Landsat-derived Normalized Difference Vegetation Index Research Articles

Establishing ecological thresholds and targets for groundwater management

AbstractGroundwater is critical for many ecosystems, yet groundwater requirements for dependent ecosystems are rarely accounted for during water and conservation planning. Here we compile 38 years of Landsat-derived normalized difference vegetation index (NDVI) to evaluate groundwater-dependent vegetation responses to changes in depth to groundwater (DTG) across California. To maximize applicability, we standardized raw NDVI and DTG values using Z scores to identify groundwater thresholds, groundwater targets and map potential drought refugia across a diversity of biomes and local conditions. Groundwater thresholds were analysed for vegetation impacts where ZNDVI dropped below −1. ZDTG thresholds and targets were then evaluated with respect to groundwater-dependent vegetation in different condition classes and rooting depths. ZNDVI scores were applied statewide to identify potential drought refugia supported by groundwater. Our approach provides a simple and robust methodology for water and conservation practitioners to support ecosystem water needs so biodiversity and sustainable water-management goals can be achieved.

Identifying the Climatic and Anthropogenic Impact on Vegetation Surrounding the Natural Springs of the Arava Valley Using Remote Sensing Methods

Natural springs, recognized as biodiversity hotspots and keystone ecosystems, exert positive ecological influences beyond their immediate extent, particularly in dryland environments. The water feeding these springs, largely governed by natural climatic conditions, is susceptible to anthropogenic impacts. The objective of this study was to determine the factors that cause fluctuations in water availability to springs of the hyper-arid Arava Valley (Israel/Jordan). Using the Standard Precipitation Index, we statistically classified the historical record of yearly rainfall for the past four decades into clusters of dry and wet sub-periods. We assessed changes in vegetation cover around the springs using the Landsat-derived Normalized Difference Vegetation Index (NDVI) for each sub-period. To assess the anthropogenic effects, we examined the correlations between vegetation cover, water extraction from the aquifer, and the status of adjacent agricultural plots that share a hydrological connection with the springs. Our findings revealed fluctuations between wet and dry sub-periods over the last four decades. We observed high responsiveness of vegetation cover around the springs to these fluctuating sub-periods. Of the 25 studied springs, 12 were directly influenced by anthropogenic factors—7 experienced a decline in vegetation, which we attributed to water extraction from the aquifers, while vegetation increase in 5 springs was attributed to water seepage from agricultural areas upstream. In conclusion, addressing vital habitats such as natural springs in arid drylands requires a holistic approach that integrates long-term climatic, ecological, and anthropogenic observations.

Exploring NDVI change patterns across the Tibetan Plateau at the hillslope scale using geomorphons

ABSTRACTWithin the Tibetan Plateau (TP), grassland degradation is of great concern, and there is a need to better understand the spatial variability and landscape patterns in grassland degradation across the TP. This study explores potential patterns in grassland degradation, estimated using Landsat-derived normalized difference vegetation index (NDVI) growing season change products generated by a prior study. Shuttle Radar Topography Mission (SRTM) digital terrain model (DTM)-derived geomorphons, topographic slope and the topographic position index (TPI) were used to explore grassland degradation and assess whether changes in growing season NDVI are associated with specific grasslands or landforms. An increase in NDVI within the TP was observed especially during the 1990 to 2018 and 2000 to 2018 time periods. Higher growing season median NDVI change values were found for the Southeast Tibet Shrublands and Meadows (SETSM) and TP Alpine Shrublands and Meadows (TPASM) grasslands in comparison to the other grassland types analysed, suggesting a more pronounced greening trend in the eastern portion of the plateau in comparison to the western portion. Small differences in NDVI change were observed for different geomorphon-based landforms, while topographic slope and TPI showed only modest correlations with NDVI change for all time periods and grassland types. More localized patterns of grassland degradation were masked by the widespread greening trends. Further analysis of only pixels that experienced a decreasing NDVI magnitude of less than or equal to −0.1 also generally suggested more change in the SETSM and TPASM. While this study generally supports the use of geomorphons as an analysis and aggregating unit for studying change patterns at the hillslope scale, more research is required to fully grasp the dynamics of landscape change at the hillslope scale in the TP.

