Content uploaded by Jose Sobrino
Author content
All content in this area was uploaded by Jose Sobrino on Apr 07, 2017
Content may be subject to copyright.
A preview of the PDF is not available
Fourth International Symposium on Recent Advances in Quantitative Remote Sensing
International Journal of Remote Sensing
Abstract
This Symposium addressed the scientific advances in quantitative remote sensing in connection with real applications. Its main goal was to assess the state of the art of both theory and applications in the analysis of remote sensing data, as well as to provide a forum for researcher in this subject area to exchange views and report their latest results. In this book 103 of the 237contributions presented in both plenary and poster sessions are arranged according to the scientific topics selected. The papers are ranked in the same order as the final programme. José A. Sobrino Symposium Chairperson Global Change Unit, Universitat de València
... These procedures guaranteed that the correlation between a trait and LAI was kept almost constant for all levels of spatial dependency and the different realisations considered. The R 2 with LAI was defined based on experiments found in the literature [49,50]. Leaf angle distribution (attribute) Erectophile (90 • ) psoil 3 Dry/Wet soil factor (unitless) 0 -Geometry tto 4 View zenith angle-VZA (degree)~U(0,5) tts 4 Solar zenith angle-SZA (degree)~U(30, 38) psi 4 Relative azimuth angle (degree)~U (0,360) -U(129,252) -1 simulated from plant traits by levels of spatial dependency and rescaled to present a Normal distribution~N (mean, standard deviation). 2 a function of another parameter: Car = Cab/5 and Cw = 4/Cm −1 . ...
Spectral, temporal and spatial dimensions are difficult to model together when predicting in situ plant traits from remote sensing data. Therefore, machine learning algorithms solely based on spectral dimensions are often used as predictors, even when there is a strong effect of spatial or temporal autocorrelation in the data. A significant reduction in prediction accuracy is expected when algorithms are trained using a sequence in space or time that is unlikely to be observed again. The ensuing inability to generalise creates a necessity for ground-truth data for every new area or period, provoking the propagation of “single-use” models. This study assesses the impact of spatial autocorrelation on the generalisation of plant trait models predicted with hyperspectral data. Leaf Area Index (LAI) data generated at increasing levels of spatial dependency are used to simulate hyperspectral data using Radiative Transfer Models. Machine learning regressions to predict LAI at different levels of spatial dependency are then tuned (determining the optimum model complexity) using cross-validation as well as the NOIS method. The results show that cross-validated prediction accuracy tends to be overestimated when spatial structures present in the training data are fitted (or learned) by the model.
Ecological time series data are widely used in ecological research thanks to the development of remote-sensing technologies and fixed ecological research stations. However, the serial correlation issue with time series, which violates the fundamental assumption of independence for traditional statistical models or analysis, is rarely considered by ecologists in vegetation–climate relationship research. In addition, the issue of time lags between climate change and vegetation response is also often ignored. Inadequate consideration of these issues produces misleading results in some cases. In this article, we propose an approach based on the Autoregressive Integrated Moving Average (ARIMA) model and the nonparametric test to address serial correlation issue and distribution requirements for the valid statistical analysis of time series data. With Hulunber meadow steppe as a case, we applied this approach to analyse the role of climate factors in vegetation dynamics based on leaf area index (LAI) data and climatic data. The results showed that the LAI dynamics of Hulunber meadow steppe were mainly related to temperature with the time lag of zero, whereas the impact of precipitation on LAI dynamics was not statistically obvious. The comparison of regression models that deal with serial correlation and residual normality to different extents showed that ignoring the serial correlation issue with time series data likely produces misleading results, highlighting the importance of serial correlation removal. The combination of nonparametric correlation tests with ARIMA-based cross-correlation analysis also proved quite useful in reducing the chance of spurious correlation and time lags resulting from outlier values in ARIMA-based cross-correlation.
