Satellite Spectral Data Research Articles

Rapid advancements in large language models for quantitative remote sensing: The case of water depth inversion

This study presents a comparative analysis of two advanced AI models, ChatGPT and ERNIE, in the context of water depth inversion. Utilizing satellite spectral data and in-situ bathymetric measurements collected from Rushikonda Beach, India, we processed and analyzed the data to generate high-resolution bathymetric maps. Both models demonstrated significant accuracy, with ChatGPT slightly outperforming ERNIE in terms of mean absolute error. The study highlights the advantages of AI models, such as efficient data processing and the ability to integrate multi-modal inputs, while also discussing challenges related to data quality, interpretability, and computational demands. The findings suggest that while both models are highly effective for water depth inversion, ongoing improvements in data handling and model transparency are essential for their broader application in environmental monitoring. This research contributes to the understanding of AI capabilities in geospatial analysis and sets the stage for future enhancements in the field.

Inversion of Soil Moisture Content in Cotton Fields Using GBR-RF Algorithm Combined with Sentinel-2 Satellite Spectral Data

Soil moisture content plays a vital role in agricultural production, significantly influencing crop growth, development, and yield. Thoroughly understanding the specific soil moisture content in cotton fields is crucial for enhancing agricultural efficiency and driving sustainable agricultural development. This study utilized the gradient-boosting regression–random forest (GBR-RF) algorithm and the GBR and RF algorithms separately, in conjunction with Sentinel-2 satellite images, to estimate cotton soil moisture content, focusing on the B1–B8 bands and in particular the sensitive B6, B7, and B8 bands. The soil data in the jujube orchard of the study area were collected using soil augers at a depth of 30 cm, with soil data collected from a depth of 20 to 30 cm. The findings revealed that the integrated learning algorithm GBR-RF demonstrated high accuracy, with R2, MAE, and MSE results of 0.8838, 1.0121, and 1.6168, respectively. In comparison, the results using just the GBR algorithm yielded R2, MAE, and MSE values of 0.8158, 1.1327, and 1.9645, respectively, while those obtained from the RF algorithm were 0.8415, 1.0680, and 1.8331, respectively. These results indicate that the algorithms exhibited strong generalization, robustness, and accuracy, with GBR-RF outperforming GBR and RF by 8.34% and 5.03%, respectively, in combination with using the B1–B8 bands for inversion. Furthermore, utilizing the full-band data resulted in R2 values that were up to 24.27% higher than those of the individual bands, affirming the efficacy of band combinations for improved accuracy. This study’s demonstration of the positive impact of integrated learning algorithms on estimating cotton soil moisture content underscores the advantages of multi-band data combinations over single-band data, highlighting their ability to enhance accuracy without significantly impacting errors. Importantly, this study’s findings, while not limited to a single experimental field, have broad applicability in cotton precision agriculture, offering valuable insights for research on yield enhancement and agricultural efficiency.

Leveraging heterogeneous LiDAR data to model successional stages at tree species level in temperate forests

Abstract In the context of the ongoing biodiversity crisis, understanding forest ecosystems, their tree species composition, and especially the successional stages of their development is crucial. They collectively shape the biodiversity within forests and thereby influence the ecosystem services that forests provide, yet this information is not readily available on a large scale. Remote sensing techniques offer promising solutions for obtaining area-wide information on tree species composition and their successional stages. While optical data are often freely available in appropriate quality over large scales, obtaining light detection and ranging (LiDAR) data, which provide valuable information about forest structure, is more challenging. LiDAR data are mostly acquired by public authorities across several years and therefore heterogeneous in quality. This study aims to assess if heterogeneous LiDAR data can support area-wide modeling of forest successional stages at the tree species group level. Different combinations of spectral satellite data (Sentinel-2) and heterogeneous airborne LiDAR data, collected by the federal government of Rhineland-Palatinate, Germany, were utilized to model up to three different successional stages of seven tree species groups. When incorporating heterogeneous LiDAR data into random forest models with spatial variable selection and spatial cross-validation, significant accuracy improvements of up to 0.23 were observed. This study shows the potential of not dismissing initially seemingly unusable heterogeneous LiDAR data for ecological studies. We advocate for a thorough examination to determine its usefulness for model enhancement. A practical application of this approach is demonstrated, in the context of mapping successional stages of tree species groups at a regional level.