Impact of freshwater diversions on vegetation in coastal wetlands based on remote sensing derived vegetation index

There exist contrasting results on the impact of large-scale freshwater and sediment diversions on land gain/loss. To improve understanding on the efficacy of diversion projects in restoring coastal wetlands, we aim to evaluate the long-term impacts of diversion-altered salinity and water level on vegetation productivity in coastal wetlands. Two freshwater diversion projects Caernarvon and Davis Pond in Louisiana, U.S. and associated reference sites were selected for inclusion in this study. We implemented multi-level Bayesian models to evaluate 1) how vegetation productivity approximated by Landsat-derived Normalized Difference Vegetation Index (NDVI) in peak biomass season changed over time (pre- and post-construction and operation of the freshwater diversions), and 2) how peak-season NDVI responded to mean and variability of water level and salinity among a suite of factors that could affect vegetation productivity. Analysis showed difference in temporal trends of NDVI between the reference and diversion site for the Caernarvon project with a significant increase at the diversion site although NDVI negatively responded to diversion at the beginning. Diversion did not seem to negatively affect NDVI for the Davis Pond project and NDVI only marginally increased at the diversion site. For both projects, NDVI negatively responded to water level, while salinity negatively affected NDVI or showed quadratic relation with NDVI. At the Caernarvon diversion project, the negative impact of water level on NDVI was greater at the diversion site than at the reference site. At the Davis Pond project, it was the salinity that showed different impact between the diversion and reference site. The temporal increase of NDVI at the Caernarvon diversion site can be explained by the lower salinity driven by diversion. The quadratic relation between salinity and NDVI at the Davis Pond diversion project led to only marginal increase at the diversion site. This study provided uncertainty estimates in temporal trend of NDVI and the impact of two key abiotic drivers on NDVI. The improved understanding on vegetation productivity will help predict landscape change in response to freshwater diversions.

Remote sensing reveals unprecedented sublethal impacts of a 40-year-old oil spill on mangroves

Oil spills cause long-lasting mangrove loss, threatening their conservation and ecosystem services worldwide. Oil spills impact mangrove forests at various spatial and temporal scales. Yet, their long-term sublethal effects on trees remain poorly documented. Here, we explore these effects based on one of the largest oil spills ever recorded, the Baixada Santista pipeline leak, which hit the mangroves of the Brazilian southeastern coast in 1983. Historical, Landsat-derived normalized difference vegetation index (NDVI) maps over the spilled mangrove reveal a large dieback of trees within a year following the oil spill, followed by a eight-year recolonization period and a stabilization of the canopy cover, however 20–30% lower than initially observed. We explain this permanent loss by an unexpected persistence of oil pollution in the sediments based on visual and geochemical evidence. Using field spectroscopy and cutting-edge drone hyperspectral imaging, we demonstrate how the continuous exposure of mangrove trees to high levels of pollution affects their health and productivity in the long term, by imposing permanent stressful conditions. Our study also reveals that tree species differ in their sensitivity to oil, giving the most tolerant ones a competitive advantage to recolonize spilled mangroves. By leveraging drone laser scanning, we estimate the loss of forest biomass caused by the oil spill to be 9.8–91.2 t ha−1, corresponding to 4.3–40.1 t C ha−1. Based on our findings, we encourage environmental agencies and lawmakers to consider the sublethal effects of oil spills on mangroves in the environmental cost of these accidents. We also encourage petroleum companies to use drone remote sensing in monitoring routines and oil spill response planning to improve mangrove preservation and impact assessment.

