1H. David, 2Y. Cohen, 3I. Bahat, 4Y. Netzer, 5A. Peeters, 6A. Ben-Gal
1. Institute of Agricultural Engineering, Agricultural Research Organization (ARO), Israel; Dept. of Geography and Environment, Bar-Ilan University, Israel
2. Institute of Agricultural Engineering, Agricultural Research Organization (ARO), Israel
3. Agricultural Research Organization (ARO), Israel; Faculty of Agriculture, Food and Environment, The Hebrew University of Jerusalem, Israel; Eastern Research and Development Center, Israel
4. Agriculture Research Department, Eastern Research and Development Center, Israel
5. TerraVision Lab, Israel
6. Institute of Soil Water and Environmental Sciences, Gilat Center, Agricultural Research Organization (ARO), Israel
High-resolution thermal images (TI) coupled with measured atmospheric conditions have been utilized to map within-field water status variability in vineyards as well as in other crops. Spaceborne TIs, though, have too coarse spatial resolution and using aerial platforms with high spatial-resolution TI sensors require substantial financial investments, which inhibit their large-scale adoption. To overcome this spatial resolution vs spatial extent trade-off, we propose to increase the spatial extent and temporal resolution of TIs acquired from unmanned aerial vehicles (UAV) by incorporating higher-temporal-lower-spatial-resolution satellite images in the VIS-NIR range. In this approach, it is suggested to correlate between thermal-based indices and vegetation indices (VIs). While VIs are not directly correlated to water status, linear correlations between VIs and water status exist provided that the analyzed scene includes both pixels representing high water deficits and well-watered conditions. The first step towards this end is presented in several vineyards in Israel.
A 2.4 ha highly variable vineyard in Central Israel was divided into 20 management cells (30X30m). During the 2017 growing season, stem water potential (SWP) measurements were measured from 5-6 vines in each management cell every 1-2 weeks. Four UAV thermal images were acquired along the growing season to map the crop water stress index (CWSI) of the vineyard. Daily clear-sky scenes for the annual growth cycle of grapevines (May to October) in 2017 were acquired from Planet Lab’s Earth observing nano-satellites constellation for vineyards in Israel. Planet “Doves” provide imagery on a daily time step using 4-channel in the VIS-NIR range at ~3 m spatial resolution. Daily 3-m NDVI time series were built from Planet Doves scenes using the maximum value criteria when several images were available for a single date. To avoid noisy data due to multi-sensor uncertainties and to fill existing gaps in the time series, the local scatterplot smoothing technique (LOESS) was applied. The breaks for additive seasonal and trend (BFAST) R package was then used to detect the time, number and direction of changes in the NDVI time series.
Results show that Planet’s NDVI time series was highly correlated with measured SWP in the vineyard (R2 = 0.80; n=12 dates). Three main phenological phases during the growth season were observed, with a flat plateau, steep decrease and a moderate decrease in both NDVI and SWP. Spatial variability in mean annual NDVI was comparable with that of the mean annual SWP in the vineyard. Lower but a significant correlation was found between end-of-season NDVI and SWP at the cell-level (R2 = 0.42, n=20).
Ongoing analysis includes 30-50 additional vineyards in Israel with following tasks: (1) correlating VIs time series derived from Planet Lab’s/Rapid-Eye/Sentinel2 satellites with weekly SWP measurements in the various vineyrads; and (2) correlating between measured SWP, satellite-derived VIs and UAV-based thermal derived CWSI maps. These correlation can then be used to derive high spatial-large extent water status maps along the grapevine growing season and utilized for precision irrigation.
