections based on the post-planting growth curve, and estimation of planting year using the most recent minimum strategy. Our method showed a very high accuracy of estimating tree crop ages, with a R-2 of 0.96 and a mean absolute error of less than half a year, when compared with 142 records provided by almond growers. We further evaluated the accuracy of the statewide mapping of planting years for all fruit and nut trees in California, and found an overall agreement of 89.2%. This automatic cloud-based application is expected to greatly strengthen our ability to forecast yield dynamics, estimate water use and fertilizer inputs, at individual field, county and statewide basis.