Comparing ground below-canopy and satellite spectral data for an improved and integrated forest phenology monitoring system

Phenology monitoring allows a better understanding of forest functioning and climate impacts. Satellite indicators are used to upscale ground phenological observations, but often differential responses are observed, and data availability can be limited. In view of climate impacts, new tools capable to detect rapid phenological changes and to work at single species level are needed. This research compares indices derived by the TreeTalker© (TT + ) below canopy upward-looking spectral data and Sentinel 2 satellite data, used to assess the phenological behavior and changepoints in several European beech forests. Overall, a mismatch between the information derived by the two sensor types is evidenced, with main differences in: start/end and length of season and phenology changepoints; larger variability captured by TT + with respect to Sentinel 2 especially in the leaf on period; mixed signal response from multiple vegetation layers in Sentinel 2 data. The complementarity of satellite and TT + indices allow exploring the phenological responses from different vegetation layers. TT + higher temporal resolution demonstrates precision in capturing the phenological changepoints in beech forests, especially if satellite image availability is limited by cloud cover and leads to miss critical phenological dates. The best settings for TT + data collection and the advantages to have two spectral data sources for improved forest phenology monitoring are also commented. The TT+, collecting additional tree parameters, can be a valuable tool for an integrated monitoring system based on spectral signals from above and below the canopy, at high temporal frequency and high spatial resolution.

Intra- and interspecific variation in spectral properties of dominant Sphagnum moss species in boreal peatlands.

Boreal peatlands store ~25 % of global soil organic carbon and host many endangered species; however, they face degradation due to climate change and anthropogenic drainage. In boreal peatlands, vegetation indicates ecohydrological conditions of the ecosystem. Applying remote sensing would enable spatially and temporally continuous monitoring of peatland vegetation. New multi- and hyperspectral satellite data offer promising approaches for understanding the spectral properties of peatland vegetation at high temporal and spectral resolutions. However, using spectral satellite data to their fullest potential requires detailed spectral analyses of dominant species in peatlands. A dominant feature of peatland vegetation is the genus Sphagnum mosses. We investigated how the reflectance spectra of common boreal Sphagnum mosses, collected from waterlogged natural conditions after snowmelt, change when the mosses are desiccated. We conducted a laboratory experiment where the reflectance spectra (350-2500 nm) and the mass of 90 moss samples (representing nine species) were measured repetitively. Furthermore, we examined (i) their inter- and intraspecific spectral differences and (ii) whether the species or their respective habitats could be identified based on their spectral signatures in varying states of drying. Our findings show that the most informative spectral regions to retrieve information about the Sphagnum species and their state of desiccation are in the shortwave infrared region. Furthermore, the visible and near-infrared spectral regions contain less information on species and moisture content. Our results also indicate that hyperspectral data can, to a limited extent, be used to separate mosses belonging to meso- and ombrotrophic habitats. Overall, this study demonstrates the importance of including data especially from the shortwave infrared region (1100-2500 nm) in remote sensing applications of boreal peatlands. The spectral library of Sphagnum mosses collected in this study is available as open data and can be used to develop new methods for remote monitoring of boreal peatlands.

Spectral determinants of biological soil crusts in the Gurbantungut Desert

ABSTRACT As thick crustal layers form on dryland surfaces, they affect the spectral information that is originally dominated by sand or rock. The spectral characteristics of organic matter replace the mineral elements as prominent features. In this case, the growth patterns and spectral characteristics of biological soil crusts (BSCs) can be observed. Satellite spectral data have been used for BSC spatial information extraction. However, the dynamic changes in BSCs can affect the spectra. Two aspects are involved here: moisture change and BSC growth. When these changes are superimposed with BSC succession, they lead to an increase in spectral complexity. This study explored three BSC types, including algal crust, lichen and moss, and discussed their spectra. By selecting BSC samples at different succession states and by combining coverage and simulated precipitation, the response of the spectra to BSC coverage and the spectral characteristics of BSCs under dry and wet conditions were measured and analysed. In addition, the spectral index variations caused by coverage and moisture of three types of BSCs were discussed, where the spectral indices include the normalized difference vegetation index (NDVI), brightness index (BI), crust index (CI), and biological soil crust index (BSCI). The results showed that the succession, moisture and growth of BSCs were the main factors affecting their spectra. BSC types can be distinguished in a particular climatic context to determine the degree of BSC succession. Precipitation in the monitoring areas needs to be considered to avoid the effects of dry and wet BSC variations on remote sensing monitoring. The coverage of different types of BSCs in mixed pixels can be determined by multiple indices. The results of this study will provide a basis for monitoring BSCs using satellite spectral information to guide regional ecological management.

