Countrywide Mapping of Plant Ecological Communities with 101 Legends including Land Cover Types for the First Time at 10 m Resolution through Convolutional Learning of Satellite Images

This paper presents next-generation mapping of plant ecological communities including land cover and agricultural types at 10 m spatial resolution countrywide. This research introduces modelling and mapping of land cover and ecological communities separately in small regions-of-interest (prefecture level), and later integrating the outputs into a large scale (country level) for dealing with regional distribution characteristics of plant ecological communities effectively. The Sentinel-2 satellite images were processed for cloud masking and half-monthly median composite images consisting of ten multi-spectral bands and seven spectral indexes were generated. The reliable ground truth data were prepared from extant multi-source survey databases through the procedure of stratified sampling, cross-checking, and noisy-labels pruning. Deep convolutional learning of the time-series of the satellite data was employed for prefecture-wise classification and mapping of 29–62 classes. The classification accuracy computed with the 10-fold cross-validation method varied from 71.1–87.5% in terms of F1-score and 70.9–87.4% in terms of Kappa coefficient across 48 prefectural regions. This research produced seamless maps of 101 ecological communities including land cover and agricultural types for the first time at a country scale with an average accuracy of 80.5% F1-score.

Assessing accuracy of GAP and LANDFIRE land cover datasets in winter habitats used by greater sage-grouse in Idaho and Wyoming, USA

Remotely sensed land cover datasets have been increasingly employed in studies of wildlife habitat use. However, meaningful interpretation of these datasets is dependent on how accurately they estimate habitat features that are important to wildlife. We evaluated the accuracy of the GAP dataset, which is commonly used to classify broad cover categories (e.g., vegetation communities) and LANDFIRE datasets, which classifies narrower cover categories (e.g., plant species) and structural features of vegetation. To evaluate accuracy, we compared classification of cover types and estimates of percent cover and height of sagebrush (Artemisia spp.) derived from GAP and LANDFIRE datasets to field-collected data in winter habitats used by greater sage-grouse (Centrocercus urophasianus). Accuracy was dependent on the type of dataset used as well as the spatial scale (point, 500-m, and 1-km) and biological level (community versus dominant species) investigated. GAP datasets had the highest overall classification accuracy of broad sagebrush cover types (49.8%) compared to LANDFIRE datasets for narrower cover types (39.1% community-level; 31.9% species-level). Percent cover and height were not accurately estimated in the LANDFIRE dataset. Our results suggest that researchers must be cautious when applying GAP or LANDFIRE datasets to classify narrow categories of land cover types or to predict percent cover or height of sagebrush within sagebrush-dominated landscapes. We conclude that ground-truthing is critical for successful application of land cover datasets in landscape-scale evaluations and management planning, particularly when wildlife use relatively rare habitat types compared to what is available.

The effects of fire on the thermal environment of sagebrush communities

Thermal heterogeneity provides options for organisms during extreme temperatures that can contribute to their fitness. Sagebrush (Artemisia spp.) communities exhibit vegetation heterogeneity that creates thermal variation at fine spatial scales. However, fire can change vegetation and thereby variation within the thermal environment of sagebrush communities. To describe spatial and temporal thermal variation of sagebrush communities following wildfire, we measured black bulb temperature (T_bb) at 144 random points dispersed within unburned and burned communities, for 24-h at each random point. We observed a wide thermal gradient in unburned (-7.3° to 63.3 °C) and burned (-4.6° to 64.8 °C) sagebrush communities. Moreover, unburned and burned sagebrush communities displayed high thermal heterogeneity relative to ambient temperature (T_air). Notably, T_bb varied by 47 °C in both unburned and burned communities when T_air was 20 °C. However, fire greatly reduced the buffering capacity and thermal refuge of Wyoming big sagebrush (A. tridentata wyomingensis) communities during low and high T_air. Furthermore, fire increased T_bb in Wyoming big sagebrush and mountain big sagebrush (A. t. vaseyana) during the mid-day hours. These results demonstrate how fire changes the thermal environment of big sagebrush communities and the importance of shrub structure which can provide thermal refuge for organisms in burned communities during extreme low and high T_air.

Characterizing Land Surface Phenology and Exotic Annual Grasses in Dryland Ecosystems Using Landsat and Sentinel-2 Data in Harmony

Invasive annual grasses, such as cheatgrass (Bromus tectorum L.), have proliferated in dryland ecosystems of the western United States, promoting increased fire activity and reduced biodiversity that can be detrimental to socio-environmental systems. Monitoring exotic annual grass cover and dynamics over large areas requires the use of remote sensing that can support early detection and rapid response initiatives. However, few studies have leveraged remote sensing technologies and computing frameworks capable of providing rangeland managers with maps of exotic annual grass cover at relatively high spatiotemporal resolutions and near real-time latencies. Here, we developed a system for automated mapping of invasive annual grass (%) cover using in situ observations, harmonized Landsat and Sentinel-2 (HLS) data, maps of biophysical variables, and machine learning techniques. A robust and automated cloud, cloud shadow, water, and snow/ice masking procedure (mean overall accuracy >81%) was implemented using time-series outlier detection and data mining techniques prior to spatiotemporal interpolation of HLS data via regression tree models (r = 0.94; mean absolute error (MAE) = 0.02). Weekly, cloud-free normalized difference vegetation index (NDVI) image composites (2016–2018) were used to construct a suite of spectral and phenological metrics (e.g., start and end of season dates), consistent with information derived from Moderate Resolution Image Spectroradiometer (MODIS) data. These metrics were incorporated into a data mining framework that accurately (r = 0.83; MAE = 11) modeled and mapped exotic annual grass (%) cover throughout dryland ecosystems in the western United States at a native, 30-m spatial resolution. Our results show that inclusion of weekly HLS time-series data and derived indicators improves our ability to map exotic annual grass cover, as compared to distribution models that use MODIS products or monthly, seasonal, or annual HLS composites as primary inputs. This research fills a critical gap in our ability to effectively assess, manage, and monitor drylands by providing a framework that allows for an accurate and timely depiction of land surface phenology and exotic annual grass cover at spatial and temporal resolutions that are meaningful to local resource managers.

Validating a Landsat Time-Series of Fractional Component Cover Across Western U.S. Rangelands

Western U.S. rangelands have been quantified as six fractional cover (0%–100%) components over the Landsat archive (1985–2018) at a 30 m resolution, termed the “Back-in-Time” (BIT) dataset. Robust validation through space and time is needed to quantify product accuracy. Here, we used field data collected concurrently with high-resolution satellite (HRS) images over multiple locations (n = 42) and years. Field observations were used to train regression tree models, predicting the component cover across each HRS image. Our objectives were to evaluate the spatial and temporal relationships between HRS and BIT component cover and compare spatio-temporal climate responses. First, for each HRS site-year (n = 77) we averaged both the HRS and BIT predictions within each site separately and regressed the averages to quantify the temporal accuracy. Next, we regressed individual pixel values of corresponding HRS and BIT predictions to quantify the spatio-temporal accuracy. Results showed strong temporal correlations with an average R2 of 0.63 and Root Mean Square Error (RMSE) of 5.47% as well as strong spatio-temporal correlations with an average R2 of 0.52 and RMSE of 7.89% across components. Our approach increased the validation sample size relative to direct comparison of field observations. Validation results showed robust spatio-temporal relationships between HRS and BIT data, providing increased user confidence in the data.