Kirsten M. de Beurs

Spatio-Temporal Statistical Methods for Modelling Land Surface Phenology

This chapter surveys 12 different spatio-temporal statistical methods to determine the start and end of the growing season using a time series of satellite images. In the first section of the chapter, we divided the methods into four categories: thresholds, derivatives, smoothing functions, and fitted models. The general use, advantages, and potential limitations of each method are discussed. In the second section of the chapter, a case study is presented to highlight one method from each category. The four study areas range from the Northwest Territories in Canada to the winter wheat areas in south-central Kansas. We concluded the case study with a discussion of the differences in results for the four methods. The chapter is finished with a synopsis discussing the use of nomenclature, the problems with a lack of statistical error structure from most methods, and the perennial issue of oversmoothing.

Kriging and thin plate splines for mapping climate variables

Four forms of kriging and three forms of thin plate splines are discussed in this paper to predict monthly maximum temperature and monthly mean precipitation in Jalisco State of Mexico. Results show that techniques using elevation as additional information improve the prediction results considerably. From these techniques, trivariate regression-kriging and trivariate thin plate splines performed best. The results of monthly maximum temperature are much clearer than the results of monthly mean precipitation, because the modeling of precipitation is more troublesome due to higher variability in the data and their non-Gaussian character.

Remote Sensing of Land Surface Phenology: A Prospectus

The process of observing land surface phenology (or LSP) using remote sensing satellites is fundamentally different from ground level observation of phenophase transition of specific organisms. The scale disparity between the spatial extent of the organisms and the spatial resolution of the sensor leads to an ill-defined mixture of target and background or signal and noise. Much progress has been made in the monitoring and modeling of land surface phenologies over the past decade. The chapter first provides a brief overview of land surface phenology, starting with the Landsat 1 in 1972, and then proceeds to a survey of current LSP products. The problem of indistinct phenometrics in remote sensing data is considered and the alternative phenometrics derived from the convex quadratic model are presented with an application in the North American Great Plains using MODIS data from 2001 to 2012. The chapter concludes with a view forward to outstanding challenges for LSP research in the coming decade.

Combined analysis of land cover change and NDVI trends in the Northern Eurasian grain belt

We present an approach to regional environmental monitoring in the Northern Eurasian grain belt combining time series analysis of MODIS normalized difference vegetation index (NDVI) data over the period 2001-2008 and land cover change (LCC) analysis of the 2001 and 2008 MODIS Global Land Cover product (MCD12Q1). NDVI trends were overwhelmingly negative across the grain belt with statistically significant (p⩽0.05) positive trends covering only 1% of the land surface. LCC was dominated by transitions between three classes; cropland, grassland, and a mixed cropland/natural vegetation mosaic. Combining our analyses of NDVI trends and LCC, we found a pattern of agricultural abandonment (cropland to grassland) in the southern range of the grain belt coinciding with statistically significant (p⩽0.05) negative NDVI trends and likely driven by regional drought. In the northern range of the grain belt we found an opposite tendency toward agricultural intensification; in this case, represented by LCC from cropland mosaic to pure cropland, and also associated with statistically significant (p⩽0.05) negative NDVI trends. Relatively small clusters of statistically significant (p⩽0.05) positive NDVI trends corresponding with both localized land abandonment and localized agricultural intensification show that land use decision making is not uniform across the region. Land surface change in the Northern Eurasian grain belt is part of a larger pattern of land cover land use change (LCLUC) in Eastern Europe, Russia, and former territories of the Soviet Union following realignment of socialist land tenure and agricultural markets. Here, we show that a combined analysis of LCC and NDVI trends provides a more complete picture of the complexities of LCLUC in the Northern Eurasian grain belt, involving both broader climatic forcing, and narrower anthropogenic impacts, than might be obtained from either analysis alone.

Land surface anomalies preceding the 2010 Russian heat wave and a link to the North Atlantic oscillation

The Eurasian wheat belt (EWB) spans a region across Eastern Ukraine, Southern Russia, and Northern Kazakhstan; accounting for nearly 15% of global wheat production. We assessed land surface conditions across the EWB during the early growing season (April–May–June; AMJ) leading up to the 2010 Russian heat wave, and over a longer-term period from 2000 to 2010. A substantial reduction in early season values of the normalized difference vegetation index occurred prior to the Russian heat wave, continuing a decadal decline in early season primary production in the region. In 2010, an anomalously cold winter followed by an abrupt shift to a warmer-than-normal early growing season was consistent with a persistently negative phase of the North Atlantic oscillation (NAO). Regression analyses showed that early season vegetation productivity in the EWB is a function of both the winter (December–January–February; DJF) and AMJ phases of the NAO. Land surface anomalies preceding the heat wave were thus consistent with highly negative values of both the DJF NAO and AMJ NAO in 2010.

Complexity metrics to quantify semantic accuracy in segmented LANDSAT images

This paper addresses semantic accuracy in relation to images obtained with remote sensing. Semantic accuracy is defined in terms of map complexity. Complexity metrics are applied as a metric to measure complexity. The idea is that a homogeneous map of a low complexity is of a high semantic accuracy. In this study, complexity metrics like aggregation index, fragmentation index and patch size are applied on two images with different objectives, one from an agricultural area in the Netherlands, and one from a rural area in Kazakhstan. Images are segmented first using region merging segmentation. Effects on metrics and semantic accuracy are discussed. On the basis of well‐defined subsets, we conclude that the complexity metrics are suitable to quantify the semantic accuracy of the map. Segmentation is the most useful for an agricultural area including various agricultural fields. Metrics are mutually comparable being highly correlated, but showing some different aspects in quantifying map homogeneity and identifying objects of a high semantic accuracy.