Mudflat surface sediment type mapping by remote sensing considering the effect of the chlorophyll-a content

Mudflats are critical interfaces between the marine and terrestrial environment of muddy coasts. Mapping mudflat surface sediment types and analyzing their dynamic changes are helpful to understand the variations of sedimentary environments and their responses to tidal current movements. To obtain the surface sediment types of the mudflats, this study first took the chlorophyll-a content as an environmental variable and combined it with satellite spectral reflectance data processed by fractional-order derivative (FOD) to establish a machine learning model for retrieving sediment component content (SCC) of the sand, silt, and clay. Then, the three SCCs were linear equilibrium corrected and inputted into Folk's ternary classification model, and the spatial distribution map of the sediment types was accurately obtained. The results showed that the use of the FODs for satellite image enhancement could provide richer spectral information and improve the accuracy of the chlorophyll-a content and SCCs inversed by the grid search-support vector machine (GS-SVM) model. When compared to direct spectral modeling, considering the effect of the chlorophyll-a content as the key input factor, the coefficients of determination (R2) for the sand, silt, and clay contents inversion were increased by 15.1, 9.2, and 38.2%, and the root mean square error (RMSE) were reduced by 9.7, 5.5, and 2.5%, respectively. Surface sediment type maps obtained from the three SCCs showed that the main sediments in the mudflats on the central coast of Jiangsu Province, China were silty sand and sandy silt. Between 2019 and 2021, the sediment types in 84.46% of the total area were unchanged. For the changing area, the sediment transitioned toward finer-grained types near the seawall of the coastal region. Whereas in the offshore area, especially along the edge of the huge tidal channels, the sediments tended to be coarser because of strong hydrodynamic screening. The findings will be helpful for improving the ability of mudflat sedimentation environmental monitoring and spatial resource management via remote sensing.

Wheat crops weediness assessment based on geobotanical, ground and satellite spectral data

Wheat crops weediness assessment based on geobotanical, ground and satellite spectral data

A Framework for Soil Salinity Monitoring in Coastal Wetland Reclamation Areas Based on Combined Unmanned Aerial Vehicle (UAV) Data and Satellite Data

Soil salinization is one of the most important causes of land degradation and desertification, often threatening land management and sustainable agricultural development. Due to the low resolution of satellites, fine mapping of soil salinity cannot be completed, while high-resolution images from UAVs can only achieve accurate mapping of soil salinity in a small area. Therefore, how to realize fine mapping of salinity on a large scale based on UAV and satellite data is an urgent problem to be solved. Therefore, in this paper, the most relevant spectral variables for soil salinity were firstly determined using Pearson correlation analysis, and then the optimal inversion model was established based on the screened variables. Secondly, the feasibility of correcting satellite data based on UAV data was determined using Pearson correlation analysis and spectral variation trends, and the correction of satellite data was completed using least squares-based polynomial curve fitting for both UAV data and satellite data. Finally, the reflectance received from the vegetated area did not directly reflect the surface reflectance condition, so we used the support vector machine classification method to divide the study area into two categories: bare land and vegetated area, and built a model based on the classification results to realize the advantages of complementing the accurate spectral information of UAV and large-scale satellite spectral data in the study areas. By comparing the modeling inversion results using only satellite data with the inversion results based on optimized satellite data, our method framework could effectively improve the accuracy of soil salinity inversion in large satellite areas by 6–19%. Our method can meet the needs of large-scale accurate mapping, and can provide the necessary means and reference for soil condition monitoring.

