**1. Introduction**

Planet Labs currently operates more than 200 PlanetScope satellites in sun-synchronous orbits and frequently launches new satellites that are designed to have a short operational lifetime (<4 years). The PlanetScope satellite constellation enables near-daily monitoring with multi-spectral imagery at high spatial resolution (3 m) [1]. PlanetScope imagery has been applied to a variety of studies to monitor phenomena that require both high spatial and temporal resolution, for instance, to monitor small water bodies [2–4], estimate methane emissions from forested wetlands [5], assess river-ice and water velocity [6], improve

**Citation:** Perin, V.; Roy, S.; Kington, J.; Harris, T.; Tulbure, M.G.; Stone, N.; Barsballe, T.; Reba, M.; Yaeger, M.A. Monitoring Small Water Bodies Using High Spatial and Temporal Resolution Analysis Ready Datasets. *Remote Sens.* **2021**, *13*, 5176. https:// doi.org/10.3390/rs13245176

Academic Editors: Alban Kuriqi and Luis Garrote

Received: 17 November 2021 Accepted: 13 December 2021 Published: 20 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

crop leaf-area-index estimation with sensor data fusion [7–9], and monitor near-real-time aboveground carbon emissions from tropical forests [10–12].

A recent global analysis of PlanetScope's temporal availability [13] showed that the annual and monthly number of PlanetScope observations does not vary uniformly across the globe. The authors attributed this finding to different PlanetScope orbits (i.e., altitude and inclinations), due to different numbers of sensors in orbit, which vary when PlanetScope satellites are decommissioned and replaced with new sensors, and due to images that cannot be geolocated [13]. In addition, it is well known that the number of observations from optical wavelength satellite imagery will vary according to dynamic and global cloud obscuration. While the PlanetScope cloud mask, Usable Data Mask 2 (UDM2), is available [1] and allows for discernment of classes like cloud, cloud shadow, and heavy, haze among others, its accuracy has not been thoroughly assessed [13–15] and it is not available for images prior to 2018 [1]. Aiming to overcome these limitations—irregular cadence and cloud obscuration—and to increase the applications of PlanetScope imagery, Planet Labs has focused on developing the next generation of tiled analysis ready datasets— Basemap [16] and Planet Fusion [17]—which are less affected by the presence of clouds and are set for a fixed temporal cadence.

Basemap is generated by mosaicking the whole or part of the highest quality PlanetScope imagery, which is selected based on cloud cover and image acutance (i.e., sharpness). For example, for a given period of interest—Basemap can be processed using different image cadences, e.g., daily, weekly, biweekly—PlanetScope images are ranked based on these metrics such that cloud-free images have higher scores than cloudy images [16,18]. Basemap is designed to monitor changes over time and for analytics-driven use cases, and it has been applied to several research projects, including monitoring of forest biomass [10–12], to assess carbon emissions from drainage canals [19], and to monitor coral reef map probabilities [20]. Planet Fusion, on the other hand, is based on the CubeSatenabled spatiotemporal enhancement method [8], and it leverages the high spatial and temporal resolution provided by PlanetScope scenes with rigorously calibrated publicly available multispectral satellites (i.e., Sentinel-2, Landsat, MODIS, and VIIRS) to provide daily and radiometrically consistent and gap-filled surface-reflectance images that are free of clouds and shadows [17]. Planet Fusion is suitable to assess inter-day changes, for timeseries analysis, and monitoring of disturbances of Earth's surface. Recently, Planet Fusion has been applied to monitor crop phenology, using the normalized difference vegetation index and leaf area index [21,22]. Given that these datasets are cloud-free and processed to have daily cadence at high spatial resolution—both Basemap and Planet Fusion have 3 m pixel size—they provide an unprecedented opportunity to improve the monitoring of dynamic small water bodies, for instance, on-farm reservoirs (OFRs) that are used by farmers to store water during the wet season and for crop irrigation during the dry season. OFRs have a dynamic surface area time series, especially during the crop-growing season, when farmers are irrigating their crops and may pump water from nearby streams [23–25].

There are more than 2.6 million OFRs in the USA alone, and these OFRs play a key role in surface hydrology by storing fresh water and as an essential component of global irrigation activities [26–28]. Nonetheless, OFRs can contribute to downstream water stress by decreasing stream discharge and peak flow in the watersheds where they are built [24,29,30]. Therefore, monitoring OFRs sub-weekly surface area changes is critical to the assessment of their seasonal and inter-annual variability, as well as to mitigation of their downstream impacts, with implications concerning how OFRs are managed and where they are built. Previous research assessed the spatial and temporal variability of OFRs by leveraging the long-term (≥25 years) Landsat-based inundation datasets [23,31,32]. Nonetheless, these datasets are limited to a few annual observations—due to clouds, sensor issues, and the 16-day repeat cycle—and Landsat's spatial resolution (30 m) limits the applications of these datasets to monitor OFRs smaller than 5 ha (i.e., high surface area uncertainties ~ 20%). Aiming to overcome these limitations, other studies [4,33,34] have applied a multi-sensor satellite imagery approach, including sensors of higher spatial and

temporal resolution (e.g., PlanetScope [3 m] and Sentinel-2 [10 m]) when compared to Landsat. However, a multi-sensor approach requires processing of satellite imagery of different spatial resolution from multiple platforms, which can be time-consuming and a limiting factor if it is necessary to process, download, and move the satellite imagery across multiple platforms [33]. In this study, we propose a novel use of the analysis ready datasets Basemap and Planet Fusion, and we aim (1) to assess the usefulness of both datasets to monitor OFRs sub-weekly surface area changes and (2) to compare the two datasets and describe their differences when used to monitor OFRs.

### **2. Methods**
