MODIS Reflectance Data Research Articles

Completing the machine learning saga in fractional snow cover estimation from MODIS Terra reflectance data: Random forests versus support vector regression

This study; i) investigates the suitability of two frequently employed machine learning algorithms in remote sensing, namely, random forests (RFs) and support vector regression (SVR) for fractional snow cover (FSC) estimation from MODIS Terra data, and ii) compares them with the previously proposed artificial neural networks (ANNs) and multivariate adaptive regression splines (MARS) methods over an heterogeneous and complex alpine terrain. The dataset comprises 20 Landsat 8 – MODIS image pairs that belong to European Alps acquired from Apr 2013 to Dec 2016. The fifteen image pairs are used to generate the training dataset necessary to build the models, whereas the remaining five are employed as a separate test dataset. The reference FSC maps are derived from the binary classified Landsat 8 snow/no snow maps at 30 m resolution. In order to assess the effect of sampling type and sample size, nine different training datasets are generated. The RF and SVR models are trained accordingly by using various settings of model tuning parameters. During the training of the models, MODIS top-of-atmosphere reflectance values of bands 1–7, NDSI, NDVI and land cover class are input as independent variables (i.e., predictors) to estimate the dependent variable (i.e., response), i.e., FSC value. The resolution of the generated FSC maps is 500 m. The results indicate that the ANN, MARS, RF and SVR models exhibit high consistency with reference FSC values as indicated by low RMSE (~0.14) and high R (~0.93) values. In order to analyze the effect of using three auxiliary variables, i.e., NDSI, NDVI and land cover class, to the predictive ability of the models; ANN, MARS, RF and SVR models are also trained without these predictor variables, i.e., by only using MODIS bands 1–7. The models trained without three auxiliary variables slightly differ from the ones trained with the full set of predictors by only resulting in a mean decrease in R <0.012 and a mean increase in RMSE <0.009, showing that they perform well in solving the complex functional dependencies by only using MODIS reflectance data. In terms of computational efficiencies of the proposed algorithms measured by the CPU times spent during model training, MARS and RF algorithms outperform ANN and SVR methods.

Improving Time Series Reconstruction by Fixing Invalid Values and its Fidelity Evaluation

MODIS time series data have been widely used in the research of regional and global ecosystems and climate change. For vegetation monitoring, vegetation indices such as NDVI (normalized difference vegetation index), EVI (enhanced vegetation index) and NBR (normalized burn ratio), are usually derived from MODIS reflectance data. However, noise usually makes it difficult to generate reliable time series of vegetation indices. Although some methods have been developed for reconstructing NDVI time series data, they still suffer from some limitations. First, there is no reliable approach for detecting and dealing with low-quality data, resulting in poor outcomes. Second, no effective evaluation of the fidelity of the corrected data to the original data has been discussed. For these reasons, we developed a new time series reconstruction approach, named Fixing Invalid Value (FIV) method. The proposed method assumes that the noise in surface reflectance data stems from invalid data, such as clouds, ice, and missing values. The FIV method first uses the spatially and temporally neighboring pixels to estimate the invalid values and then applies morphology operations to remove the residual noise. Finally, the Savitzky-Golay (S-G) filter is employed to generate the final results. The FIV method is tested on 8-day composite MODIS surface reflectance time series data from 2001 to 2012 in Jiangxi and Fujian provinces, China. The results show that the FIV method outperforms the conventional S-G filter and the HANTS method both in terms of visual inspection and quantitative evaluation. Furthermore, the fidelity evaluation reveals that the proposed FIV method produces high-quality time series data under all weather conditions.

Predicting forest fires burned area and rate of spread from pre-fire multispectral satellite measurements