Monitoring of water resources and a better understanding of the eco-hydrological processes governing their dynamics are necessary to anticipate and develop measures to adapt to climate and water-use changes. Focusing on this aim, a research project carried out within the framework of French–Moroccan cooperation demonstrated how remote sensing can help improve the monitoring and modelling of water resources in semi-arid Mediterranean regions. The study area is the Tensift Basin located near Marrakech (Morocco) – a typical Southern Mediterranean catchment with water production in the mountains and downstream consumption mainly driven by agriculture. Following a description of the institutional context and the experimental network, the main recent research results are presented: (1) methodological development for the retrieval of key components of the water cycle in a snow-covered area from remote-sensing imagery (disaggregated soil moisture from soil moisture and ocean salinity) at the kilometre scale, based on the Moderate Resolution Imaging Spectroradiometer (MODIS); (2) the use of remote-sensing products together with land-surface modelling for the monitoring of evapotranspiration; and (3) phenomenological modelling based only on time series of remote-sensing data with application to forecasting of cereal yields. Finally, the issue of transfer of research results is also addressed through two remote sensing-based tools developed together with the project partners involved in water management and irrigation planning.
This article presents a methodology to quantitatively extract the solar-induced fluorescence (SIF) using the canopy reflectance index. The sensitivity analysis was conducted with a spectral vegetation Fluorescence Model (FluorMOD), and the results demonstrate that Sun zenith angle (θ), fluorescence quantum efficiency (Fi), leaf inclination distribution function (LIDF), leaf temperature (T), leaf area index, and leaf chlorophyll a + b content (chl-a+b) had large effects on the fluorescence radiance at 761 nm (LF,761). Based on the results of the sensitivity analysis, the input parameters θ, Fi, LIDF, T, and chl-a+b varied within a certain range during the generation of the simulated data. Based on the simulated data, R740/R630, R685/R850, and R750/R710 were thought to be the best candidates to extract the fluorescence radiation. The quantitative relationships between the fluorescence retrieved by R740/R630, R685/R850, and R750/R710 and LF,761 were analysed and expressed as functions of θ, Fi, T, and reflectance index. The correlation coefficients (r) between the fluorescence retrieved using R685/R850, R740/R630, and R750/R710 and LF,761 are 0.94, 0.95, and 0.95, respectively, and the root mean square errors (RMSEs) were 0.32, 0.29, and 0.30 W m−2 μm−1 sr−1, respectively. Through comparison with FLD and 3FLD, the method presented in this article yielded better results, and could be used to estimate the fluorescence. This methodology provides new insights into the quantitative retrieval of SIF from the reflectance spectrum.
Soil moisture is an important parameter that influences the exchange of water and energy fluxes between the land surface and the atmosphere. Through the simulation by a Soil–Vegetation–Atmosphere Transfer model, Carlson proposed the universal spatial information-based method to determine soil moisture that is insensitive to the initial atmospheric and surface conditions, net radiation, and atmospheric correction. In this study, a practical normalized soil moisture model is established to describe the relationship among the normalized soil moisture (M), the normalized land surface temperature (T*), and the fractional vegetation cover. The dry and wet points are determined using the surface energy balance principle, which has a robust physical basis. This method is applied to retrieve soil moisture for the Soil Moisture-Atmosphere Coupling Experiment campaign in the Walnut Creek watershed, which has a humid climate, and at the Linzestation, which has a semi-arid climate. The validation data are obtained on days of year (DOYs) 182 and 189 in 2002 in the humid region and on DOYs 148 and 180 in 2008 for the semi-arid region; these data collection days are coincident with the overpass of the Landsat Thematic Mapper/Enhanced Thematic Mapper Plus. When the estimates are compared with the in situ measurements of soil water content, the root mean square error is approximately 0.10 m3 m−3 with a bias of 0.05 m3 m−3 for the humid region and 0.08 m3 m−3 with a bias of 0.03 m3 m−3 for the semi-arid region. These results demonstrate that the practical normalized soil moisture model is applicable in both humid and semi-arid regions.
Linking observed or estimated ground incoming solar radiation with cloud coverage is difficult since the latter is usually poorly described in standard meteorological observation protocols. To investigate the benefits of detailed observation and characterization of cloud coverage and distribution, a fieldwork campaign has been set up in order to collect data about cloud cover conditions and daily evolution to directly analyse their impacts on solar radiation fluxes. To do so, daytime hemispherical images have been collected at a very high frequency, simultaneously to ground measurements of solar radiation fluxes in a scientific station close to Lake NamCo, China. After calibration, one of the main tasks was the classification of those hemispherical images and the extraction of meaningful indices to describe the cloud cover, such as cloud fraction or cloud cover distribution. The classification is based on automatic detection of threshold on the red channel histogram. The results show that several cloud indices could be successfully derived from the hemispherical images, even if very thin clouds can be difficult to detect. The indices are then correlated to the measured solar radiation values and the impact of cloud cover on surface radiation fluxes were analysed. This analysis highlights that, more than the cloud fraction, the cloud distribution in the hemisphere is of importance when modelling radiation fluxes in the solar domain.