Gene-environment interaction in the association of residential greenness and 25(OH) vitamin D

There is increasing awareness for beneficial health effects of green space surrounding the home, but the underlying mechanisms are not yet fully understood and challenging to study given the correlation with other exposures. Here, the association of residential greenness and vitamin D including a gene-environment interaction is investigated. 25-hydroxyvitamin D (25(OH)D) was measured by electrochemiluminescence at ages 10 and 15 years in participants of two German birth cohorts GINIplus and LISA. Greenness was measured using the Landsat-derived Normalized Difference Vegetation Index (NDVI) in a 500 m buffer surrounding the home. Linear and logistic regression models were applied at both time points adjusted for several covariates (N10Y = 2,504, N15Y = 2,613). In additional analyses vitamin D-related genes, physical activity, time spent outdoors, supplements, and measurement season were investigated as potential confounders or effect modifiers. A 1.5-SD increase in NDVI was significantly associated with increased 25(OH)D values at ages 10 and 15 years (β10y = 2.41 nmol/l, p=<0.01; β15y = 2.03 nmol/l, p = 0.02). In stratified analyses, the associations were not seen in participants spending more than 5 h/day outside in summer, having a high physical activity level, taking supplements, or being examined during the winter season. In a subset (n = 1,732) with genetic data, a significant gene-environment interaction of NDVI with CYP2R1, an upstream gene in 25(OH)D synthesis, was observed at age 10 years. When investigating 25(OH)D sufficiency, defined as values above 50 nmol/l, a 1.5-SD increase in NDVI was associated with significantly higher odds of having sufficient 25 (OH)D levels at age 10 years (OR = 1.48, 1.19–1.83). In conclusion, robust associations between residential greenness and 25 (OH)D levels were observed in children and adolescents independent of other confounders and additionally supported by the presence of a gene-environment interaction. Effects of NDVI were stronger in those having lower vitamin D levels at age 10 years due to their covariate profile or genetically lower 25(OH)D synthesis.

A new method for calculating C factor when projecting future soil loss using the Revised Universal soil loss equation (RUSLE) in semi-arid environments

Understanding how reservoir sedimentation rates may evolve due to climate change is essential for projecting future changes in reservoir water storage capacity. The Revised Universal Soil Loss Equation (RUSLE) is commonly used to assess regional soil loss rates because of its suitability for working with the coarse temporal and spatial scale of climate model outputs. Application of the RUSLE for projecting future erosion rates is constrained by the relatively limited number of classes used in projected changes of native vegetation: this typically reduces a wide range of variation in RUSLE cover (C) factors to a single value per class, and these classes may be insensitive to changes in species composition and canopy density with time and space. This paper develops a low-cost, efficient, and objective approach to projecting future C values directly based on widely available Landsat data, downscaled climate model data, and a limited number of terrain variables. Observed C is estimated from Landsat-derived Normalized Difference Vegetation Index (NDVI) and Modified Soil-Adjusted Vegetation Index (MSAVI) values using standard methods. A linear relationship is established between the observed C values and average annual antecedent temperature, average annual antecedent precipitation, latitude, and percent sand (adjusted R2 = 0.75604, 0.7543, and RMSE = 0.09448, 0.07043, respectively). A proof of principal study demonstrates that, unlike when land cover classes are used to estimates C, a RUSLE model driven by the regression-based C provides a correct order-of-magnitude estimate of observed long-term erosion and sediment yield rates (e.g., an estimate of 2.9–3.2 t ha−1 y-1 is obtained where observed rates range from 1.2 to 3.9 t ha−1 y-1 for the same time period, and up to 8 t ha−1 y-1 historically, while standard formulations of RUSLE yield 16.1 and 0.3 t ha−1 y-1). Using climate model data to estimate both precipitation intensity (R factor) and C in the RUSLE model results in basin wide projected end 21st century soil erosion rates for relative concentration pathway 8.5 of 5.5–7.5 t ha−1 y-1, consistent with expectations for accelerated soil loss in the study area as climate change reduces soil moisture while increasing precipitation intensity. The same model driven by land cover class estimates of C produces soil loss rates an order of magnitude smaller (0.3–0.4 t ha−1 y-1), consistent with a much more heavily vegetated landscape than could be supported under the increasingly arid projected climate.

