Detailed maps of forest structure attributes are crucial for sustainable forest management, conservation, and forest ecosystem science at the landscape level. Mapping the structure of broad heterogeneous forests is chal lenging, but the integration of extensive field inventory plots with wall-to-wall metrics derived from synthetic aperture radar (SAR) and optical remote sensing offers a potential solution. Our goal was to map forest structure attributes (diameter at breast height, basal area, mean height, dominant height, wood volume and canopy cover) at 30-m resolution across the diverse 463,000 km2 of native forests of Argentina based on SAR Sentinel-1, vegetation metrics from Sentinel-2 and geographic coordinates. We modelled the forest structure attributes based on the latest national forest inventory, generated uncertainty maps, quantified the contribution of the predictors, and compared our height predictions with those from GEDI (Global Ecosystem Dynamics tion) and GFCH (Global Forest Canopy Height). We analyzed 3788 forest inventory plots (1000 m2 each) from Argentina’s Second Native Forest Inventory (2015–2020) to develop predictive random forest regression models. From Sentinel-1, we included both VV (vertical transmitted and received) and VH (vertical transmitted and horizontal received) polarizations and calculated 1st and 2nd order textures within 3 ×3 pixels to match the size of the inventory plots. For Sentinel-2, we derived EVI (enhanced vegetation index), calculated DHIs (dynamic habitat indices (annual cumulative, minimum and variation) and the EVI median, then generated 1st and 2nd order textures within 3 ×3 pixels of these variables. Our models including metrics from Sentinel-1 and 2, plus latitude and longitude predicted forest structure attributes well with root mean square errors (RMSE) ranging from 23.8% to 70.3%. Mean and dominant height models had notably good performance presenting relatively low RMSE (24.5% and 23.8%, respectively). Metrics from VH polarization and longitude were overall the most important predictors, but optimal predictors differed among the different forest structure attributes. Height predictions (r =0.89 and 0.85) outperformed those from GEDI (r =0.81) and the GFCH (r =0.66), suggesting that SAR Sentinel-1, DHIs from Sentinel-2 plus geographic coordinates provide great opportunities to map multiple forest structure attributes for large areas. Based on our models, we generated spatially-explicit maps of multiple forest structure attributes as well as uncertainty maps at 30-m spatial resolution for all Argentina’s native forest areas in support of forest management and conservation planning across the country.
File: Silveira_RSE_2022.pdf
Soil organic carbon (SOC) content supports several ecosystem services. Quantifying SOC requires: (i) accurate C estimates of forest components, and (ii) soil estimates. However, SOC is difficult to measure, so predictive models are needed. Our objective was to model SOC stocks within 30 cm depth in Patagonian forests based on climatic, topographic and vegetation productivity measures from satellite images, including Dynamic Habitat Indices and Land Surface Temperature derived from Landsat-8. We used data from 1320 stands of different forest types in Patagonia, and random forest regression to map SOC. The model captured SOC variability well (R2 = 0.60, RMSE = 22.1%), considering the huge latitudinal extension (36.4deg to 55.1deg SL) and the great diversity of forest types. Mean SOC was 134.4 ton C ha-1 +/- 25.2, totaling 404.2 million ton C across Patagonia. Overall, SOC values were highest in valleys of the Andes mountains and in southern Tierra del Fuego, ranging from 53.5 to 277.8 ton C ha-1 for the whole Patagonia region. Soil organic carbon is a metric relevant to many applications, connecting major issues such as forest management, conservation, and livestock production, and having spatially explicit estimates of SOC enables managers to fulfil the international agreements that Argentina has joined.