Operational forest fire danger rating systems rely on the recent evolution of meteorological variables to estimate dead fuel condition. Further combining the latter with meteorological and environmental variables, they predict fire occurrence and spread. In this study we retrieved live fuel condition from MODIS multispectral measurements in the near infrared and shortwave infrared. Next, we combined these retrievals with an extensive dataset on actual forest fires in Campania (13,595 km2), Italy, to determine how live fuel condition affects the probability distribution functions of fire characteristics. Accordingly, the specific objective of this study was to develop and evaluate a new approach to estimate the probability distribution functions of fire burned area, duration and rate of spread as a function of the Perpendicular Moisture Index (PMI), whose value decreases with decreasing live fuel moisture content (LFMC). To this purpose, available fire data was intersected with MODIS 8-day composited reflectance data so to associate each fire event with the corresponding pre-fire PMI observation. Fires were then grouped in ten decile bins of PMI, and the conditional probability distribution functions of burned area, fire duration and rate of spread were determined in each bin. Distributions of burned area and rate of spread vary across PMI decile bins, while no significant difference was observed for fire duration. Further testing this result with a likelihood ratio test confirmed that PMI is a covariate of burned area and rate of spread, but not of fire duration. We defined an extreme event as a fire whose burned area (respectively rate of spread) exceeds the 95th percentile of the frequency distribution of all observed fire events. The probability distribution functions in the ten decile bins of PMI were combined to obtain a conditional probability distribution function, which was then used to predict the probability of extreme fires, as defined. It was found that the probability of extreme events steadily increases with decreasing PMI. Overall, at the end of the dry season the probability of extreme events is about the double than at the beginning. These results may be used to produce frequently (e.g. daily) updated maps of the probability of extreme events given a PMI map retrieved from e.g. MODIS reflectance data.

Assimilating Multiresolution Leaf Area Index of Moso Bamboo Forest from MODIS Time Series Data Based on a Hierarchical Bayesian Network Algorithm

The highly accurate multiresolution leaf area index (LAI) is an important parameter for carbon cycle simulation for bamboo forests at different scales. However, current LAI products have discontinuous resolution with 1 km mostly, that makes it difficult to accurately quantify the spatiotemporal evolution of carbon cycle at different resolutions. Thus, this study used MODIS LAI product (MOD15A2) and MODIS reflectance data (MOD09Q1) of Moso bamboo forest (MBF) from 2015, and it adopted a hierarchical Bayesian network (HBN) algorithm coupled with a dynamic LAI model and the PROSAIL model to obtain high-precision LAI data at multiresolution (i.e., 1000, 500, and 250 m). The results showed the LAIs assimilated using the HBN at the three resolutions corresponded with the actual growth trend of the MBF and correlated significantly with the observed LAI with a determination coefficient (R2) value of &gt;0.80. The highest-precision assimilated LAI was obtained at 1000-m resolution with R2 values of 0.91. The LAI assimilated using the HBN algorithm achieved better accuracy than the MODIS LAI with increases in the R2 value of 2.7 times and decreases in the root mean square error of 87.8%. Therefore, the HBN algorithm applied in this study can effectively obtain highly accurate multiresolution LAI time series data for bamboo forest.

Validation of VIIRS and MODIS reflectance data in coastal and oceanic waters: An assessment of methods

Satellite ocean color datasets have vast potentials for assessing and monitoring of marine environments. However, with the MODIS sensor aging and the VIIRS sensor reaching maturity, it is important to continuously evaluate the quality of reflectance data from both instruments. Here, we critically assess the statistical performance of both MODIS and VIIRS, including analysis of two separate (and commonly used) VIIRS processing routines. In addition, we note variability in the literature as to the methods used to identify and remove low-quality data during similar validation exercises. Although most studies use some implementation of satellite quality flags (L2 flags) and many exclude data based on spatial heterogeneity or large temporal gap from satellite overpasses, critical assessment of these methods indicates variable performance. Indeed, we found little improvement in validation statistics after implementation of these data culling techniques, with substantial variability in effectiveness between wavebands and sensors. Overall, these findings highlight the need to critically assess the impact (on both data quantity and quality) of exclusion criteria, towards more effective techniques to ensure quality and consistency of satellite ocean color datasets.

Regional Crop Gross Primary Productivity and Yield Estimation Using Fused Landsat-MODIS Data