Detecting Spatiotemporal Differences in Cropland Abandonment and Reforestation Across the Three-North Region of China Based on Landsat Time Series

The Three-North region of China is an important ecological protection area. While the depopulation of rural areas and policy intervention have allowed cropland abandonment and reforestation (CAR), the spatial distribution and temporal differences underlying this landscape pattern have not been analyzed. CAR results in the gradual disappearance of the soil signal representing tilled or harvested crops, followed by a gradual increase in the interannual vegetation signal. This long-term gradual greening can be exploited to map the spatiotemporal distribution of CAR. In this study, Landsat-based detection of trends in disturbance and recovery (LandTrendr) was applied to the 1990–2020 Landsat-derived normalized difference vegetation index (NDVI) time series. The data were combined with a differential analysis of the phenological characteristics of the surveyed area to detect land surface greening trends. The relationship with actual information on CAR was then determined. The results showed widespread long-term greening in the Three-North region, covering up to 80% of the region) during the annual growth (May–June) and harvest (September–October) periods. Together, these two periods indicated 7.5% overall greening. The highest greening intensity was in areas where cropland undergone conversion to forest, which strongly suggested CAR as the succession type. The average accuracy of CAR mapping was ~80% (mapping accuracy of a greening intensity > 0.4 was as high as 87%). This study demonstrates the utility of our method in studies of driving factors of CAR and ecological benefits of secondary forest restoration.

Multi-Year Mapping of Disturbance and Reclamation Patterns over Tronox’s Hillendale Mine, South Africa with DBEST and Google Earth Engine

This study was devised to examine the pattern of disturbance and reclamation by Tronox, which instigated a closure process for its Hillendale mine site in South Africa, where they recovered zirconium- and titanium-bearing minerals from 2001 to 2013. Restoring mined-out areas is of great importance in South Africa, with its ominous record of almost 6000 abandoned mines since the 1860s. In 2002, the government enacted the Mineral and Petroleum Resources Development Act (No. 28 of 2002) to enforce extracting companies to restore mined-out areas before pursuing closure permits. Thus, the trajectory of the Hillendale mine remains unstudied despite advances in the satellite remote sensing technology that is widely used in this field. Here, we retrieved a collection of Landsat-derived normalized difference vegetation index (NDVI) within the Google Earth Engine and applied the Detecting Breakpoints and Estimating Segments in Trend (DBEST) algorithm to examine the progress of vegetation transformation over the Hillendale mine between 2001 and 2019. Our results showed key breakpoints in NDVI, a drop from 2001, reaching the lowest point in 2009–2011, with a marked recovery pattern after 2013 when the restoration program started. We also validated our results using a random forests strategy that separated vegetated and non-vegetated areas with an accuracy exceeding 78%. Overall, our findings are expected to encourage users to replicate this affordable application, particularly in emerging countries with similar cases.

Climate Variability May Delay Post-Fire Recovery of Boreal Forest in Southern Siberia, Russia

