Breaks For Additive Season And Trend Research Articles

Quantifying the impacts of climatic trend and fluctuation on crop yields in northern China.

Climate change plays a critical role in crop yield variations, which has attracted a great deal of concern worldwide. However, the mechanisms of how climatic trend and fluctuations affect crop yields are not well understood and need to be further investigated. Thus, using the GIS-based Environmental Policy Integrated Climate (EPIC) model, we simulated the yields of major crops (i.e., wheat, maize, and rice) and evaluated the impacts of climatic factors on crop yields in the Agro-Pastoral Transitional Zone (APTZ) of northern China between 1980 and 2010. The partial least squares regression model was used to assess the contribution rates of climatic factors (i.e., precipitation, photosynthetically active radiation (PAR), minimum temperature (T min), maximum temperature (T max)) to the variation of crop yields. The Breaks for Additive Season and Trend (BFAST) model was adopted to decompose the climate factors into trend and fluctuation components, and the relative contributions of climate trend and fluctuation were then evaluated. The results indicated that the contributions of climatic factors to yield variations of wheat, maize, and rice were 31.7, 37.7, and 23.1%, respectively. That is, climate change had larger impacts on maize than wheat and rice. More cultivated areas were significantly and positively correlated with precipitation than with other climatic factors due to the limited precipitation in the APTZ. Also, climatic trend component had positive impacts on crop yields in the whole region, whereas the climate fluctuation was associated mainly with the areas where the crop yields decreased. This study helps improve our understanding of the mechanisms of climate change impacts on crop yields, and provides useful scientific information for designing regional-scale strategies of adaptation to climate change.

Breaks in MODIS time series portend vegetation change: verification using long‐term data in an arid grassland ecosystem

Frequency and severity of extreme climatic events are forecast to increase in the 21st century. Predicting how managed ecosystems may respond to climatic extremes is intensified by uncertainty associated with knowing when, where, and how long effects of extreme events will be manifest in an ecosystem. In water-limited ecosystems with high inter-annual variability in rainfall, it is important to be able to distinguish responses that result from seasonal fluctuations in rainfall from long-term directional increases or decreases in precipitation. A tool that successfully distinguishes seasonal from directional biomass responses would allow land managers to make informed decisions about prioritizing mitigation strategies, allocating human resource monitoring efforts, and mobilizing resources to withstand extreme climatic events. We leveraged long-term observations (2000-2013) of quadrat-level plant biomass at multiple locations across a semiarid landscape in southern New Mexico to verify the use of Normalized Difference Vegetation Index (NDVI) time series derived from 250-m Moderate Resolution Imaging Spectroradiometer (MODIS) data as a proxy for changes in aboveground productivity. This period encompassed years of sustained drought (2000-2003) and record-breaking high rainfall (2006 and 2008) followed by subsequent drought years (2011 through 2013) that resulted in a restructuring of plant community composition in some locations. Our objective was to decompose vegetation patterns derived from MODIS NDVI over this period into contributions from (1) the long-term trend, (2) seasonal cycle, and (3) unexplained variance using the Breaks for Additive Season and Trend (BFAST) model. BFAST breakpoints in NDVI trend and seasonal components were verified with field-estimated biomass at 15 sites that differed in species richness, vegetation cover, and soil properties. We found that 34 of 45 breaks in NDVI trend reflected large changes in mean biomass and 16 of 19 seasonal breaks accompanied changes in the contribution to biomass by perennial and/or annual grasses. The BFAST method using satellite imagery proved useful for detecting previously reported ground-based changes in vegetation in this arid ecosystem. We demonstrate that time series analysis of NDVI data holds potential for monitoring landscape condition in arid ecosystems at the large spatial scales needed to differentiate responses to a changing climate from responses to seasonal variability in rainfall.