Use of Landsat and MODIS data to remotely estimate Russia’s sown area

The intensity of crop management is one of the most important management decisions that affect soil carbon stocks in croplands. In this study, we use satellite data at two spatial resolutions (30 m Landsat and 500 m MODIS) and field observations to determine arable lands in a portion of the Russian grain belt. Once arable lands are established, we map cropping intensity between 2002 and 2009 to get a better understanding of the activity occurring on arable lands. Our arable land estimates compare favourably with the 2006 All-Russia Agricultural Census. We also compare three global data sets that quantify croplands against the census data. Finally, we show that our cropping intensity map compares very well to the available regional statistical data. We reveal that areas in the southern regions of Russia are successfully cropped during fewer years than more centrally located areas.

Long-term impacts of land cover changes on stream channel loss.

Land cover change and stream channel loss are two related global environmental changes that are expanding and intensifying. Here, we examine how different types and transitions of land cover change impact stream channel loss across a large urbanizing watershed. We present historical land cover in the 666-km(2) Lake Thunderbird watershed in central Oklahoma (USA) over a 137 year period and coinciding stream channel length changes for the most recent 70 years of this period. Combining these two datasets allowed us to assess the interaction of land cover changes with stream channel loss. Over this period, the upper third of the watershed shifted from predominantly native grassland to an agricultural landscape, followed by widespread urbanization. The lower two-thirds of the watershed changed from a forested landscape to a mosaic of agriculture, urban, forest, and open water. Most channel length lost in the watershed over time was replaced by agriculture. Urban development gradually increased channel loss and disconnection from 1942 to 2011, particularly in the headwaters. Intensities of channel loss for both agriculture and urban increased over time. The two longest connected segments of channel loss came from the creation of two large impoundments, resulting in 46 km and 25 km of lost stream channel, respectively. Overall, the results from this study demonstrate that multiple and various land-use changes over long time periods can lead to rapid losses of large channel lengths as well as gradual (but increasing) losses of small channel lengths across all stream sizes. When these stream channel losses are taken into account, the environmental impacts of anthropogenic land-use change are compounded.

Recent trends in agricultural production of Africa based on AVHRR NDVI time series

African agriculture is expected to be hard-hit by ongoing climate change. Effects are heterogeneous within the continent, but in some regions resulting production declines have already impacted food security. Time series of remote sensing data allow us to examine where persistent changes occur. In this study, we propose to examine recent trends in agricultural production using 26 years of NDVI data. We use the 8-km resolution AVHRR NDVI 15-day composites of the GIMMS group (1981-2006). Temporal data-filtering is applied using an iterative Savitzky-Golay algorithm to remove noise in the time series. Except for some regions with persistent cloud cover, this filter produced smooth profiles. Subsequently two methods were used to extract phenology indicators from the profiles for each raster cell. These indicators include start of season, length of season, time of maximum NDVI, maximum NDVI, and cumulated NDVI over the season. Having extracted the indicators for every year, we aggregate them for agricultural areas at sub-national level using a crop mask. The aggregation was done to focus the analysis on agriculture, and allow future comparison with yield statistics. Trend analysis was performed for yearly aggregated indicators to assess where persistent change occurred during the 26-year period. Results show that the phenology extraction method chosen has an important influence on trend outcomes. Consistent trends suggest a rising yield trend for 500-1100 mm rainfall zones ranging from Senegal to Sudan. Negative yield trends are expected for the southern Atlantic coast of West Africa, and for western Tanzania.

Phenological Response of an Arizona Dryland Forest to Short-Term Climatic Extremes

Baseline information about dryland forest phenology is necessary to accurately anticipate future ecosystem shifts. The overarching goal of our study was to investigate the variability of vegetation phenology across a dryland forest landscape in response to climate alterations. We analyzed the influence of site characteristics and climatic conditions on the phenological patterns of an Arizona, USA, ponderosa pine (Pinus ponderosa) forest during a five-year period (2005 to 2009) that encompassed extreme wet and dry precipitation regimes. We assembled 80 synthetic Landsat images by applying the spatial and temporal adaptive reflectance fusion method (STARFM) to 500 m MODIS and 30 m Landsat-5 Thematic Mapper (TM) data. We tested relationships between site characteristics and the timing of peak Normalized Difference Vegetation Index (NDVI) to assess the effect of climatic stress on the green-up of individual pixels during or after the summer monsoon. Our results show that drought-induced stress led to a fragmented phenological response that was highly dependent on microsite parameters, as both the spatial autocorrelation of peak timing and the number of significant site variables increased during the drought year. Pixels at lower elevations and with higher proportions of herbaceous vegetation were more likely to exhibit dynamic responses to changes in precipitation conditions. Our study demonstrates the complexity of responses within dryland forest ecosystems and highlights the need for standardized monitoring of phenology trends in these areas. The spatial and temporal variability of phenological signals may provide a quantitative solution to the problem of how to evaluate dryland land surface trends across time.