File: Martinez-Pastur-et-al_remotesensing-14-05702.pdf
The ecosystem services that forests provide depend on tree species composition. Therefore, it is important to map not only forest extent and its dynamics, but also composition. Open access to Landsat has resulted in considerable improvements in remote sensing methods for mapping tree species, but most approaches fail to perform when there is a shortage of clear observations. Our main goal was to map forest composition with Landsat imagery in various data availability conditions, and to investigate how the missing data, either due to clouds or scan line problems affect classification accuracy. We tested a data driven approach that is based on multi-temporal analysis of the tree species’ spectral characteristics making it applicable to regional-scale mapping even when the gap-free imagery is not available. Our study area consisted of one Landsat footprint (26/28) located in Northern Wisconsin, USA. We selected this area because of numerous tree species (23), heterogenic composition of forests where the majority of stands are mixed, and availability of high-quality reference data. We quantified how classification accuracy at the species level was affected by a) the amount of missing data due to cloud cover and Scanning Line Corrector (SLC) gaps, b) the number of acquisitions, and c) the seasonal availability of images. We applied a decision tree classifier, capable of handling missing data to both single- and a three-year Landsat-7 and Landsat-8 observations. We classified the dominant tree species in each pixel and grouped results to forest stands to match our reference data. Our results show four major findings. First, producer’s and user’s accuracies range from 46.2% to 96.2% and from 59.9% to 93.7%, respectively for the most abundant forest types in the study area (all types covering greater than 2% of the forest area). Second, all tree species were mapped with overall accuracy above 70% even in when we restricted our data set to images having gaps larger than 30% of the study area. Third, the classification accuracy improved with more acquisitions, especially when images were available for the fall, spring, and summer. Finally, producer’s accuracies for pure-stands were higher than those for mixed stands by 10 to 30 percentage points. We conclude that inclusion of Landsat imagery with missing data allows to map forest types with accuracies that previously could be achieved only for those rare years for which several gap-free images were available. The approach presented here is directly applicable to Landsat-like observations and derived products such as seasonal composites and temporal statistics that miss 30% or more of the data for any single date to develop forest composition maps that are important for both forest management and ecology.
File: Turlej_IJAEOG_2022.pdf
Human activities alter ecosystems everywhere, causing rapid biodiversity loss and biotic homogenization. These losses necessitate coordinated conservation actions guided by biodiversity and species distribution spatial data that cover large areas yet have fine-enough resolution to be management-relevant (i.e., ≤5 km). However, most biodiversity products are too coarse for management or are only available for small areas. Furthermore, many maps generated for biodiversity assessment and conservation do not explicitly quantify the inherent tradeoff between resolution and accuracy when predicting biodiversity patterns. Our goals were to generate predictive models of overall breeding bird species richness and species richness of different guilds based on nine functional or life-history-based traits across the conterminous United States at three resolutions (0.5, 2.5, and 5 km) and quantify the tradeoff between resolution and accuracy and, hence, relevance for management of the resulting biodiversity maps. We summarized 18 years of North American Breeding Bird Survey data (1992–2019) and modeled species richness using random forests, including 66 predictor variables (describing climate, vegetation, geomorphology, and anthropogenic conditions), 20 of which we newly derived. Among the three spatial resolutions, the percentage variance explained ranged from 27% to 60% (median = 54%; mean = 57%) for overall species richness and 12% to 87% (median = 61%; mean = 58%) for our different guilds. Overall species richness and guild-specific species richness were best explained at 5-km resolution using ~24 predictor variables based on percentage variance explained, symmetric mean absolute percentage error, and root mean square error values. However, our 2.5-km-resolution maps were almost as accurate and provided more spatially detailed information, which is why we recommend them for most management applications. Our results represent the first consistent, occurrence-based, and nationwide maps of breeding bird richness with a thorough accuracy assessment that are also spatially detailed enough to inform local management decisions. More broadly, our findings highlight the importance of explicitly considering tradeoffs between resolution and accuracy to create management-relevant biodiversity products for large areas.