Leveraging Multi-Sensor Time Series Datasets to Map Short- and Long-Term Tropical Forest Disturbances in the Colombian Andes

The spatial distribution of disturbances in Andean tropical forests and protected areas has commonly been calculated using bi or tri-temporal analysis because of persistent cloud cover and complex topography. Long-term trends of vegetative decline (browning) or improvement (greening) have thus not been evaluated despite their importance for assessing conservation strategy implementation in regions where field-based monitoring by environmental authorities is limited. Using Colombia’s Cordillera de los Picachos National Natural Park as a case study, we provide a temporally rigorous assessment of regional vegetation change from 2001–2015 with two remote sensing-based approaches using the Breaks For Additive Season and Trend (BFAST) algorithm. First, we measured long-term vegetation trends using a Moderate Resolution Imaging Spectroradiometer (MODIS)-based Multi-Angle Implementation of Atmospheric Correction (MAIAC) time series, and, second, we mapped short-term disturbances using all available Landsat images. MAIAC-derived trends indicate a net greening in 6% of the park, but in the surrounding 10 km area outside of the park, a net browning trend prevails at 2.5%. We also identified a 12,500 ha area within Picachos (4% of the park’s total area) that has shown at least 13 years of consecutive browning, a result that was corroborated with our Landsat-based approach that recorded a 12,642 ha (±1440 ha) area of disturbed forest within the park. Landsat vegetation disturbance results had user’s and producer’s accuracies of 0.95 ± 0.02 and 0.83 ± 0.18, respectively, and 75% of Landsat-detected dates of disturbance events were accurate within ±6 months. This study provides new insights into the contribution of short-term disturbance to long-term trends of vegetation change, and offers an unprecedented perspective on the distribution of small-scale disturbances over a 15-year period in one of the most inaccessible national parks in the Andes.

Detecting spatiotemporal changes of peak foliage coloration in deciduous and mixedforests across the Central and Eastern United States

The timing of fall foliage coloration, especially peak coloration, is of great importance to the climate change research community as it has implications for carbon storage in forests. However, its long-term variation and response to climate change are poorly understood. To address this issue, we examined the long-term trends and breakpoints in satellite derived peak coloration onset from 1982 to 2014 using an innovative approach that combines Singular Spectrum Analysis (SSA) with Breaks for Additive Seasonal and Trend (BFAST). The peak coloration trend was then evaluated using both field foliage coloration observations and flux tower measurements. Finally, interannual changes in peak coloration onset were correlated with temperature and precipitation variation. Results showed that temporal trends in satellite-derived peak coloration onset were comparable with both field observations and flux tower measurements of gross primary productivity. Specifically, a breakpoint in long-term peak coloration onset was detected in 25% of pixels which were mainly distributed at latitudes north of 37°N. The breakpoint tended to occur between 1998 and 2004. Peak coloration onset was delayed before the breakpoint while it was transformed to an early trend after the breakpoint in nearly all pixels. The remaining 75% of pixels exhibited monotonic trends, 35% of which revealed a late trend and 40% an early trend. The results indicate that the onset of peak coloration experienced a late trend during the 1980s and 1990s in most deciduous and mixed forests. However, the trend was reversed during the most recent decade when the timing of peak coloration became earlier. The onset of peak coloration was significantly correlated with late summer and autumn temperature in 55.5% of pixels from 1982 to 2014. This pattern of temperature impacts was also verified using field observations and flux tower measurements. In the remaining 44.5% of pixels, 12.2% of pixels showed significantly positive correlation between the onset of peak coloration and cumulative precipitation during late summer and autumn period from 1982 to 2014. Our findings can improve understanding of the impact of changes in autumn phenology on carbon uptake in forests, which in turn facilitate more reliable measures of carbon dynamics in vegetation–climate interactions models.

