Pavel Propastin

Modifying geographically weighted regression for estimating aboveground biomass in tropical rainforests by multispectral remote sensing data

Abstract The present study uses a local regression approach for estimation of aboveground biomass (AGB) in a tropical rainforest area with highly diverse terrain conditions from remote sensing-based multi-spectral vegetation indices (VI). By incorporating altitudinal effects into the spatial weighting matrices of the common geographically weighted regression (GWR), an extended GWR model, geographically and altitudinal weighted regression (GAWR), has been developed to deal with both spatial (horizontal) and altitudinal (vertical) non-stationarity in the data set. Unlike the common GWR model, the presented GAWR approach captures both horizontal and altitudinal drifts in the relationships between aboveground biomass and remote sensing data. In order to test its improved performance, the GAWR method was compared with the traditional GWR technique and global ordinary least squares regression (OLS) in a region of mountainous tropical rainforest in Sulawesi, Indonesia. The relationships between AGB and VIs were found to be significantly spatially variable. The results showed that there were substantial benefits in capturing both horizontal and vertical non-stationarity simultaneously. The GAWR method significantly improved AGB prediction in all simulations relative to both the traditional GWR and OLS methods, as indicated by accuracy and precision statistics. From the results of empirical tests, it seems proper to say that for this data set, the GAWR model is better than the traditional GWR model.

Spatial Patterns of NDVI Variation over Indonesia and Their Relationship to ENSO Warm Events during the Period 1982–2006

The present study is based on the assumption that vegetation in Indonesia is significantly affected by climate anomalies that are related to El Nino-Southern Oscillation (ENSO) warm phases (El Nino) during the past decades. The analysis builds upon a monthly time series from the normalized difference vegetation index (NDVI) gridded data from the Advanced Very High Resolution Radiometer (AVHRR) and two ENSO proxies, namely, sea surface temperature anomalies (SSTa) and Southern Oscillation index (SOI), and aims at the analysis of the spatially explicit dimension of ENSO impact on vegetation on the Indonesian archipelago. A time series correlation analysis between NDVI anomalies and ENSO proxies for the most recent ENSO warm events (1982-2006) showed that, in general, anomalies in vegetation productivity over Indonesia can be related to an anomalous increase of SST in the eastern equatorial Pacific and to decreases in SOI, respectively. The net effect of these variations is a significant decrease in NDVI values throughout the affected areas during the ENSO warm phases. The 1982/83 ENSO warm episode was rather short but—in terms of ENSO indices—the most extreme one within the study period. The 1997/98 El Nino lasted longer but was weaker. Both events had significant impact on vegetation in terms of negative NDVI anomalies. Compared to these two major warm events, the other investigated events (1987/88, 1991/92, 1994/95, and 2002/03) had no sig- nificant effect on vegetation in the investigated region. The land cover-type specific sensitivity of vegetation to ENSO anomalies revealed thresholds of vegetation response to ENSO warm events. The results for the 1997/98 ENSO warm event confirm the hypothesis that the vulnerability of vegetated tropical land surfaces to drought conditions is considerably affected by land use intensity. In particular, it could be shown that natural forest areas are more resistant to drought stress than degraded forest areas or cropland. Comparing the spatially explicit patterns of El Nino-related vegetation variation during the major El Nino phases, the spatial distribution of affected areas reveals distinct core regions of ENSO drought impact on vegetation for Indonesia that coincide with forest conversion and agricultural intensification hot spots.

A physically based approach to model LAI from MODIS 250 m data in a tropical region