Multiscale assessment of ground, aerial and satellite spectral data for monitoring wheat grain nitrogen content

Wheat grain quality characteristics have experienced increasing attention as a central factor affecting wheat end-use products quality and human health. Nonetheless, in the last decades a reduction in grain quality has been observed. Therefore, it is central to develop efficient quality-related phenotyping tools. In this sense, one of the most relevant wheat features related to grain quality traits is grain nitrogen content, which is directly linked to grain protein content and monitorable with remote sensing approaches. Moreover, the relation between nitrogen fertilization and grain nitrogen content (protein) plays a central role in the sustainability of agriculture. Both aiming to develop efficient phenotyping tools using remote sensing instruments and to advance towards a field-level efficient and sustainable monitoring of grain nitrogen status, this paper studies the efficacy of various sensors, multispectral and visible red–greenblue (RGB), at different scales, ground and unmanned aerial vehicle (UAV), and phenological stages (anthesis and grain filling) to estimate grain nitrogen content. Linear models were calculated using vegetation indices at each sensing level, sensor type and phenological stage. Furthermore, this study explores the up-scalability of the best performing model to satellite level Sentinel-2 equivalent data. We found that models built at the phenological stage of anthesis with UAV-level multispectral cameras using red-edge bands outperformed grain nitrogen content estimation (R2 = 0.42, RMSE = 0.18%) in comparison with those models built with RGB imagery at ground and aerial level, as well as with those built with widely used ground-level multispectral sensors. We also demonstrated the possibility to use UAV-built multispectral linear models at the satellite scale to determine grain nitrogen content effectively (R2 = 0.40, RMSE = 0.29%) at actual wheat fields.

Complex analysis of the ionosphere variations during the geomagnetic storm at 20 January 2010 performed by Detection of Ionosphere Anomalies (DIA) software and DEMETER satellite data

Abstract. The Detection of Ionosphere Anomalies (DIA) software was developed and used for ionosphere research in the Institute of Space Technique and Technology of the National Center of Space Research and Technology (Almaty, Republic of Kazakhstan). The software was designed for ionosphere anomalies detection, identification, and analysis from satellite spectral and wave data from scientific payloads installed on the DEMETER (Detection of Electro-Magnetic Emissions Transmitted from Earthquake Regions) satellite. The main task of DIA software is to provide the researcher with a convenient package for detection and identification of the sources of electromagnetic radiation, variations of the ionic and electronic components of the ionosphere, and other ionosphere parameters from satellite data. Using this software, a complex research of the state of the medium-latitude ionosphere during a geomagnetic storm on 20 January 2010 was done. Electric and magnetic components of the field in ultralow frequency (ULF), extremely low frequency (ELF), and very low frequency (VLF) bands are processed and analyzed, as well as temperature, velocity, and density of ionic and electronic plasma components and fluxes of energetic electrons at satellite altitude during a storm.

Soil-sample geochemistry normalised by class membership from machine-learnt clusters of satellite and geophysics data

The primary aim of mineral exploration is to focus the search space from regional scale (1000 s km2) to prospect scale (1–10 km2), identify areas that warrant further investigation, and ultimately drill a target. Mineral deposits typically display geochemical anomalism related to the conditions of their formation, or the weathering and mass transport of sediment away from a mineral deposit. However, this anomalism can be obscured in soil-sample surveys due to background variations in host bedrock chemistry and the transformation and redistribution of original minerals during surface weathering, regolith development, and supergene processes. This contribution uses machine-learnt map products to aid in the interpretation of a regional-scale soil-sample dataset. ASTER (satellite spectral) data were combined with airborne radiometric data using Self-Organising Maps (SOM) to generate a map of regolith type. A map of bedrock geology was previously derived by the authors from airborne geophysical data using Random Forests. These two products are used to classify soil samples with an objective bedrock-geology and regolith-type classification. A z-score normalisation was implemented on the soil-sample geochemical data, using two different matrices (regolith type and bedrock geology) to remove the bias of bedrock geology and regolith type from the soil sample chemistry. The signal that remains likely reflects metasomatic processes, and associated elemental anomalism, potentially linked to mineralisation. These data can then be plotted to provide maps of (pathfinder) element anomalies, thus identifying potential exploration targets. We explore this approach to normalise geochemical data in a dataset of 9,924 soil samples collected over an area of 1,028 km2, from Kerkasha, Eritrea (Nubian Shield). The Kerkasha project is an early-stage exploration project, with a limited understanding of the local geology and mineral deposit types; however, it contains many Au and Cu prospects that have received minimal exploration work.