Monitoring ecosystem dynamics in northwestern Ethiopia using NDVI and climate variables to assess long term trends in dryland vegetation variability

Dryland ecosystems are highly vulnerable to environmental changes. Monitoring is vital in order to evaluate their response to fluctuating rainfall and temperature patterns for long-term ecosystem safeguarding. Monitoring of long term changes of normalized difference vegetation index (NDVI) and climate variables are fundamental for better understanding of change trajectories in dryland ecosystem, and to ascertain their potential interaction with anthropogenic drivers. In this study, we identify determinant factors of dryland changes by using MODIS NDVI, precipitation and temperature data for Breaks for Additive Seasonal and Trend (BFAST) and Mann Kendall test statistic. BFAST predicts iteratively time and number of changes within a time series data to depict the size and direction of changes. Analysis of NDVI, precipitation and temperature time series data showed substantial changes during the study period of 2000–2014. There is a reduction trend in vegetation showed by the decline in NDVI, with significant breakpoints till 2009 and recovery afterwards, without a significant change in annual trends of precipitation (α < 0.05) for the same study period. Furthermore 2 positive climate trends were founded: a) a significant positive trend on long term annual rainfall during the main rainy seasons and; 2) a significant (α < 0.05) annual increment of the long term mean minimum and mean maximum temperature of 0.03 °C/year and 0.04 °C/year, respectively. This assessment showed that climate variables cannot be considered as the main factors in explaining the observed patterns of vegetation dynamics. Seasonal and interannual precipitation changes have a lower weight as driving factors for the reduction in vegetation trends. Hence, the decline in vegetation productivity of the region can be attributed to the increasing pressure of human activities.

ON-LINE CHANGE MONITORING WITH TRANSFORMED MULTI-SPECTRAL TIME SERIES, A STUDY CASE IN TROPICAL FOREST

Abstract. In recent years, the methods for detecting structural changes in time series have been adapted for forest disturbance monitoring using satellite data. The BFAST (Breaks For Additive Season and Trend) Monitor framework, which detects forest cover disturbances from satellite image time series based on empirical fluctuation tests, is particularly used for near real-time deforestation monitoring, and it has been shown to be robust in detecting forest disturbances. Typically, a vegetation index that is transformed from spectral bands into feature space (e.g. normalised difference vegetation index (NDVI)) is used as input for BFAST Monitor. However, using a vegetation index for deforestation monitoring is a major limitation because it is difficult to separate deforestation from multiple seasonality effects, noise, and other forest disturbance. In this study, we address such limitation by exploiting the multi-spectral band of satellite data. To demonstrate our approach, we carried out a case study in a deciduous tropical forest in Bolivia, South America. We reduce the dimensionality from spectral bands, space and time with projective methods particularly the Principal Component Analysis (PCA), resulting in a new index that is more suitable for change monitoring. Our results show significantly improved temporal delay in deforestation detection. With our approach, we achieved a median temporal lag of 6 observations, which was significantly shorter than the temporal lags from conventional approaches (14 to 21 observations).

Mapping Clearances in Tropical Dry Forests Using Breakpoints, Trend, and Seasonal Components from MODIS Time Series: Does Forest Type Matter?

Tropical environments present a unique challenge for optical time series analysis, primarily owing to fragmented data availability, persistent cloud cover and atmospheric aerosols. Additionally, little is known of whether the performance of time series change detection is affected by diverse forest types found in tropical dry regions. In this paper, we develop a methodology for mapping forest clearing in Southeast Asia using a study region characterised by heterogeneous forest types. Moderate Resolution Imaging Spectroradiometer (MODIS) time series are decomposed using Breaks For Additive Season and Trend (BFAST) and breakpoints, trend, and seasonal components are combined in a binomial probability model to distinguish between cleared and stable forest. We found that the addition of seasonality and trend information improves the change model performance compared to using breakpoints alone. We also demonstrate the value of considering forest type in disturbance mapping in comparison to the more common approach that combines all forest types into a single generalised forest class. By taking a generalised forest approach, there is less control over the error distribution in each forest type. Dry-deciduous and evergreen forests are especially sensitive to error imbalances using a generalised forest model i.e., clearances were underestimated in evergreen forest, and overestimated in dry-deciduous forest. This suggests that forest type needs to be considered in time series change mapping, especially in heterogeneous forest regions. Our approach builds towards improving large-area monitoring of forest-diverse regions such as Southeast Asia. The findings of this study should also be transferable across optical sensors and are therefore relevant for the future availability of dense time series for the tropics at higher spatial resolutions.