Grassland vegetation growth directly reflects plant growth conditions, and growth processes are an important component of ecological status assessments of grasslands and can provide timely guidance for agricultural production. In this study, the Xilingol Grassland in Inner Mongolia was used as the study area, and a monitoring indicator system for the remote sensing of grassland vegetation growth was established based on 16 days of Moderate Resolution Imaging Spectroradiometer (MODIS) data and ground sampling data, which were used to assess the suitability of the indicator system. A monitoring indicator system for the remote sensing of grassland vegetation that included the modified growth index (MGI)-normalized difference vegetation index (NDVI), MGI-enhanced vegetation index (EVI), growth index (GI)-NDVI, and GI-EVI was established by using the year 2000 as the base year, two vegetation indices and difference and normalized difference methods. A model for estimating the ground growth (g) was then constructed by using expert opinion scoring and the Analytic Hierarchy Process (AHP) to determine the weights of vegetation coverage (c), height (h), and yield (y) in the ground plots, with the model calculated as follows: g = 0.2543c + 0.1848 h + 0.5609y. Additionally, the ground growth value was calculated according to the ground growth model, and the values obtained from the remote-sensing indicators in the corresponding region were subjected to a correlation analysis based on this partition. The remote-sensing growth indices suitable for temperate steppe, meadow steppe, and desert steppe regions were GI-EVI, MGI-NDVI, and GI-NDVI, respectively. Finally, vegetation growth in the Xilingol Grassland was evaluated using the optimal remote-sensing GI for each area, and the results indicated that the areas with the greatest growth improvement occurred in the temperate steppe region followed by the meadow steppe region, whereas vegetation growth improvement was insignificant in the desert steppe region. The results of this study have important significance for the economic development and ecological environmental improvement of pastoral and semi-pastoral regions and can be used to determine the optimal use and scientific management of grasslands.
Timely and accurate estimates of crop areas are critical to enhancing agriculture management and ensuring national food security. This study aims to combine remote-sensing data and an optimized spatial sampling scheme to improve the accuracy of crop area estimates and decrease the cost of crop surveys at a regional scale. This study focuses on winter wheat in Mengcheng County in Anhui Province, China. Advanced Land Observing Satellite (ALOS) Advanced Visible light and Near Infrared Radiometer (AVNIR)-2, and Landsat5 Thematic Mapper (TM) images from 2009 and 2010, respectively, are used to extract the winter wheat area and distribution. Additionally, a spatial sampling scheme was optimized by combining remotely sensed data, geographical information system (GIS), Geostatistics, and traditional sampling methods. The experimental results demonstrate that the variability in the proportion of winter wheat acreage in one sampling unit (PWS) increases with increasing sampling unit size. The PWS coefficient of variation (CV) varies from 32.75 to 43.46% among the eight sampling unit sizes. The spatial correlation thresholds of PWS increase with increasing sampling unit size. For small sampling unit sizes (500 m × 500 m–2000 m × 2000 m), the relative error and CV of the population extrapolation for the optimized sample layout are obviously lower than those of the simple random sampling method. For larger sampling unit sizes (2500 m × 2500 m–4000 m × 4000 m), the sample size is obviously lower for the optimized sample layout compared with that of the simple random sampling method, but there are no differences in the relative errors or CVs. By combining remote-sensing data and the optimized spatial sampling scheme, this research can improve the accuracy of crop area estimation at a regional scale.