Evaluation of the status of land use/land cover change using remote sensing and GIS in Jewha Watershed, Northeastern Ethiopia

Evaluation of land use/land cover (LULC) status of watersheds is vital to environmental management. This study was carried out in Jewha watershed, which is found in the upper Awash River basin of central Ethiopia. The total catchment area is 502 km2. All climatic zones of Ethiopia, including lowland arid (‘Kola’), midland semi-arid (‘Woinadega’), humid highland (Dega) and afro alpine (‘Wurch’) can be found in the watershed. The study focused on LULC classification and change detection using GIS and remote sensing techniques by analyzing satellite images. The data preprocessing and post-process was done using multi-temporal spectral satellite data. The images were used to evaluate the temporal trends of the LULC class by considering the years 1984, 1995, 2005 and 2015. Accuracy assessment and change detection of the classification were undertaken by accounting these four years images. The land use types in the study area were categorized into six classes: natural forest, plantation forest, cultivated land, shrub land, grass land and bare land. The result shows the cover classes which has high environmental role such as forest and shrub has decreased dramatically through time with cultivated land increasing during the same period in the watershed. The forest cover in 1984 was about 6.5% of the total catchment area, and it had decreased to 4.2% in 2015. In contrast, cultivated land increased from 38.7% in 1984 to 51% in 2015. Shrub land decreased from 28 to 18% in the same period. Bare land increased due to high gully formation in the catchment. In 1984, it was 1.8% which turned to 0.6% in 1995 then increased in 2015 to 2.7%. Plantation forest was not detected in 1984. In 1995, it covers 1.5% which turned to be the same in 2015. The study clearly demonstrated that there are significant changes of land use and land cover in the catchment. The findings will allow making informed decision which will allow better land use management and environmental conservation interventions.

Estimating food resource availability in arid environments with Sentinel 2 satellite imagery.

BackgroundIn arid environments, plant primary productivity is generally low and highly variable both spatially and temporally. Resources are not evenly distributed in space and time (e.g., soil nutrients, water), and depend on global (El Niño/ Southern Oscillation) and local climate parameters. The launch of the Sentinel2-satellite, part of the European Copernicus program, has led to the provision of freely available data with a high spatial resolution (10 m per pixel). Here, we aimed to test whether Sentinel2-imagery can be used to quantify the spatial variability of a minor tussock grass (Enneapogon spp.) in an Australian arid area and whether we can identify different vegetation cover (e.g., grass from shrubs) along different temporal scenarios. Although short-lasting, the Enneapogon grassland has been identified as a key primary food source to animals in the arid environment. If we are able to identify and monitor the productivity of this species remotely, it will provide an important new tool for examining food resource dynamics and subsequent animal responses to them in arid habitat.MethodsWe combined field vegetation surveys and Sentinel2-imagery to test if satellite spectral data can predict the spatial variability of Enneapogon over time, through GLMMs. Additionally, a cluster analysis (‘gower’ distance, ‘complete’ method), based on Enneapogon seed-productivity, and total vegetation cover in October 2016, identified three clusters: bare ground, grass dominated and shrub dominated. We compared the vegetation indices between these different clusters from October 2016 to January 2017.ResultsWe found that MSAVI2 and NDVI correlated with the proportion of Enneapogon with seeds across the landscape and this relationship changed over time. Both vegetation indices (MSAVI2 and NDVI) were higher in patches with high seed-productivity of Enneapogon than in bare soil, but only in October, a climatically-favorable period during which this dominant grass reached peak seed-productivity.DiscussionMSAVI2 and NDVI provided reliable estimates of the heterogeneity of vegetation type across the landscape only when measured in the Austral spring. This means that grass cover is related to seed-productivity and it is possible to remotely and reliably predict food resource availability in arid habitat, but only in certain conditions. The lack of significant differences between clusters in the summer was likely driven by the short-lasting nature of the vegetation in the study and the sparseness of the grass-dominated vegetation, in contrast to the shrub vegetation cluster that was particularly well measured by the NDVI.ConclusionsOverall, our study highlights the potential for Sentinel2-imagery to estimate and monitor the change in grass seed availability remotely in arid environments. However, heterogeneity in grassland cover is not as reliably measured as other types of vegetation and may only be well detected during periods of peak productivity (e.g., October 2016).