Influence of climate variability on anchovy reproductive timing off northern Chile

We investigated the relationship between environmental variables and the Gonadosomatic Monthly Mean (GMM) index of anchovy (Engraulis ringens) to understand how the environment affects the dynamics of anchovy reproductive timing. The data examined corresponds to biological information collected from samples of the landings off northern Chile (18°21′S, 24°00′S) during the period 1990–2010. We used the Humboldt Current Index (HCI) and the Multivariate ENSO Index (MEI), which combine several physical-oceanographic factors in the Tropical and South Pacific regions. Using the GMM index, we studied the dynamics of anchovy reproductive timing at different intervals of length, specifically females with a length between 11.5 and 14cm (medium class) and longer than 14cm (large class). Seasonal Autoregressive Integrated Mobile Average (SARIMA) was used to predict missing observations. The trends of the environment and reproductive indexes were explored via the Breaks For Additive Season and Trend (BFAST) statistical technique and the relationship between these indexes via cross-correlation functions (CCF) analysis. Our results showed that the habitat of anchovy switched from cool to warm condition, which also influenced gonad development. This was revealed by two and three significant changes (breaks) in the trend of the HCI and MEI indexes, and two significant breaks in the GMM of each time series of anchovy females (medium and large). Negative cross-correlation between the MEI index and GMM of medium and large class females was found, indicating that as the environment gets warmer (positive value of MEI) a decrease in the reproductive activity of anchovy can be expected. Correlation between the MEI index and larger females was stronger than with medium females. Additionally, our results indicate that the GMM index of anchovy for both length classes reaches two maximums per year; the first from August to September and the second from December to January. The intensity (maximum GMM values at rise point) of reproductive activity was not equal though, with the August–September peak being the highest. We also discuss how the synchronicity between environment and reproductive timing, the negative correlation found between MEI and GMM indexes, and the two increases per year of anchovy GMM relate to previous studies. Based on these findings we propose ways to advance in the understanding of how anchovy synchronize gonad development with the environment.

Performance of vegetation indices from Landsat time series in deforestation monitoring

The performance of Landsat time series (LTS) of eight vegetation indices (VIs) was assessed for monitoring deforestation across the tropics. Three sites were selected based on differing remote sensing observation frequencies, deforestation drivers and environmental factors. The LTS of each VI was analysed using the Breaks For Additive Season and Trend (BFAST) Monitor method to identify deforestation. A robust reference database was used to evaluate the performance regarding spatial accuracy, sensitivity to observation frequency and combined use of multiple VIs. The canopy cover sensitive Normalized Difference Fraction Index (NDFI) was the most accurate. Among those tested, wetness related VIs (Normalized Difference Moisture Index (NDMI) and the Tasselled Cap wetness (TCw)) were spatially more accurate than greenness related VIs (Normalized Difference Vegetation Index (NDVI) and Tasselled Cap greenness (TCg)). When VIs were fused on feature level, spatial accuracy was improved and overestimation of change reduced. NDVI and NDFI produced the most robust results when observation frequency varies.