Vegetation dynamics, particularly vegetation growth, are often used as indicators of potential grassland degradation. Grassland vegetation growth can be monitored using remotely sensed data, which has rapid and broad coverage. Grassland ecosystems are an important component of the regional landscape. In this study, we developed an applicable method for monitoring grassland growth. The dynamic variation in the grassland was analysed using Moderate Resolution Imaging Spectroradiometer (MODIS) data. The normalized difference vegetation index (NDVI) was calculated from 2001 to 2010 during the grassland growing season. To evaluate the grassland growth, the use of the growth index (GI) was proposed. According to the GI values, five growth grades were identified: worse, slightly worse, balanced, slightly better, and better. We explored the spatial-temporal variation of grassland growth and the relationship between grassland growth and meteorological factors (i.e. precipitation and temperature factors). Our results indicated that, compared with the multi-year average, the spatial-temporal variation of grassland growth was significantly different between 2001 and 2010. The vegetation growth was worse in 2009 compared with the multi-year average. A GI of ‘worse’ accounted for 66.73% of the area. The vegetation growth in 2003 was the best of the years between 2001 and 2010, and a better GI accounted for 58.08% of the area in 2003. The GI from 2004 to 2008 exhibited significant fluctuations. The correlation coefficient between the GI and precipitation or temperature indicated that meteorological factors likely affected the inter-annual variations in the grassland growth. The peak of the grassland growth season was positively correlated with the spatial patterns of precipitation and negatively correlated with those of temperature. Precipitation during the growing season was the main influence in the arid and semi-arid regions. Monitoring grassland growth using remote sensing can accurately reveal the grassland growth status at the macro-scale in a timely manner. This research proposes an effective method for monitoring grassland growth and provides a reference for the sustainable development of grassland ecosystems.
Calibration and validation (cal/val) are key activities to test the data quality acquired from satellite-based instruments, as well as to report the accuracy of derived products such as the land surface temperature (LST). Calibration of thermal infrared (TIR) data and validation of LST products at low spatial resolution requires the identification of large and homogeneous areas, which is a difficult task. In this work, spatial and temporal homogeneity of LST was analysed over three Spanish regions: the agricultural area of Barrax, Doñana National Park, and Cabo de Gata Natural Park. For this purpose, very high spatial resolution (approximately 3 m) imagery acquired with the Airborne Hyperspectral Scanner (AHS) in the framework of different field campaigns and high–medium spatial resolution (approximately 100 m) imagery acquired with the Landsat-8 (L8) TIR sensor (TIRS) have been used to retrieve homogeneity of high–medium and low spatial resolution sensors, respectively. Different LST retrieval algorithms were applied to AHS and TIRS to compare the LST for a given pixel against the LST of neighbour pixels through the computation of the root mean square error (RMSE). The results obtained from the analysis of LST derived from AHS data over Barrax and Doñana test sites show that part of these regions have an RMSE lower than 1 K, which is consistent with the accuracy of the LST validation (between 0.5 and 1.5 K). The analysis of LST derived from the TIRS shows that some parts of Doñana and Cabo de Gata sites have a mean RMSE of 1 K over the period of a year, with maximal homogeneity in autumn and winter (lower than 1 K) and minimal in spring and summer (around 2 K). These results are lower than the accuracy of the LST validation (approximately 2 K). The results show the usefulness of these three test sites to perform cal/val activities for both low and high spatial resolution sensors. The methodology presented in this study also allows the identification of suitable areas for future cal/val activities.
Tibet, the largest region of the Qinghai–Tibet Plateau, is undergoing extensive grassland deterioration and desertification due to both human and natural factors. Alpine meadow and grassland restoration is difficult after degradation; consequently, the desertification of the Tibetan grassland has attracted substantial social attention. This article considered Amdo, Baingoin, Coqên, and Zhongba counties in Tibet as the study areas, employed remote-sensing data, and developed Tibetan grassland desertification classification indices based on field surveys. Moreover, this study used spectral mixture analysis (SMA) methods to interpret remote-sensing image data from the study areas during three periods (1990, 2000, and 2009) and considered the bare sand (gravel) area proportion as the main basis for the evaluation of grassland desertification. The results of this study demonstrate that the slightly, moderately, and severely desertified grasslands of the monitoring zone covered a total area of 114,113.16 km2 in 1990, accounting for 82.12% of the study area. The area exhibited no change in 2000 and decreased by 4472.31 km2 in 2009. The severely desertified grassland area declined from 1990 to 2009. The degree of grassland desertification in these four Tibetan counties diminished from 1990 to 2009, and the grassland desertification area exhibited a gradual reduction during the same period. Regarding other soil coverage types, the ice and snow area markedly changed and declined to approximately one-third of its original extent during these 20 years, and most of the ice and snow area was converted to bare land and various types of desertified grassland.