File: Carroll_et_al-2022_Mapping_breeding.pdf
Wars are frequent and can affect land use substantially, but the effects of wars can vary greatly depending on their characteristics, such as intensity or duration. Furthermore, the spatial scale of the effects can differ. The effects of wars may be localized and thus close to conflict locations if direct mechanisms matter most (e.g., abandonment because active fighting precludes farming), or wide-ranging, e.g., farther away from conflict locations, if indirect mechanisms predominate (e.g., no access to agricultural inputs). Our goal was to quantify how the very different wars in the Caucasus region during post-Soviet times most likely affected agricultural abandonment at different scales. We analyzed data on conflict locations plus Landsat-derived land-cover data from 1987 to 2015, and applied matching statistics, difference-in-differences estimators, and logistic panel regressions. We examined the localized versus wide-ranging effects of the different wars on permanent agricultural abandonment and inferred to direct and indirect mechanisms that may have resulted in agricultural abandonment. While permanent agricultural abandonment was overall surprisingly limited across the Caucasus, up to one third of abandonment was most likely related to the wars. Among the wars, the war in Chechnya was by far the most intense and longest, but its effect on abandonment was similar to the less intense and relatively short war in Abkhazia. 47 % and 45 % of agricultural abandonment was related to each war, respectively. The reason was that the effect of the war in Chechnya was more localized, and abandonment occurred near conflict locations, in contrast to Abkhazia, where the effect was wide-ranging and abandonment occurred farther away from conflict locations. In contrast, the war in South Ossetia showed no significant effect on abandonment, and the war in Nagorno-Karabakh had the surprising pattern that abandonment was higher where no war had occurred. For each of the wars, abandonment was predominately related to the nearest conflict locations, but in Abkhazia additional conflict locations within 10 km further increased the probability of abandonment. We infer that the direct mechanisms of the war such as bombing, and active fighting most likely resulted in a localized effect close to conflict locations in Chechnya and in Nagorno-Karabakh. However, in Nagorno-Karabakh subsidies for new settlers after the war, (i.e., a positive wide-ranging effect), potentially reduced the amount of abandonment there. In contrast, negative wide-ranging effects such as refugee movements and post-war restrictions on their return is related to broad-scale abandonment in Abkhazia. In summary, permanent agricultural abandonment was not necessarily higher in a war with a high overall intensity. Instead, the effect of a given war varied in scale, and was related to the relative importance of direct and localized versus indirect and wide-ranging mechanisms, including postwar events and policies, which is likely the case for other wars, too.
File: Buchner-GCB_Buchner_2022.pdf
In mountainous environments, topography strongly affects the reflectance due to illumination effects and cast shadows, which introduce errors in land cover classifications. However, topographic correction is not routinely implemented in standard data pre-processing chains (e.g., Landsat Analysis Ready Data), and there is a lack of consensus whether topographic correction is necessary, and if so, how to conduct it. Furthermore, methods that correct simultaneously for atmospheric and topographic effects are becoming available, but they have not been compared directly. Our objects were to investigate (1) the effectiveness of two topographic correction approaches that integrate atmospheric and topographic correction, (2) improvements in classification accuracy when analyzing topographically corrected single-date imagery (14 July 2016 and 2 October 2016), versus a full Landsat time series from 2014 to 2016, and 3) improvements in classification accuracy when including additional terrain information (i.e. topographic slope, elevation, and aspect). We developed a physical based model and compared it with an enhanced C-correction, both of which integrate atmospheric and topographic correction. We compared classification accuracies with and without topographic correction using combinations of single-date imagery, image composites and spectral-temporal metrics generated from the full Landsat time series, and additional terrain information in the Caucasus Mountains. We found that both the enhanced C-correction and the physical model performed very well and largely eliminated the correlation (Pearson’s correlation coefficient r ranges from 0.06 to 0.24) between surface reflectance and illumination condition, but the physical model performed best (r ranges from 0.05 to 0.11). Both image composites, and spectral-temporal metrics generated from corrected imagery, resulted in significantly (p ≤ 0.05) higher classification accuracies and better forest classifications, especially for the mixed forests. Adding terrain information reduced classification error significantly, but not as much as topographic correction. In summary, topographic correction remains necessary, even when analyzing a full Landsat time series and including a digital elevation model in the classification. We recommend that topographic correction should be applied when analyzing Landsat satellite imagery in mountainous region for forest cover classification.