ESTIMATING RELIABILITY OF DISTURBANCES IN SATELLITE TIME SERIES DATA BASED ON STATISTICAL ANALYSIS

Abstract. Normally, the status of land cover is inherently dynamic and changing continuously on temporal scale. However, disturbances or abnormal changes of land cover — caused by such as forest fire, flood, deforestation, and plant diseases — occur worldwide at unknown times and locations. Timely detection and characterization of these disturbances is of importance for land cover monitoring. Recently, many time-series-analysis methods have been developed for near real-time or online disturbance detection, using satellite image time series. However, the detection results were only labelled with “Change/ No change” by most of the present methods, while few methods focus on estimating reliability (or confidence level) of the detected disturbances in image time series. To this end, this paper propose a statistical analysis method for estimating reliability of disturbances in new available remote sensing image time series, through analysis of full temporal information laid in time series data. The method consists of three main steps. (1) Segmenting and modelling of historical time series data based on Breaks for Additive Seasonal and Trend (BFAST). (2) Forecasting and detecting disturbances in new time series data. (3) Estimating reliability of each detected disturbance using statistical analysis based on Confidence Interval (CI) and Confidence Levels (CL). The method was validated by estimating reliability of disturbance regions caused by a recent severe flooding occurred around the border of Russia and China. Results demonstrated that the method can estimate reliability of disturbances detected in satellite image with estimation error less than 5% and overall accuracy up to 90%.

ESTIMATING RELIABILITY OF DISTURBANCES IN SATELLITE TIME SERIES DATA BASED ON STATISTICAL ANALYSIS

Normally, the status of land cover is inherently dynamic and changing continuously on temporal scale. However, disturbances or abnormal changes of land cover — caused by such as forest fire, flood, deforestation, and plant diseases — occur worldwide at unknown times and locations. Timely detection and characterization of these disturbances is of importance for land cover monitoring. Recently, many time-series-analysis methods have been developed for near real-time or online disturbance detection, using satellite image time series. However, the detection results were only labelled with “Change/ No change” by most of the present methods, while few methods focus on estimating reliability (or confidence level) of the detected disturbances in image time series. To this end, this paper propose a statistical analysis method for estimating reliability of disturbances in new available remote sensing image time series, through analysis of full temporal information laid in time series data. The method consists of three main steps. (1) Segmenting and modelling of historical time series data based on Breaks for Additive Seasonal and Trend (BFAST). (2) Forecasting and detecting disturbances in new time series data. (3) Estimating reliability of each detected disturbance using statistical analysis based on Confidence Interval (CI) and Confidence Levels (CL). The method was validated by estimating reliability of disturbance regions caused by a recent severe flooding occurred around the border of Russia and China. Results demonstrated that the method can estimate reliability of disturbances detected in satellite image with estimation error less than 5% and overall accuracy up to 90%.

Spatio-temporal change detection from multidimensional arrays: Detecting deforestation from MODIS time series

Growing availability of long-term satellite imagery enables change modeling with advanced spatio-temporal statistical methods. Multidimensional arrays naturally match the structure of spatio-temporal satellite data and can provide a clean modeling process for complex spatio-temporal analysis over large datasets. Our study case illustrates the detection of breakpoints in MODIS imagery time series for land cover change in the Brazilian Amazon using the BFAST (Breaks For Additive Season and Trend) change detection framework. BFAST includes an Empirical Fluctuation Process (EFP) to alarm the change and a change point time locating process. We extend the EFP to account for the spatial autocorrelation between spatial neighbors and assess the effects of spatial correlation when applying BFAST on satellite image time series. In addition, we evaluate how sensitive EFP is to the assumption that its time series residuals are temporally uncorrelated, by modeling it as an autoregressive process. We use arrays as a unified data structure for the modeling process, R to execute the analysis, and an array database management system to scale computation. Our results point to BFAST as a robust approach against mild temporal and spatial correlation, to the use of arrays to ease the modeling process of spatio-temporal change, and towards communicable and scalable analysis.