Accurate crop yield assessments using satellite remote sensing-based methods are of interest for regional monitoring and the design of policies that promote agricultural resiliency and food security. However, the application of current vegetation productivity algorithms derived from global satellite observations is generally too coarse to capture cropland heterogeneity. The fusion of data from different sensors can provide enhanced information and overcome many of the limitations of individual sensors. In thitables study, we estimate annual crop yields for seven important crop types across Montana in the continental USA from 2008–2015, including alfalfa, barley, maize, peas, durum wheat, spring wheat and winter wheat. We used a satellite data-driven light use efficiency (LUE) model to estimate gross primary productivity (GPP) over croplands at 30-m spatial resolution and eight-day time steps using a fused NDVI dataset constructed by blending Landsat (5 or 7) and Terra MODIS reflectance data. The fused 30-m NDVI record showed good consistency with the original Landsat and MODIS data, but provides better spatiotemporal delineations of cropland vegetation growth. Crop yields were estimated at 30-m resolution as the product of estimated GPP accumulated over the growing season and a crop-specific harvest index (HIGPP). The resulting GPP estimates capture characteristic cropland productivity patterns and seasonal variations, while the estimated annual crop production results correspond favorably with reported county-level crop production data (r = 0.96, relative RMSE = 37.0%, p &lt; 0.05) from the U.S. Department of Agriculture (USDA). The performance of estimated crop yields at a finer (field) scale was generally lower, but still meaningful (r = 0.42, relative RMSE = 50.8%, p &lt; 0.05). Our methods and results are suitable for operational applications of crop yield monitoring at regional scales, suggesting the potential of using global satellite observations to improve agricultural management, policy decisions and regional/global food security.

Spatial Rice Yield Estimation Based on MODIS and Sentinel-1 SAR Data and ORYZA Crop Growth Model

Crop insurance is a viable solution to reduce the vulnerability of smallholder farmers to risks from pest and disease outbreaks, extreme weather events, and market shocks that threaten their household food and income security. In developing and emerging countries, the implementation of area yield-based insurance, the form of crop insurance preferred by clients and industry, is constrained by the limited availability of detailed historical yield records. Remote-sensing technology can help to fill this gap by providing an unbiased and replicable source of the needed data. This study is dedicated to demonstrating and validating the methodology of remote sensing and crop growth model-based rice yield estimation with the intention of historical yield data generation for application in crop insurance. The developed system combines MODIS and SAR-based remote-sensing data to generate spatially explicit inputs for rice using a crop growth model. MODIS reflectance data were used to generate multitemporal LAI maps using the inverted Radiative Transfer Model (RTM). SAR data were used to generate rice area maps using MAPScape-RICE to mask LAI map products for further processing, including smoothing with logistic function and running yield simulation using the ORYZA crop growth model facilitated by the Rice Yield Estimation System (Rice-YES). Results from this study indicate that the approach of assimilating MODIS and SAR data into a crop growth model can generate well-adjusted yield estimates that adequately describe spatial yield distribution in the study area while reliably replicating official yield data with root mean square error, RMSE, of 0.30 and 0.46 t ha−1 (normalized root mean square error, NRMSE of 5% and 8%) for the 2016 spring and summer seasons, respectively, in the Red River Delta of Vietnam, as evaluated at district level aggregation. The information from remote-sensing technology was also useful for identifying geographic locations with peculiarity in the timing of rice establishment, leaf growth, and yield level, and thus contributing to the spatial targeting of further investigation and interventions needed to reduce yield gaps.

Modeling three-dimensional forest structures to drive canopy radiative transfer simulations of bidirectional reflectance factor

ABSTRACTThree-dimensional (3-D) Monte Carlo-based radiative transfer (MCRT) models are usually used for benchmarking in intercomparisons of the canopy radiative transfer (RT) simulations. However, the 3-D MCRT models are rarely applied to develop remote sensing algorithms to estimate essential climate variables of forests, due mainly to the difficulties in obtaining realistic stand structures for different forest biomes over regional to global scales. Fortunately, some of important tree structure parameters such as canopy height and tree density distribution have been available globally. This enables to run the intermediate complexities of the 3-D MCRT models. We consequently developed a statistical approach to generate forest structures with intermediate complexities depending on the inputs of canopy height and tree density. It aims at facilitating applications of the 3-D MCRT models to develop remote sensing retrieval algorithms. The proposed approach was evaluated using field measurements of two boreal forest stands at Estonia and USA, respectively. Results demonstrated that the simulations of bidirectional reflectance factor (BRF) based on the measured forest structures agreed well with the BRF based on the generated structures from the proposed approach with the root mean square error (RMSE) and relative RMSE (rRMSE) ranging from 0.002 to 0.006 and from 0.7% to 19.8%, respectively. Comparison of the computed BRF with corresponding MODIS reflectance data yielded RMSE and rRMSE lower than 0.03 and 20%, respectively. Although the results from the current study are limited in two boreal forest stands, our approach has the potential to generate stand structures for different forest biomes.

