Characterization of Planetscope-0 Planetscope-1 surface reflectance and normalized difference vegetation index continuity

Abstract

Planetscope data are acquired from a constellation of low-cost satellites to provide 3 ​m red, green, blue, and near infrared (NIR) data with near-daily global coverage. Differences in the spectral characteristics of the different Planetscope sensor generations imply that they may not provide consistent reflectance time series needed for certain quantitative analyses. This study provides a comprehensive comparison of the surface reflectance and derived normalized difference vegetation index (NDVI) differences between the first two generations of Planetscope sensors, known as Planetscope-0 and Planetscope-1. More than 9100 Planetscope images, acquired with cloud cover <30% for different months at 500 locations across Africa were compared. Histograms of the Planetscope image overpass time differences over the 500 locations provide insights into the temporal intervals between successive images. The most frequent image overpass difference was one day for Planetscope-0, the same day for Planetscope-1, and one day considering both sensors. The overpass time difference between images sensed on the same day varied from <1 ​min to 124 ​min, and 64% of the Planetscope-0 and Planetscope-1 overpass time differences were sensed >60 ​min apart and 25% were sensed ≤30 ​min apart. Transformation functions were developed and the coefficients are provided so that the user community can adjust the surface reflectance and NDVI from the Planetscope-0 and Planetscope-1 sensors to each other to provide more consistent 3 ​m time series. This was undertaken by statistical comparison of 9.9 million contemporaneous quality filtered Planetscope-0 and Planetscope-1 observations sensed on the same day within 30 ​min (average 11.8 ​min) and ≤5° solar zenith difference. Two linear regression models, reduced major axis (RMA) and Theil-Sen, were used to provide a general depiction of reflectance and NDVI differences between the two sensor generations, and to develop the transformation functions between them, respectively. The Planetscope-0 blue surface reflectance was on average greater than the Planetscope-1 blue surface reflectance by 20.1%, whereas the Planetscope-0 red, green and NIR surface reflectance were 4.6%, 5.1%, and 12.1% less than the comparable Planetscope-1 bands respectively. The Planetscope-0 atmospherically corrected NDVI was typically about 3.7% smaller than the Planetscope-1 atmospherically corrected NDVI. These differences are likely primarily due to the different spectral response functions and bandwidths of the two sensor generations. This is supported by our finding that the surface reflectance from the three Planetscope-0 visible band values were more highly correlated with each other than for the Planetscope-1 visible bands. To demonstrate that the transformations performed correctly, the Planetscope-0 surface reflectance and NDVI data were transformed to Planetscope-1 equivalent values, and vice versa, and the temporal consistency of monthly time series containing observations from both sensors were evaluated using a temporal smoothness index. The results of this adjustment demonstrated that the temporal consistency of monthly time series containing Planetscope-0 and Planetscope-1 observations were improved for all bands and for the NDVI.