File: Yin_IJAEOG_Yin_2022.pdf
Forest biodiversity conservation and species distribution modeling greatly benefit from broad-scale forest maps depicting tree species or forest types rather than just presence and absence of forest, or coarse classifications. Ideally, such maps would stem from satellite image classification based on abundant field data for both model training and accuracy assessments, but such field data do not exist in many parts of the globe. However, different forest types and tree species differ in their vegetation phenology, offering an opportunity to map and characterize forests based on the seasonal dynamic of vegetation indices and auxiliary data. Our goal was to map and characterize forests based on both land surface phenology and climate patterns, defined here as forest phenoclusters. We applied our methodology in Argentina (2.8 million km2), which has a wide variety of forests, from rainforests to cold-temperate forests. We calculated phenology measures after fitting a harmonic curve of the enhanced vegetation index (EVI) time series derived from 30-m Sentinel 2 and Landsat 8 data from 2018–2019. For climate, we calculated land surface temperature (LST) from Band 10 of the thermal infrared sensor (TIRS) of Landsat 8, and precipitation from Worldclim (BIO12). We performed stratified X-means cluster classifications followed by hierarchical clustering. The resulting clusters separated well into 54 forest phenoclusters with unique combinations of vegetation phenology and climate characteristics. The EVI 90th percentile was more important than our climate and other phenology measures in providing separability among different forest phenoclusters. Our results highlight the potential of combining remotely sensed phenology measures and climate data to improve broad-scale forest mapping for different management and conservation goals, capturing functional rather than structural or compositional characteristics between and within tree species. Our approach results in classifications that go beyond simple forest–nonforest in areas where the lack of detailed ecological field data precludes tree species–level classifications, yet conservation needs are high. Our map of forest phenoclusters is a valuable tool for the assessment of natural resources, and the management of the environment at scales relevant for conservation actions.
File: Silveira-Ecological-Applications-2022-Silveira-Forest-phenoclusters-for-Argentina-based-on-vegetation-phenology-and-climate.pdf
Grasslands are important for global biodiversity, food security, and climate change analyses, which makes mapping and monitoring of vegetation changes in grasslands necessary to better understand, sustainably manage, and protect these ecosystems. However, grassland vegetation monitoring at spatial and temporal resolution relevant to land management (e.g., ca. 30-m, and at least annually over long time periods) is challenging due to complex spatio-temporal pattern of changes and often limited data availability. Here we assess both shortand long-term changes in grassland vegetation cover from 1987 to 2019 across the Caucasus ecoregion at 30-m resolution based on Cumulative Endmember Fractions (i.e., annual sums of monthly ground cover fractions) derived from the full Landsat record, and temporal segmentation with LandTrendr. Our approach combines the benefits of physically-based analyses, missing data prediction, annual aggregations, and adaptive identification of changes in the time-series. We analyzed changes in vegetation fraction cover to infer the location, timing, and magnitude of vegetation change episodes of any length, quantified shifts among all ground cover fractions (i.e., green vegetation, non-photosynthetic vegetation, soil, and shade), and identified change pathways (i.e., green vegetation loss, desiccation, dry vegetation loss, revegetation green fraction, greening, or revegetation dry fraction). We found widespread long-term positive changes in grassland vegetation (32.7% of grasslands), especially in the early 2000s, but negative changes pathways were most common before the year 2000. We found little association between changes in green vegetation and meteorological conditions, and varied relationships with livestock populations. However, we also found strong spatial heterogeneity in vegetation dynamics among neighboring fields and pastures, demonstrating capability of our approach for grassland management at local levels. Our results provide a detailed assessment of grassland vegetation change in the Caucasus Ecoregion, and present an approach to map changes in grasslands even where availability of Landsat data is limited.