Coupled LAI assimilation and BEPS model for analyzing the spatiotemporal pattern and heterogeneity of carbon fluxes of the bamboo forest in Zhejiang Province, China

Bamboo forests are widely distributed in the tropical and sub-tropical regions, and have a great carbon sequestration capacity. However, the carbon dynamics of bamboo forest are still poorly understood. This study developed an algorithm for assimilating LAI dynamics based on an integrated ensemble Kalman filter using MODIS LAI data, MODIS reflectance data, and canopy reflectance data modeled by PROSAIL for two typical bamboo forests. The assimilated LAI was highly consistent with the observed LAI, yielding correlation coefficient (r) of 0.84 and 0.94 for Moso bamboo forest and Lei bamboo forest, respectively. Then, the spatiotemporal pattern of the CO2 fluxes within bamboo forest were modeled for Zhejiang Province during 2011–2014 using the Boreal Ecosystem Productivity Simulator model driven by assimilated LAI. The modeled CO2 fluxes were in agreement with the CO2 fluxes measured by eddy covariance technology at two selected sites within the study region. Average r, absolute deviation, and root mean square error values were 0.77, 0.65gCm−2day−1, and 0.82gCm−2day−1, respectively. Bamboo stands with high Net Ecosystem Exchange (NEE) values were found to be distributed throughout mountainous areas in the northwest, southwest, and northeast of Zhejiang, while low NEE stands were located in the mountain basin areas, high altitude areas, and the majority of coastal areas. Among the eleven cities in Zhejiang Province, Hangzhou was the highest annual total NEE city (0.34±0.06TgCyr−1), and Huzhou was the highest NEE density city (7.65±1.02MgChm−2yr−1). The carbon sequestrated by bamboo forest was 1.74TgCyr−1, accounted 6.36Tg CO2 yr−1, which indicated the bamboo forested area in Zhejiang Province was a carbon sink. The spatial heterogeneity of CO2 fluxes of bamboo forest can be revealed by an exponential semivariance model. The analysis of the model structure indicated that the spatial heterogeneity of CO2 fluxes of bamboo forest was characterized by significant autocorrelation, with an average autocorrelation range of 3.00km, however the effect of random factors (management levels, management measures, etc.) cannot be ignored. This is the first time that the CO2 fluxes of bamboo forest in Zhejiang Province have been mapped and this study provides a new estimate of CO2 fluxes in this region.

Assimilating leaf area index of three typical types of subtropical forest in China from MODIS time series data based on the integrated ensemble Kalman filter and PROSAIL model

Subtropical forest ecosystems play essential roles in the global carbon cycle and in carbon sequestration functions, which challenge the traditional understanding of the main functional areas of carbon sequestration in the temperate forests of Europe and America. The leaf area index (LAI) is an important biological parameter in the spatiotemporal simulation of the carbon cycle, and it has considerable significance in carbon cycle research. Dynamic retrieval based on remote sensing data is an important method with which to obtain large-scale high-accuracy assessments of LAI. This study developed an algorithm for assimilating LAI dynamics based on an integrated ensemble Kalman filter using MODIS LAI data, MODIS reflectance data, and canopy reflectance data modeled by PROSAIL, for three typical types of subtropical forest (Moso bamboo forest, Lei bamboo forest, and evergreen and deciduous broadleaf forest) in China during 2014–2015. There were some errors of assimilation in winter, because of the bad data quality of the MODIS product. Overall, the assimilated LAI well matched the observed LAI, with R2 of 0.82, 0.93, and 0.87, RMSE of 0.73, 0.49, and 0.42, and aBIAS of 0.50, 0.23, and 0.03 for Moso bamboo forest, Lei bamboo forest, and evergreen and deciduous broadleaf forest, respectively. The algorithm greatly decreased the uncertainty of the MODIS LAI in the growing season and it improved the accuracy of the MODIS LAI. The advantage of the algorithm is its use of biophysical parameters (e.g., measured LAI) in the LAI assimilation, which makes it possible to assimilate long-term MODIS LAI time series data, and to provide high-accuracy LAI data for the study of carbon cycle characteristics in subtropical forest ecosystems.