Prolonged dry periods and increased temperatures that result from anthropogenic climate change have been shown to increase the frequency and severity of wildfires in the boreal region. There is growing evidence that such changes in fire regime can reduce forest resilience and drive shifts in post-fire plant successional trajectories. The response of post-fire vegetation communities to climate variability is under-studied, despite being a critical phase determining the ultimate successional conclusion. This study investigated the responses of post-fire recruited species to climate change and inter-annual variability at 16 study sites that experienced high-severity fire events, mostly in early 2000, within the Scots pine forest-steppe zone of southeastern Siberia, Russia. These sites were originally dominated by Scots pine, and by 2018, they were recruited by different successional species. Additionally, three mature Scots pine stands were included for comparison. A Bayesian Additive Regression Trees (BART) approach was used to model the relationship between Landsat-derived Normalized Difference Vegetation Index (NDVI) time series, temperature and precipitation in the 15 years after a stand-replacing fire. Using the resulting BART models, together with six projected climate scenarios with increased temperature and enhanced inner-annual precipitation variability, we simulated NDVI at 5-year intervals for 15 years post-fire. Our results show that the BART models performed well, with in-sample Pseudo-R2 varying from 0.49 to 0.95 for fire-disturbed sites. Increased temperature enhanced greenness across all sites and across all three time periods since fires, exhibiting a positive feedback in a warming environment. Repeatedly dry spring periods reduced NDVI at all the sites and wetter summer periods following such dry springs could not compensate for this, indicating that a prolonged dry spring has a strong impact consistently over the entire early developmental stages from the initial 5 years to 15 years post-fire. Further, young forests showed higher climate sensitivity compared to the mature forest, irrespective of species and projected climatic conditions. Our findings suggest that a dry spring not only increases fire risk, but also delays recovery of boreal forests in southern Siberia. It also highlights the importance of changing rainfall seasonality as well as total rainfall in a changing climate for post-fire recovery of forest.

Greening trends and their relationship with agricultural land abandonment across Poland

Substantial land surface greening has been observed globally over the last few decades. This greening has been primarily attributed to climate change and CO2 concentrations, but recent research studies have emphasized the large role of human land use in this process, especially agricultural land abandonment (ALA). The aim of this study was to investigate whether long-term gradual greening of agricultural land can aid in mapping abandoned land across Poland. Trend estimation and temporal segmentation were applied to the 1986–2019 annual Landsat-derived Normalized Difference Vegetation Index (NDVI) time series to detect periods of long-term greening and to assess its relationship with actual information on ALA. The results show that long-term greening is widespread in Poland (covering up to 60% of its territory), regardless of former and current land uses, most of which has stemmed from modifications within stable land use classes (as they cover 91% of Poland). The highest greening rates and intensities were observed due to conversions from agricultural to non-agricultural land use, which strongly suggests land abandonment and the proceeding succession. However, some proportion of managed agricultural land has also experienced high-intensity greening, and a large proportion of abandoned agricultural land is not greening intensively. Thus, setting an exact greening intensity threshold allowing to clearly distinguish the status of land was not possible. This study also presents the consistency of the Landsat annual time series and its good performance for long-term greening detection, indicating that temporal segmentation may better capture ALA patterns than trend estimation

Positive response of tree productivity to warming is reversed by increased tree density at the Arctic tundra–taiga ecotone

The transition zone between the northern boreal forest and the Arctic tundra, known as the tundra–taiga ecotone (TTE) has undergone rapid warming in recent decades. In response to this warming, tree density, growth, and stand productivity are expected to increase. Increases in tree density have the potential to negate the positive impacts of warming on tree growth through a reduction in the active layer and an increase in competitive interactions. We assessed the effects of tree density on tree growth and climate–growth responses of Cajander larch (Larix cajanderi Mayr.) and on trends in the normalized difference vegetation index (NDVI) in the TTE of Northeast Siberia. We examined 19 mature forest stands that all established after a fire in 1940 and ranged in tree density from 300 to 37 000 stems·ha−1. High-density stands with shallow active layers had lower tree growth, higher stand productivity, and more negative growth responses to growing season temperatures compared with low-density stands with deep active layers. Variation in stand productivity across the density gradient was not captured by Landsat-derived NDVI, but NDVI did capture annual variations in stand productivity. Our results suggest that the expected increases in tree density following fires at the TTE may effectively limit tree growth and that NDVI is unlikely to capture increasing productivity associated with changes in tree density.