A time series of leaf area index (LAI) has been developed based on 16-day normalized difference vegetation index (NDVI) data from the Moderate Resolution Imaging Spectroradiometer (MODIS) at 250 m resolution (MOD250_LAI). The MOD250_LAI product uses a physical radiative transfer model which establishes a relationship between LAI, fraction of vegetation cover (FVC) and given patterns of surface reflectance, view-illumination conditions and optical properties of vegetation. In situ measurements of LAI and FVC made at 166 plots using hemispherical photography served for calibration of model parameters and validation of modelling results. Optical properties of vegetation cover, summarized by the light extinction coefficient, were computed at the local (pixel) level based on empirical models between ground-measured tree crown architecture at 85 sampling plots and spectral values in Landsat ETM+ bands. Influence of view-illumination conditions on optical properties of canopy was simulated by a view angle geometry model incorporating the solar zenith angle and the sensor viewing angle. The results revealed high compatibility of the produced MOD250_LAI data set with ground truth information and the 30 m resolution Landsat ETM+ LAI estimated using the similar algorithm. The produced MOD250_LAI was also compared with the global MODIS 1000-m LAI product (MOD15A2 LAI). Results show good consistency of the spatial distribution and temporal dynamics between the two LAI products. However, the results also showed that the annual LAI amplitude by the MOD15A2 product is significantly higher than by the MOD250_LAI. This higher amplitude is caused by a considerable underestimation of the tropical rainforest LAI by the MOD15A2 during the seasonal phases of low leaf production.

Modified light use efficiency model for assessment of carbon sequestration in grasslands of Kazakhstan: combining ground biomass data and remote-sensing

A modified light use efficiency (LUE) model was tested in the grasslands of central Kazakhstan in terms of its ability to characterize spatial patterns and interannual dynamics of net primary production (NPP) at a regional scale. In this model, the LUE of the grassland biome (ϵn) was simulated from ground-based NPP measurements, absorbed photosynthetically active radiation (APAR) and meteorological observations using a new empirical approach. Using coarse-resolution satellite data from the Sea-viewing Wide Field-of-view Sensor (SeaWiFS), monthly NPP was calculated from 1998 to 2008 over a large grassland region in Kazakhstan. The modelling results were verified against scaled up plot-level observations of grassland biomass and another available NPP data set derived from a field study in a similar grassland biome. The results indicated the reliability of productivity estimates produced by the model for regional monitoring of grassland NPP. The method for simulation of ϵn suggested in this study can be used in grassland regions where no carbon flux measurements are accessible.

Modeling Net Ecosystem Exchange for Grassland in Central Kazakhstan by Combining Remote Sensing and Field Data

Carbon sequestration was estimated in a semi-arid grassland region in Central Kazakhstan using an approach that integrates remote sensing, field measurements and meteorological data. Carbon fluxes for each pixel of 1 × 1 km were calculated as a product of photosynthetically active radiation (PAR) and its fraction absorbed by vegetation (fPAR), the light use efficiency (LUE) and ecosystem respiration (Re). The PAR is obtained from a mathematical model incorporating Earth-Sun distance, solar inclination, solar elevation angle, geographical position and cloudiness information of localities. The fPAR was measured in field using hemispherical photography and was extrapolated to each pixel by combination with the Normalized Difference Vegetation Index (NDVI) obtained by the Vegetation instrument on board the Satellite Pour l’Observation de la Terra (SPOT) satellite. Gross Primary Production (GPP) of the aboveground and belowground vegetation of 14 sites along a 230 km west-east transect within the study region were determined at the peak of growing season in different land cover types and linearly related to the amount of PAR absorbed by vegetation (APAR). The product of this relationship is LUE = 0.61 and 0.97 g C/MJ APAR for short grassland and steppe, respectively. The Re is estimated using complex models driven by climatic data. Growing season carbon sequestration was calculated for the modelling year of 2004. Overall, the short grassland was a net carbon sink, whereas the steppe was carbon neutral. The evaluation of the modelled carbon sequestration against independent reference data sets proved high accuracy of the estimations.

Assessing Satellite-Observed Nighttime Lights for Monitoring Socioeconomic Parameters in the Republic of Kazakhstan

This paper describes an initial assessment of human-induced nighttime lights acquired by the Defence Meteorological Satellite Program (DMSP) Operational Linescan System (OLS) with respect to its applicability in monitoring settlement patterns, population, electricity consumption, gross domestic product (GDP), and carbon dioxide emissions at different spatial levels in the Republic of Kazakhstan. The results revealed the suitability of DMSP-OLS data to detect both urban expansion and contraction over last two decades caused by the new economic situation following the independence of Kazakhstan in 1991. Relationships between DMSP-OLS urban lit area and the socioeconomic parameters were quantified. The DMSP-OLS data proved to be an effective tool in the monitoring of both the spatial and temporal variability of the examined socioeconomic parameters.