Assessing total water storage and identifying flood events over Tonlé Sap basin in Cambodia using GRACE and MODIS satellite observations combined with hydrological models

In this study, satellite observations including gravity (GRACE), terrestrial reflectance (MODIS), and global precipitation (TRMM) data, along with the output from the PCR-GLOBWB hydrological model, are used to generate monthly and sub-monthly terrestrial water storage (TWS) estimates and quantify flood events over the Tonle Sap basin between 2002 and 2014. This study is the first time GRACE data have been used to investigate the hydrological processes over the Tonle Sap basin. To improve the accuracy of the TWS estimates from GRACE, a signal restoration method was applied in an effort to recover the signal loss (i.e., signal leakage) inherent in the standard GRACE post-processing scheme. The method applies the correction based on the GRACE observations only, requiring no external data or hydrological models. The effectiveness of the technique over the Tonle Sap basin was validated against several independent data sets. Based on the GRACE observations since 2002, the 2011 and 2013 flood events were clearly identified, and measured to have basin-averaged TWS values of 42 cm (40% higher than the long-term mean peak value) and 36 cm (34% higher) equivalent water height, respectively. Those same years also coincide with the largest observed flood extents, estimated from the MODIS data as 6561 km2 (91% above the long-term mean peak value) and 5710 km2 (66% above), respectively. Those flood events are also linked to the observed inter-annual variations of water storage between 2010 and 2014. It was shown that those inter-annual variations mainly reflect the variations in the surface water and groundwater storage components, influenced by the change of the precipitation intensity. In addition, this study presents a new approach for deriving monthly and sub-monthly TWS variations over a regularly inundated area by using MODIS reflectance data in addition to GRACE solutions. The results of this study show that GRACE data can be considered as an effective tool for monitoring certain small-scale (82,000 km2) hydrological basins.

Remote Sensing of River Delta Inundation: Exploiting the Potential of Coarse Spatial Resolution, Temporally-Dense MODIS Time Series

River deltas belong to the most densely settled places on earth. Although they only account for 5% of the global land surface, over 550 million people live in deltas. These preferred livelihood locations, which feature flat terrain, fertile alluvial soils, access to fluvial and marine resources, a rich wetland biodiversity and other advantages are, however, threatened by numerous internal and external processes. Socio-economic development, urbanization, climate change induced sea level rise, as well as flood pulse changes due to upstream water diversion all lead to changes in these highly dynamic systems. A thorough understanding of a river delta’s general setting and intra-annual as well as long-term dynamic is therefore crucial for an informed management of natural resources. Here, remote sensing can play a key role in analyzing and monitoring these vast areas at a global scale. The goal of this study is to demonstrate the potential of intra-annual time series analyses at dense temporal, but coarse spatial resolution for inundation characterization in five river deltas located in four different countries. Based on 250 m MODIS reflectance data we analyze inundation dynamics in four densely populated Asian river deltas—namely the Yellow River Delta (China), the Mekong Delta (Vietnam), the Irrawaddy Delta (Myanmar), and the Ganges-Brahmaputra (Bangladesh, India)—as well as one very contrasting delta: the nearly uninhabited polar Mackenzie Delta Region in northwestern Canada for the complete time span of one year (2013). A complex processing chain of water surface derivation on a daily basis allows the generation of intra-annual time series, which indicate inundation duration in each of the deltas. Our analyses depict distinct inundation patterns within each of the deltas, which can be attributed to processes such as overland flooding, irrigation agriculture, aquaculture, or snowmelt and thermokarst processes. Clear differences between mid-latitude, subtropical, and polar deltas are illustrated, and the advantages and limitations of the approach for inundation derivation are discussed.

Sensitivity analysis of 6S-based look-up table for surface reflectance retrieval

We created a look-up table (LUT) based on the Second Simulation of the Satellite Signal in the Solar Spectrum (6S) radiative transfer model, which reduces large errors in the surface reflectance retrieval under high solar zenith angle (SZA) conditions. The LUT was calculated in 10° SZA intervals containing pre-computed atmospheric correction coefficients as a function of discretized pre-defined input parameters. In order to validate the performance of the LUT, we compared the retrieved surface reflectance using the LUT against a retrieval performed using the simplified method of atmospheric correction (SMAC). These results were validated against MODIS reflectance data (MOD09). The surface reflectance obtained using the LUT was highly correlated with the MOD09, with a coefficient of determination (R2) of 0.88 (red band) and 0.94 (NIR). The retrieved surface reflectance had a root mean-squared error of 0.0132 (red band) and 0.0191 (NIR). Accuracy of surface reflectance retrieved using our LUT with a 10° SZA interval was better than that of the obtained using SMAC. However, certain errors were still present particularly at high SZAs. In order to increase the accuracy at high SZAs, new LUT was computed with a finer SZA interval (5°) at high SZAs. In both red and NIR bands, the R2, fine SZA interval LUT (0.92) were compared to the coarse SZA interval LUT (0.74) of around 65°. Additionally, the run time for surface reflectance retrievals with our LUT was almost comparable to that of the SMAC, an operational model. This study demonstrates that proper SZAs interval for making LUT in high SZA range.