Conservation-induced resettlement as a driver of land cover change in India: An object-based trend analysis

Located in the foothills of the Indian Himalaya, Rajaji National Park was established to protect and enhance the habitat of the Asian elephant (Elephas maximus) and tiger (Panthera tigris). In 2002 the Van Gujjars, indigenous forest pastoralists, were voluntarily resettled from the Chilla Range (an administrative unit of Rajaji National Park) to Gaindikhata, a nearby area where they were granted land for agriculture. In this study we used a variety of remote sensing approaches to identify changes in land cover associated with the resettlement. The methods comprise two main approaches. First, we used object-based image analysis (OBIA) to identify the pre-resettlement land cover classes of use areas (representing agricultural expansion and adjacent areas of grazing, and collection of fuelwood and fodder) and recovery areas (representing areas where settlements were removed, and the adjacent areas of resource use). Secondly, we used trend analysis to assess the gradual and abrupt changes in vegetation that took place in use and recovery areas. To conduct the trend analysis we used BFAST (Breaks For Additive Season and Trend), which separates seasonal variation from long-term trends, and identifies breaks that can be linked back to disturbances or land cover changes. We found that the OBIA classification yielded high average class accuracies, and we were able to make class distinctions that would have been difficult to make using a traditional pixel-based approach. Pre-resettlement, the recovery areas were classified as mixed forest and riparian vegetation. In contrast, the use areas were classified primarily as grass dominated, brush dominated, and plantation forest, and were located relatively far away from riparian areas. Following the resettlement, the trend analysis showed a sudden change in the seasonal variation of NDVI in areas converted to agriculture. Areas neighboring the new agricultural land experienced sudden decreases in NDVI, suggestive of disturbances, at a higher rate than the same land cover types elsewhere. At the same time, these neighboring areas experienced a gradual overall increase in NDVI which could be caused by an expansion of leafy invasive shrubs such as Lantana camara in areas heavily used for biomass collection. The recovery areas also experienced a gradual increase in NDVI as well as sudden breaks to this trend, but we lacked evidence to connect these changes to the resettlement. Our findings support the claim that the resettlement has shifted pressure from more ecologically valuable to less ecologically valuable land cover types, and suggest that to some degree resource use pressure has increased outside the park. The study employs a novel synthesis of OBIA and trend analysis that could be applied to land change studies more broadly.

Drought Detection of West Java's Paddy Field Using MODIS EVI Satellite Images (Case Study: Rancaekek and Rancaekek Wetan)

Nowadays, drought phenomenon occurred in several area in Indonesia. The length of the dry season, especially in the southern equatorial allegedly caused by El Nino phenomenon. This causes crop failures in many center area of agriculture. West Java Province as one of the centers of agricultural activities in Indonesia experienced a severe drought within a period of 6 months (April-September in 2003). The monitoring of drought is useful to understanding the characterization of drought itself. In the next period, we can decide what should we do to decreased the impact of this phenomenon. The study aimed to implement Breaks for Additive Seasonal and Trend (BFAST) algorithm for detecting and monitoring paddy field areas experiencing drought in The West Java during the period 2000-2015. This study used remote sensing data to study the response of vegetation on the drought phenomenon. MODIS EVI time series were decomposed into seasonal, trend, and remainder component using BFAST which enables the detection of trend changes within the time series. The result of study shows that BFAST able to detect drought in MODIS EVI time series. The result also compared to a new drought index, called SPEI.

Analysis of Agricultural Land Use Changes in Jombang Regency, East Java, Indonesia Using BFAST Method

Jombang is one of the regency in East Java province which is dominated by agricultural lands. This is caused by high levels of soil fertility due to volcanic eruption in the past. Recently, conversion of agricultural land to non-agricultural land has been increasing as accelerated by the growth of industrial sector. If this conversion occurs continuously, it will lead to decreasing of the agricultural production in Jombang. Therefore, it is an urgent to provide accurate and up-to-date information of agricultural land use changes. This research will investigate the temporal vegetation dynamics in paddy fields and to detect their changes in Jombang Regency. The Breaks For Additive Season and Trend (BFAST) method was used to identify the change pattern events in this study. This method can estimate the time and amount of the abrupt change in the time series datasets. Based on the temporal vegetation pattern, the agriculture land use in Jombang could be categorized into ten typical cropping patterns, namely: paddy-paddy-paddy, paddy-paddy-secondary crops, paddy-paddy-bareland, paddy-secondary crops-bareland, paddy-secondary crops-paddy, paddy-bareland-paddy, forest, sugarcane, non-agriculture, and mixed vegetation. Moreover, the results show that by the BFAST change detection, the change distribution of agriculture land use from the past ten years can be detected.