The pattern, change and driven factors of vegetation cover in the Qin Mountains region

The Qin Mountains region is one of the most important climatic boundaries that divide the North and South of China. This study investigates vegetation covers changes across the Qin Mountains region over the past three decades based on the Landsat-derived Normalized Difference Vegetation Index (NDVI), which were extracted from the Google Earth Engine (GEE). Our results show that the NDVI across the Qin Mountains have increased from 0.624 to 0.776 with annual change rates of 0.0053/a over the past 32 years. Besides, its abrupt point occurred in 2006 and the change rates after this point increased by 0.0094/a (R2 = 0.8159, p < 0.01) (2006–2018), which is higher than that in 1987–1999 and 1999–2006. The mean NDVI have changed in different elevation ranges. The NDVI in the areas below 3300 m increased, such increased is especially most obviously in the cropland. Most of the forest and grassland locate above 3300 m with higher increased rate. Before 2006, the temperature and reference evapotranspiration (PET) were the important driven factors of NDVI change below 3300 m. After afforestation, human activities become important factors that influenced NDVI changes in the low elevation area, but hydro-climatic factors still play an important role in NDVI increase in the higher elevations area.

Monitoring tropical forest succession at landscape scales despite uncertainty in Landsat time series.

Forecasting rates of forest succession at landscape scales will aid global efforts to restore tree cover to millions of hectares of degraded land. While optical satellite remote sensing can detect regional land cover change, quantifying forest structural change is challenging. We developed a state-space modeling framework that applies Landsat satellite data to estimate variability in rates of natural regeneration between sites in a tropical landscape. Our models work by disentangling measurement error in Landsat-derived spectral reflectance from process error related to successional variability. We applied our modeling framework to rank rates of forest succession between 10 naturally regenerating sites in Southwestern Panama from about 2001 to 2015 and tested how different models for measurement error impacted forecast accuracy, ecological inference, and rankings of successional rates between sites. We achieved the greatest increase in forecasting accuracy by adding intra-annual phenological variation to a model based on Landsat-derived normalized difference vegetation index (NDVI). The best-performing model accounted for inter- and intra-annual noise in spectral reflectance and translated NDVI to canopy height via Landsat-lidar fusion. Modeling forest succession as a function of canopy height rather than NDVI also resulted in more realistic estimates of forest state during early succession, including greater confidence in rank order of successional rates between sites. These results establish the viability of state-space models to quantify ecological dynamics from time series of space-borne imagery. State-space models also provide a statistical approach well-suited to fusing high-resolution data, such as airborne lidar, with lower-resolution data that provides better temporal and spatial coverage, such as the Landsat satellite record. Monitoring forest succession using satellite imagery could play a key role in achieving global restoration targets, including identifying sites that will regain tree cover with minimal intervention.

Assessing salt marsh extent and condition changes with 35 years of Landsat imagery: Tagus Estuary case study

Reliable information on salt marsh extent and condition is crucial to promote effective management strategies towards their maintenance and recovery. Most of previous studies on salt marsh extent assessment used Landsat-derived Normalized Difference Vegetation Index (NDVI), being limited the current knowledge about the performance of other Vegetation Indices (VI). Based on Landsat imagery this study proposes a new methodology to map salt marsh extent in estuarine systems by combining Normalized Difference Water Index (NDWI) and VI, exploring their performance when using different VI. Moreover, it aims to assess the extent and condition changes between 1984 and 2018 in two salt marshes (Pancas and Corroios) located within Tagus Estuary. The VI best-performing salt marsh extent was determined and the methodology applied to assess salt marsh extent changes. Condition change was investigated by statistically analyzing spatially averaged VI over salt marsh extent change regions. Results demonstrated that NDWI and VI combined can be used to efficiently map the marsh extent and NDVI was the VI with the best performance. Corroios revealed mostly stable without noticeable changes in its extent and condition. Oppositely, Pancas registered a continuous seaward progression at a mean rate of 3 ha/year since 1984, while restricted upper regions dieback after 2004–2005, likely due to high soil salinity conditions. In general, NDVI performs better salt marsh extent, but others VI corrected to minimize soil effects perform better when tidal flats are entirely, or almost entirely, exposed, opening perspectives to the development of new methods combining the use of different VI to optimize salt marsh extent detection.