Monitoring the spreading of the Amazon freshwater plume by MODIS, SMOS, Aquarius, and TOPAZ

AbstractA synergistic tool for studying the Amazon River plume dynamics based on a novel algorithm for deriving sea surface salinity (SSS) from MODIS reflectance data together with SSS data from the SMOS and Aquarius satellites and the TOPAZ data assimilation system is proposed. The new algorithm is based on a neural network to relate spectral remote sensing reflectance measured by MODIS with SSS measured by SMOS in the Amazon River plume. The algorithm is validated against independent in situ data and is found to be valid in the range of SSS from 29 to 35 psu, for the period of highest rates of Amazon River discharge with RMSE = 0.79 psu and r2 = 0.84. Monthly SSS fields were reconstructed from the MODIS data for late summers from 2002 to 2012 at a 10 km resolution and compared to surface currents and SSS derived from the TOPAZ reanalysis system. The two data sets reveal striking agreement, suggesting that the TOPAZ system could be used for a detailed study of the Amazon River plume dynamics. Both the position and speed of the North Brazilian Current as well as the spreading of the Amazon River plume are monitored. In particular, a recurrent mechanism was observed for the spreading of the rivers plumes, notably that the fresh water is usually advected toward the Caribbean Sea by the North Brazilian Current but get diverted into the tropical Atlantic when North Brazilian Current Rings are shed.

江苏省农作物最大光能利用率时空特征及影响因子分析

基于遥感数据的光能利用率模型被广泛应用于计算陆地生态系统的生产力，但其结果对最大光能利用率（<em>ε</em>_max）参数非常敏感。由于该参数存在明显的时空变异性，在区域尺度上难以确定。利用MODIS遥感数据、分县产量统计数据和VPM模型推算了2001-2010年江苏省各县逐年的农田<em>ε</em>_max，分析了其时空变化特征及其可能原因。结果表明，在2001-2010年期间，江苏省61个县区农田<em>ε</em>_max平均值的变化范围为0.757-3.435 g C/MJ，呈现北高南低、中间高四周较低的空间分布特征；各县区农田的<em>ε</em>_max都呈现出上升趋势，但在2001-2006年期间存在明显的年际波动，2002年、2004年和2006年的<em>ε</em>_max相对较低，2007年后全省农田<em>ε</em>_max稳定上升；<em>ε</em>_max的年际波动呈现由北向南递减的趋势；全省大部分地区<em>ε</em>_max的年际变化与单位耕地面积农用化肥施用量呈正相关性，苏北北部尤为明显；同时也与C₄作物产量所占比例相关。研究表明在利用光能利用率模型计算农田生产力时，需要发展能考虑<em>ε</em>_max时空变化的参数化方案。;The productivity is a critical component of the terrestrial carbon cycle. Remote sensing driven light use efficiency models have been widely utilized to calculate the productivity of terrestrial ecosystems. They calculate the productivity as a linear product of maximum light use efficiency <em>ε</em>_max, scaling factors of environmental stress, and absorbed photosynthetically active radiation (APAR). The outputs of these models are very sensitive to t maximum light use efficiency (<em>ε</em>_max), which represents the utilization rate of photosynthetically active radiation under ideal conditions without restrictions. However, it is difficult to determine this parameter properly since it is affected by numerous factors, including canopy nitrogen content, atmospheric concentration of CO₂, plant types and species, and exhibit significant spatial and temporal variations. In currently used light use efficiency models, this parameter only changes with plant types. Such simplified parameterization might induce large uncertainties in calculated productivity of terrestrial ecosystems at regional and global scales. In this study, MODIS reflectance data, county\level yield census data, locally observed meteorological data, and the vegetation photosynthesis model (VPM) were employed to derive annual mean county\level cropland <em>ε</em>_max in Jiangsu province for the period from 2001 to 2010. The county\level yield census data was converted into total net primary productivity of each county for estimating the <em>ε</em>_max parameter in the VPM model. Then, the spatial and temporal variations of <em>ε</em>_max and possible driving factors were analyzed were analyzed. The results show that the county\level means of cropland <em>ε</em>_max in 61 counties of Jiangsu province varied between 0.757-3.435 g C/MJ during the period from 2001 to 2010, with an overall gradient decreasing from the north to the south. It is higher in the central areas than in the surrounding areas. Annual mean cropland <em>ε</em>_max increased during the period from 2001 to 2010 in all 61 counties, but showed obvious interannual fluctuations during the period from 2001 to 2006. It was relatively lower in years 2002, 2004 and 2006 and steadily increased in all counties since 2007. The interannual fluctuations of <em>ε</em>_max were normally higher in the north than in the south. The annual means of <em>ε</em>_max were positively related to the amount of fertilizer used in per unit area of cultivated cropland in most counties, particularly in the northern areas of Jiangsu Province. The increasing rates of <em>ε</em>_max were positively correlated with the increasing rates of fertilizer utilization per unit area. The interannual fluctuations of <em>ε</em>_max were also related to the yield fraction of C₄ crops (corn). This study proves that it is of importance to develop a parameterization scheme accounting for the temporal and spatial variations of <em>ε</em>_max for improving the calculation of productivity in croplands using light use efficiency models and remote sensing data. It should be kept in mind that some factors might induce uncertainties estimated <em>ε</em>_max, such as the uncertainties in yield census data, the parameters used to converts yield into net primary productivity, methodologies used to calculate APAR and scaling factors of environmental stress. The underlying mechanisms for driving significant spatial and temporal variations of <em>ε</em>_max need further thorough investigation. Nevertheless, the findings in this study can provide a value reference for optimizing the calculation of carbon fluxes between the atmosphere and terrestrial ecosystems.