File: 1-s2.0-S2666017221000225-main.pdf
Global change analyses are facilitated by the growing number of remote-sensing datasets that have both broad spatial extent and repeated observations over decades. These datasets provide unprecedented power to detect patterns of time trends involving information from all pixels on a map. However, rigorously testing for time trends requires a solid statistical foundation to identify underlying patterns and test hypotheses. Appropriate statistical analyses are challenging because environmental data often have temporal and spatial autocorrelation, which can either obscure underlying patterns in the data or suggest false associations between patterns in the data and independent values used to explain them. Existing statistical methods that account for temporal and spatial autocorrelation are not practical for remote-sensing datasets that often contain millions of pixels. Here, we first analyze simulated data to show the need to account for both spatial and temporal autocorrelation in time-trend analyses. Second, we present a new statistical approach, PARTS (Partitioned Autoregressive Time Series), to identify underlying patterns and test hypotheses about time trends using all pixels in large remote sensing datasets. PARTS is flexible and can include, for example, the effects of multiple independent variables, such as land-cover or latitude, on time trends. Third, we use PARTS to analyze global trends in NDVI, focusing on trends in pixels that have not experienced land-cover change. We found that despite the appearance of overall increases in NDVI in all continents, there is little statistical support for these trends except for Asia and Europe, and only in some land-cover classes. Furthermore, we found no overall latitudinal trend in greening for any continent, but some latitude by land-cover class interactions, implying that latitudinal patterns differed among land-cover classes. PARTS makes it possible to identify patterns and test hypotheses that involve the aggregate information from many pixels on a map, thereby increasing the value of existing remote-sensing datasets.
File: 1-s2.0-S0034425721003989-main.pdf
Bird species richness is highly dependent on the amount of energy available in an ecosystem, with more available
energy supporting higher species richness. A good indicator for available energy is Gross Primary Productivity
(GPP), which can be estimated from satellite data.
Our question was how temporal dynamics in GPP affect bird species richness. Specifically, we evaluated the
potential of the Dynamic Habitat Indices (DHIs) derived from MODIS GPP data together with environmental and
climatic variables to explain annual patterns in bird richness across the conterminous United States. By focusing
on annual DHIs, we expand on previous applications of multi-year composite DHIs, and could evaluate lag-effects
between changes in GPP and species richness.
We used 8-day GPP data from 2003 to 2013 to calculate annual DHIs, which capture three aspects of vegetation
productivity: (1) annual cumulative productivity, (2) annual minimum productivity, and (3) annual
seasonality expressed as the coefficient of variation in productivity. For each year from 2003 to 2013, we
calculated total bird species richness and richness within six functional guilds, based on North American
Breeding Bird Survey data.
The DHIs alone explained up to 53% of the variation in annual bird richness within the different guilds
(adjusted deviance-squared D2adj = 0.20–0.52), and up to 75% of the variation (D2adj = 0.28–0.75) when
combined with other environmental and climatic variables. Annual DHIs had the highest explanatory power for
habitat-based guilds, such as grassland (D2adj = 0.67) and woodland breeding species (D2adj = 0.75). We found
some inter-annual variability in the explanatory power of annual DHIs, with a difference of 5–7 percentage
points in explained variation among years in DHI-only models, and 3–7 points for models combining DHI,
environmental and climatic variables. Our results using lagged year models did not deviate substantially from
same-year annual models.
We demonstrate the relevance of annual DHIs for biodiversity science, as effective predictors of temporal
variation in species richness patterns. We suggest that the use of annual DHIs can improve conservation planning,
by conveying the range of patterns of biodiversity response to global changes, over time.
File: Hobi-et-al-2021_BirdSpeciesRichness_DynamicHabIndices_EcolIndicators.pdf