Mapping impervious surface distribution in China using multi-source remotely sensed data

ABSTRACTImpervious surface area (ISA) data are required for such studies as urban environmental modeling, hydrological modeling, and socioeconomic analysis, but updating these datasets in a large area remains a challenge due to the complex urban landscapes consisting of different materials and colors with various spatial patterns. This research explores the integration of multi-source remotely sensed data for mapping China’s ISA distribution at 30-m spatial resolution. The integration of Visible Infrared Imaging Radiometer Suite Day/Night Band (VIIRS DNB) and Moderate Resolution Imaging Spectroradiometer (MODIS) normalized difference vegetation index (NDVI) data were used to extract initial ISA with spatial resolution of 250 m using a thresholding approach. The Landsat-derived NDVI and Modified Normalized Difference Water Index (MNDWI) were used to remove vegetation and water areas from the mixed pixels that existed in the initial ISA data. The spectral signatures of these ISA data were further extracted from Landsat multispectral images and used to refine the ISA data using expert knowledge. The results indicate that the integration of multi-source data can successfully map ISA distribution with 30-m spatial resolution in China with producer’s and user’s accuracies of 83.1 and 91.9%, respectively. These ISA data are valuable for better management of urban landscapes and for use as an input in other studies such as socioeconomic and environmental modeling.

Wetland Dynamics Inferred from Spectral Analyses of Hydro-Meteorological Signals and Landsat Derived Vegetation Indices

The dynamic response of coastal wetlands (CWs) to hydro-meteorological signals is a key indicator for understanding climate driven variations in wetland ecosystems. This study explored the response of CW dynamics to hydro-meteorological signals using time series of Landsat-derived normalized difference vegetation index (NDVI) values at six locations and hydro-meteorological time-series from 1984 to 2015 in Apalachicola Bay, Florida. Spectral analysis revealed more persistence in NDVI values for forested wetlands in the annual frequency domain, compared to scrub and emergent wetlands. This behavior reversed in the decadal frequency domain, where scrub and emergent wetlands had a more persistent NDVI than forested wetlands. The wetland dynamics were found to be driven mostly by the Apalachicola Bay water level and precipitation. Cross-spectral analysis indicated a maximum time-lag of 2.7 months between temperature and NDVI, whereas NDVI lagged water level by a maximum of 2.2 months. The quantification of persistent behavior and subsequent understanding that CW dynamics are mostly driven by water level and precipitation suggests that the severity of droughts, floods, and storm surges will be a driving factor in the future sustainability of CW ecosystems.

Mapping Annual Urban Change Using Time Series Landsat and NLCD

Annual urban change information is important for an improved understanding of urban dynamics and continuous modeling of urban ecosystem processes. This study examined Landsat-derived Normalized Difference Vegetation Index (NDVI) time series for characterizing annual urban change. To reduce impacts from cloud contamination and missing data, United States Geological Survey (USGS) Landsat Analysis Ready Data were processed to derive annual NDVI layers using a maximum value composite algorithm. National Land Cover Database land cover products from 2001 and 2011 were used as references for generating a decadal urban change mask. Within the decadal urban change mask and using annual NDVI as input, we examined three time-series change detection methods to pinpoint specific year of urban change: (a) minimum-value method, (b) break-point detection, and (c) simple-threshold identification. For accuracy assessment, we divided change pixels into urbanization and urban-intensification pixel groups, defined by initial land cover types. We used Google Earth's High-Resolution Imagery Archive as primary reference data for detailed accuracy assessment. Overall, the urbanization pixel group has good change detection accuracies of above 82% for all three change detection algorithms. The break-point detection method resulted in the highest overall accuracy of 88%. Overall accuracies for urban intensification pixel group were in the range of 35%–76%, depending on choice of change detection algorithm, length of input time-series, and further division of pixel subgroups.