Reconstructing land use history from Landsat time-series: Case study of a swidden agriculture system in Brazil

We developed a method to reconstruct land use history from Landsat images time-series. The method uses a breakpoint detection framework derived from the econometrics field and applicable to time-series regression models. The Breaks For Additive Season and Trend (BFAST) framework is used for defining the time-series regression models which may contain trend and phenology, hence appropriately modelling vegetation intra and inter-annual dynamics. All available Landsat data are used for a selected study area, and the time-series are partitioned into segments delimited by breakpoints. Segments can be associated to land use regimes, while the breakpoints then correspond to shifts in land use regimes. In order to further characterize these shifts, we classified the unlabelled breakpoints returned by the algorithm into their corresponding processes. We used a Random Forest classifier, trained from a set of visually interpreted time-series profiles to infer the processes and assign labels to the breakpoints. The whole approach was applied to quantifying the number of cultivation cycles in a swidden agriculture system in Brazil (state of Amazonas). Number and frequency of cultivation cycles is of particular ecological relevance in these systems since they largely affect the capacity of the forest to regenerate after land abandonment. We applied the method to a Landsat time-series of Normalized Difference Moisture Index (NDMI) spanning the 1984–2015 period and derived from it the number of cultivation cycles during that period at the individual field scale level. Agricultural fields boundaries used to apply the method were derived using a multi-temporal segmentation approach. We validated the number of cultivation cycles predicted by the method against in-situ information collected from farmers interviews, resulting in a Normalized Residual Mean Squared Error (NRMSE) of 0.25. Overall the method performed well, producing maps with coherent spatial patterns. We identified various sources of error in the approach, including low data availability in the 90s and sub-object mixture of land uses. We conclude that the method holds great promise for land use history mapping in the tropics and beyond.

Using spatial context to improve early detection of deforestation from Landsat time series

Mapping deforestation using medium spatial resolution satellite data (e.g. Landsat) is increasingly shifting from decadal and annual scales to sub-annual scales in recent years, but this shift has brought new challenges on how to account for seasonality in the satellite data when detecting deforestation. A seasonal model is typically used to account for seasonality, but fitting a seasonal model is difficult when there are not enough data in the time series. Here, we propose a new approach that reduces seasonality in satellite image time series using spatial context. With this spatial context approach, each pixel value in the image is spatially normalised using the median value calculated from neighbouring pixels whose pixel values are above the 90th percentile. Using Landsat data, we compared our spatial context approach to a seasonal model approach at a humid tropical forest in Brazil and a dry tropical forest with strong seasonality in Bolivia. After reducing seasonal variations in Landsat data, we detected deforestation from the same data using the Breaks For Additive Season and Trend (BFAST) method. We show that, in dry tropical forest, deforestation events are detected much earlier when the spatial context approach is used to reduce seasonal variations in Landsat data than when a seasonal model is used. In the dry tropical forest, the median temporal detection delay for deforestation from the spatial context approach was two observations, seven times shorter than the median temporal detection delay from the seasonal model approach (15 observations). In the humid tropical forest, the difference in the temporal detection delay between the spatial context and seasonal model approach was not significant. The differences in overall spatial accuracy between the spatial context and seasonal model were also not significant in both dry and humid tropical forests. The main benefit for using spatial context is early detection of deforestation events in forests with strong seasonality. Therefore, the spatial context approach we propose here provides opportunity to monitor deforestation events in dry tropical forests at sub-annual scales using Landsat data.