Carrot Yield Mapping: A Precision Agriculture Approach Based on Machine Learning

Carrot yield maps are an essential tool in supporting decision makers in improving their agricultural practices, but they are unconventional and not easy to obtain. The objective was to develop a method to generate a carrot yield map applying a random forest (RF) regression algorithm on a database composed of satellite spectral data and carrot ground-truth yield sampling. Georeferenced carrot yield sampling was carried out and satellite imagery was obtained during crop development. The entire dataset was split into training and test sets. The Gini index was used to find the five most important predictor variables of the model. Statistical parameters used to evaluate model performance were the root mean squared error (RMSE), coefficient of determination (R2) and mean absolute error (MAE). The five most important predictor variables were the near-infrared spectral band at 92 and 79 days after sowing (DAS), green spectral band at 50 DAS and blue spectral band at 92 and 81 DAS. The RF algorithm applied to the entire dataset presented R2, RMSE and MAE values of 0.82, 2.64 Mg ha−1 and 1.74 Mg ha−1, respectively. The method based on RF regression applied to a database composed of spectral bands proved to be accurate and suitable to predict carrot yield.

Estimation of the spatial distribution of spring barley yield using ground-based and satellite spectrophotometric data

The article presents a method for estimating the spatial distribution of spring barley yield, based on the use of optical ground and satellite spectral data (PlanetScope data with a spatial resolution of 3 meters). This approach is highly relevant for the development of precision agriculture technologies. Yield mapping is carried out on the basis of data on the spatial distribution of the actual yield and the spatial distribution of the spectral optical characteristics. The method’s characteristic feature is the use of the integral values of vegetation indices (NDVI, MSAVI2, ClGreen) at various stages of crop development. The method was tested on the basis of stationary field experience, where traditional agriculture (deep plowing) is compared with resource-saving technologies (subsurface and surface plowing, and direct seeding with zero tillage).

Multi Spectral Satellite Data to Investigate Land Expansion and Related Micro Climate Change as Threats to the Environment

Technology development of satellite earth observation offers multi-spectral data, variations sensitivity of satellite sensors are very useful to detect existing conditions by using the remote sensing algorithm to extract the land cover and the phenomenon that has occurred. Ecosystems have bound tolerance to nature condition and organisms have adaptation limit to climate change. Method used in this study was the extraction of land cover by applying multi spectral analysis as Maximum Likelihood, object base image analysis, (LST) land surface temperature, and also the formula Threat of Land Expansion (TLE) to the environment. Result showed that multispectral and time series analysis gave the increasing dynamic of land expansion, especially for forest high density, which areas ranged from 29098.73 ha in 2000 to 19216.7 ha in 2018. The surface temperature increase was ± 2.8 C° in eighteen years. The temperature dynamic represented the micro climate changes, has followed stream of land expansion. TLE represented zone threats of land expansion to natural ecosystems. It was found that threat zone because land expansion. The area of high threat vale 0.4 around central human activity was a built up area.

Leaf chlorophyll content retrieval of wheat by simulated RapidEye, Sentinel-2 and EnMAP data