Mapping paddy rice fields by applying machine learning algorithms to multi-temporal Sentinel-1A and Landsat data

ABSTRACTSentinel-1A synthetic aperture radar (SAR) data present an opportunity for acquiring crop information without restrictions caused by weather and illumination conditions, at a spatial resolution appropriate for individual rice fields and a temporal resolution sufficient to capture the growth profiles of different crop species. This study investigated the use of multi-temporal Sentinel-1A SAR data and Landsat-derived normalized difference vegetation index (NDVI) data to map the spatial distribution of paddy rice fields across parts of the Sanjiang plain, in northeast China. The satellite sensor data were acquired throughout the rice crop-growing season (May–October). A co-registered set of 10 dual polarization (VH/VV) SAR and NDVI images depicting crop phenological development were used as inputs to Support Vector Machine (SVM) and Random Forest (RF) machine learning classification algorithms in order to map paddy rice fields. The results showed a significant increase in overall classification when the NDVI time-series data were integrated with the various combinations of multi-temporal polarization channels (i.e. VH, VV, and VH/VV). The highest classification accuracies overall (95.2%) and for paddy rice (96.7%) were generated using the RF algorithm applied to combined multi-temporal VH polarization and NDVI data. The SVM classifier was most effective when applied to the dual polarization (i.e. VH and VV) SAR data alone and this generated overall and paddy rice classification accuracies of 91.6% and 82.5%, respectively. The results demonstrate the practicality of implementing RF or SVM machine learning algorithms to produce 10 m spatial resolution maps of paddy rice fields with limited ground data using a combination of multi-temporal SAR and NDVI data, where available, or SAR data alone. The methodological framework developed in this study is apposite for large-scale implementation across China and other major rice-growing regions of the world.

Evidence of compounded disturbance effects on vegetation recovery following high-severity wildfire and spruce beetle outbreak.

Spruce beetle (Dendroctonus rufipennis) outbreaks are rapidly spreading throughout subalpine forests of the Rocky Mountains, raising concerns that altered fuel structures may increase the ecological severity of wildfires. Although many recent studies have found no conclusive link between beetle outbreaks and increased fire size or canopy mortality, few studies have addressed whether these combined disturbances produce compounded effects on short-term vegetation recovery. We tested for an effect of spruce beetle outbreak severity on vegetation recovery in the West Fork Complex fire in southwestern Colorado, USA, where much of the burn area had been affected by severe spruce beetle outbreaks in the decade prior to the fire. Vegetation recovery was assessed using the Landsat-derived Normalized Difference Vegetation Index (NDVI) two years after the fire, which occurred in 2013. Beetle outbreak severity, defined as the basal area of beetle-killed trees within Landsat pixels, was estimated using vegetation index differences (dVIs) derived from pre-outbreak and post-outbreak Landsat images. Of the seven dVIs tested, the change in Normalized Difference Moisture Index (dNDMI) was most strongly correlated with field measurements of beetle-killed basal area (R2 = 0.66). dNDMI was included as an explanatory variable in sequential autoregressive (SAR) models of NDVI2015. Models also included pre-disturbance NDVI, topography, and weather conditions at the time of burning as covariates. SAR results showed a significant correlation between NDVI2015 and dNDMI, with more severe spruce beetle outbreaks corresponding to reduced post-fire vegetation cover. The correlation was stronger for models which were limited to locations in the red stage of outbreak (outbreak ≤ 5 years old at the time of fire) than for models of gray-stage locations (outbreak > 5 years old at the time of fire). These results indicate that vegetation recovery processes may be negatively impacted by severe spruce beetle outbreaks occurring within a decade of stand-replacing wildfire.