Temperature and Precipitation Variability and Its Effects on Streamflow in the Upstream Regions of the Lancang–Mekong and Nu–Salween Rivers

Abstract Hydrological regimes of alpine rivers are highly sensitive to climate variability/change. Temperature and precipitation variability and its effects on streamflow in the upstream regions of the Lancang–Mekong River (LMR) and Nu–Salween River (NSR) are examined in this study based on long-term observational data from 16 meteorological stations and 2 hydrological stations between the 1950s and 2010. This study employs the Mann–Kendall nonparametric test, together with the trend-free prewhitening (TFPW) approach to test trends and the Breaks For Additive Season and Trend (BFAST) method to detect abrupt changes in the hydrometeorological time series. The relations between air temperature, precipitation, and streamflow trends are assessed using random forest regression. The results show significant climate warming and related prevalent positive precipitation trends both at the annual and seasonal scale. A substantial precipitation increase paralleling climate warming, especially in spring, was also observed. However, no consistent abrupt change in meteorological time series was found. The increasing trends of streamflow with climate warming are seen both for the outlets of the LMR and NSR upstream regions, with the abrupt changes occurring in the mid-1960s and the late 1990s, respectively. The relation of streamflow to annual and wet season precipitation is pronounced, especially for the upstream region of the LMR with a percent variance explained of more than 65%. However, the relatively minor linkage of streamflow to air temperature and dry season precipitation may be confounded by the climate warming–driven changes in snowpack, permafrost, glacier, and evapotranspiration. These results could provide further a reference for the regional water resources management under climate change scenarios.

Detecting Change Dates from Dense Satellite Time Series Using a Sub-Annual Change Detection Algorithm

Multi-temporal satellite images are available at very high revisit frequency, allowing the characterization of land-cover transitions and trajectories in greater detail. However, most change detection methods aim to capture a snapshot of land cover, for instance on an annual scale, which do not describe changes that occurred between the annual time points. In this study, we present a sub-annual change detection (SCD) approach to detect change dates from dense satellite time series. SCD estimates change dates by analyzing differences between two consecutive annual segments in two steps. To validate the proposed method, SCD was applied to real and simulated time series of MODIS 16-day NDVI from 2000 to 2012 for MODIS tile h11v05 and a sub-area in southeast Ohio, USA. The results show that SCD can successfully detect from dense time series the dates of changes representing land-cover transitions among various vegetated land covers. In addition, SCD can achieve comparable accuracy on the date of abrupt land-cover changes for the real time series tested in the study area when compared with the trend and seasonal changes identified by a Breaks for Additive Seasonal and Trend (BFAST) method. Future effort will be given to apply the proposed approach to various remotely sensed time series for other areas with extreme climates.

Abrupt spatiotemporal land and water changes and their potential drivers in Poyang Lake, 2000–2012

Driven by various natural and anthropogenic factors, Poyang Lake, the largest freshwater lake in China, has experienced significant land use/cover changes in the past few decades. The aim of this study is to investigate the spatial–temporal patterns of abrupt changes and detect their potential drivers in Poyang Lake, using time-series Moderate Resolution Imaging Spectroradiometer (MODIS) 16-day maximum value composite vegetation indices between 2000 and 2012. The breaks for additive seasonal and trend (BFAST) method was applied to the smoothed time-series normalized difference vegetation index (NDVI), to detect the timing and magnitude of abrupt changes in the trend component. Large part of Poyang Lake (98.9% for trend component) has experienced abrupt changes in the past 13years, and the change patterns, including the distributions in timing and magnitudes of major abrupt trend changes between water bodies and land areas were clearly differentiated. Most water bodies had abrupt increasing NDVI changes between 2010 and 2011, caused by the sequential severe flooding and drought in the two years. In contrast, large parts of the surrounding land areas had abrupt decreasing NDVI changes. Large decreasing changes occurred around 2003 at the city of Nanchang, which were driven by urbanization. These results revealed spatial–temporal land cover changing patterns and potential drivers in the wetland ecosystem of Poyang Lake.