Leaf chlorophyll content (LCC) is an important physiological indicator of the actual health status of individual plants. An accurate estimation of LCC can therefore provide valuable information for precision field management. Red-edge information from hyperspectral data has been widely used to estimate crop LCC. However, after the advent of red-edge bands in satellite imagery, no systematic evaluation of the performance of satellite data has been conducted. Toward this end, we analyze herein the performance of winter wheat LCC retrieval of currant and forthcoming satellites (RapidEye, Sentinel-2 and EnMAP) and their new red-edge bands by using partial least squares regression (PLSR) and a vegetation-index-based approach. These satellite spectral data were obtained by resampling ground-measured hyperspectral data under various field conditions and according to specific spectral response functions and spectral resolution. The results showed: 1) This study confirmed that RapidEye, Sentinel-2 and EnMAP data are suitable for winter wheat LCC retrieval. For the PLSR approach, Sentinel-2 data provided more accurate estimates of LCC (R2=0.755, 0.844, 0.805 for 2002, 2010, and 2002+2010) than do RapidEye data (R2=0.689, 0.710, 0.707 for 2002, 2010, and 2002+2010) and EnMAP data (R2=0.735, 0.867, 0.771 for 2002, 2010, and 2002+2010). For index-based approaches, the MERIS terrestrial chlorophyll index, which is a vegetation index with two red-edge bands, was the most sensitive and robust index for LCC for both the Sentinel-2 and EnMAP data (R2≥0.628), and the indices (NDRE1, SRRE1 and CIRE1) with a single red-edge band were the most sensitive and robust indices for the RapidEye data (R2≥0.420); 2) According to the analysis of the effect of the wavelength and number of used red-edge spectral bands on LCC retrieval, the short-wavelength red-edge bands (from 699 to 734 nm) provided more accurate predictions when using the PLSR approach, whereas the long-wavelength red-edge bands (740 to 783 nm) gave more accurate predictions when using the vegetation indice (VI) approach. In addition, the prediction accuracy of RapidEye, Sentinel-2 and EnMAP data was improved gradually because of more number of red-edge bands and higher spectral resolution; VI regression models that contain a single or multiple red-edge bands provided more accurate predictions of LCC than those without red-edge bands, but for normalized difference vegetation index (NDVI)-, simple ratio (SR)- and chlorophyll index (CI)-like index, two red-edge bands index didn't significantly improve the predictive accuracy of LCC than those indices with a single red-edge band. Although satellite data with higher spectral resolution and a greater number of red-edge bands marginally improve the accuracy of estimates of crop LCC, the level of this improvement remains insufficient because of higher spectral resolution, which results in a worse signal-to-noise ratio. The results of this study are helpful to accurately monitor LCC of winter wheat in large-area and provide some valuable advice for design of red-edge spectral bands of satellite sensor in future.

ESTIMATING CHLOROPHYLL A CONCENTRATIONS OF SEVERAL INLAND WATERS WITH HYPERSPECTRAL DATA AND MACHINE LEARNING MODELS

Abstract. Water is a key component of life, the natural environment and human health. For monitoring the conditions of a water body, the chlorophyll a concentration can serve as a proxy for nutrients and oxygen supply. In situ measurements of water quality parameters are often time-consuming, expensive and limited in areal validity. Therefore, we apply remote sensing techniques. During field campaigns, we collected hyperspectral data with a spectrometer and in situ measured chlorophyll a concentrations of 13 inland water bodies with different spectral characteristics. One objective of this study is to estimate chlorophyll a concentrations of these inland waters by applying three machine learning regression models: Random Forest, Support Vector Machine and an Artificial Neural Network. Additionally, we simulate four different hyperspectral resolutions of the spectrometer data to investigate the effects on the estimation performance. Furthermore, the application of first order derivatives of the spectra is evaluated in turn to the regression performance. This study reveals the potential of combining machine learning approaches and remote sensing data for inland waters. Each machine learning model achieves an R2-score between 80% to 90% for the regression on chlorophyll a concentrations. The random forest model benefits clearly from the applied derivatives of the spectra. In further studies, we will focus on the application of machine learning models on spectral satellite data to enhance the area-wide estimation of chlorophyll a concentration for inland waters.

Spatial Analysis of Temperate Forest Structure: A Geostatistical Approach to Natural Forest Potential

Forest ecosystems represent an important means of ecosystem services; they are key as carbon sinks, water collectors, soil stabilizers, suppliers of great biological diversity, among other benefits. In addition, regionalization based on forest conditions provides a valuable approach to understanding and analyzing spatial patterns, which is useful as a tool for the implementation of forest ecosystem protection and conservation programs. In this research, the structure of a temperate forest in the western Sierra Madre region of Mexico was analyzed and characterized. The study unit was the watershed and the analysis used a geospatial approach combined with multivariate techniques such as: Principal Component Analysis, Cluster Analysis (CA), Discriminant Analysis (DA) and Multivariate Analysis of Variance. We evaluated the relationships among spectral satellite data, thematic maps and structural forest variables. A total of 345 watersheds were grouped based on these variables. The grouping of watersheds under low, medium and high production conditions was carried out with CA, defining 3 groups. The validation of the grouping was performed through DA, estimating errors with the restitution method, as well as with the cross-validation method. Significant differences were found among the groups. The grouping of watersheds provides observable evidence of the variability of the forest condition throughout the area. This study allows identifying forest areas with different levels of productivity and can help to detect levels of vulnerability and ecological fragility in natural forests in temperate zones.