Reducing Uncertainty in Modeling the NDVI-Precipitation Relationship: A Comparative Study Using Global and Local Regression Techniques

The spatial relationship between vegetation and rainfall in Central Kazakhstan was modeled using the Normalized Difference Vegetation Index (NDVI) and rainfall data from weather stations. The modeling is based on the application of two statistical approaches: conventional ordinary least squares (OLS) regression, and geographically weighted regression (GWR). The results support the assumption that the average impression provided by the OLS model may not accurately represent conditions locally. The GWR approach, dealing with spatial non-stationarity, significantly increases the models accuracy and prediction power. The GWR provides a better solution to the problem of spatially autocorrelated errors in spatial modeling compared to the OLS modeling.

Assessment of vegetation vulnerability to ENSO warm events over Africa

Abstract The study investigates the vulnerability of vegetation over Africa to El-Nino Southern Oscillation (ENSO) events using the moving window statistical correlation analysis technique. The correlation analysis was done between Normalized Difference Vegetation Index (NDVI) data from Advanced Very High Resolution Radiometer (AVHRR) and an ENSO index, namely Multivariate ENSO Index (MEI). The study develops a new monitoring approach (ENSO vulnerability assessment system) to quantify the relationships between monthly maximum NDVI anomalies and their month-to-month correlations with the ENSO indices for the vegetative land areas of Africa. The data used in this study was for the period 1982–2006. The new monitoring approach was used in the assessment of the long-time vegetation sensitivity to the ENSO warm events that occurred during the study period. A map of Africa indicating the vegetation vulnerability to ENSO is produced. Different areas of vegetation vulnerability are identified within the main vegetation cover classes. For the African vegetative land, 16% of the total area was characterized by moderate vulnerability of vegetation to El-Nino, whereas 1.18% showed high vulnerability. Results suggest that the vulnerability of vegetative land surfaces across Africa to climate extremes, such as ENSO depends considerably on the vegetation type. In particular, results show that areas of equatorial rainforest are more resistant to drought stress than the wooded and non-wooded vegetation categories.

A remote sensing based monitoring system for discrimination between climate and human‐induced vegetation change in Central Asia

Purpose – This paper aims to demonstrate the importance of taking into account precipitation and the vegetation response to it when trying to analyse changes of vegetation cover in drylands with high inter‐annual rainfall variability.Design/methodology/approach – Linear regression models were used to determine trends in NDVI and precipitation and their interrelations for each pixel. Trends in NDVI that were entirely supported by precipitation trends were considered to impose climate‐induced vegetation change. Trends in NDVI that were not explained by trends in precipitation were considered to mark human‐induced vegetation change. Modelling results were validated by test of statistical significance and by comparison with the data from higher resolution satellites and fieldtrips to key test sites.Findings – More than 26 percent of all vegetated area in Central Asia experienced significant changes during 1981‐2000. Rainfall has been proved to enforce most of these changes (21 percent of the entire vegetated ...

Large-scale mapping of aboveground biomass of tropical rainforest in Sulawesi, Indonesia, using Landsat ETM+ and MODIS data

This study combined 30-m spatial resolution Landsat Enhanced Thematic Mapper Plus (ETM+) data and 500-m spatial resolution Moderate Resolution Imaging Spectroradiometer (MODIS) data, together with field data, to up-scale aboveground biomass (AGB) of tropical rainforests in Sulawesi, Indonesia. Contrasting to usual up-scaling approaches, this study uses in the intermediate step an estimation method based on application of geostatistics in the Landsat ETM+ spectral feature space. To connect ground-based measurements with spectral reflectance in Landsat ETM+ bands, this method employs ordinary Kriging in the three-dimesional (3-D) space where spectral features of satellite data are used instead of geographic coordinates. The most appropriate combination of the 3-D space spectral bands (bands 4 and 5) was then assigned with corresponding MODIS bands. In the final step, AGB could be mapped over the large area covered by the MODIS data. The results of the study indicate the effectiveness of the developed up-scaling method with respect to dealing with the problem of resolution mismatch between the ground sampling plots and the MODIS data.