A dynamic Bayesian network data fusion algorithm for estimating leaf area index using time-series data from in situ measurement to remote sensing observations

Leaf area index (LAI) products retrieved from remote sensing observations have been widely used in the fields of ecosphere, atmosphere etc. However, because satellite-observed images are captured instantaneously and sometimes screened by cloud, some current LAI products are inherently discontinuous in time and their accuracy may not meet the needs of users well. To solve these problems, we proposed a dynamic Bayesian network (DBN)-based data fusion algorithm that integrates dynamic crop growth information, a canopy reflectance (CR) model and remote sensing observations from the perspective of Bayesian probability. Using the proposed algorithm, LAI was estimated using data sets from both field measurements for winter wheat in Beijing, China, and MODIS reflectance data at two American flux tower sites. Results showed good agreement between the LAI estimated by the DBN-based data fusion method and the true ground LAI, with a correlation coefficient of (R) 0.95 and 0.96, respectively, and a corresponding root mean square error (RMSE) of 0.35 and 0.49, respectively. In addition, the LAI estimated by the DBN-based data fusion method formed a continuous time series and was consistent with the variety law of vegetation growth at both plot and flux tower site scales. It has been demonstrated that the proposed DBN-based data fusion algorithm has the potential to be used to accurately estimate LAI and to fill the temporal gap by integrating information from multiple sources.

Correlation between potato yield and MODIS‐derived vegetation indices

Monitoring of crop growth and forecasting its yield well before harvest is very important for crop and food management. Remote sensing images are capable of identifying crop health, as well as predicting its yield. Vegetation indices (VIs), such as the normalized difference vegetation index (NDVI), leaf area index (LAI) and fraction of photosynthetically active radiation (fPAR) calculated from remotely sensed data have been widely used to monitor crop growth and to predict crop yield. This study used 8 day TERRA MODIS reflectance data of 500 m resolution for the years 2005 to 2006 to estimate the yield of potato in the Munshiganj area of Bangladesh. The satellite data has been validated using ground truth data from fields of 50 farmers. Regression models are developed between VIs and field level potato yield for six administrative units of Munshiganj District. The yield prediction equations have high coefficients of correlation (R 2) and are 0.84, 0.72 and 0.80 for the NDVI, LAI and fPAR, respectively. These equations were validated by using data from 2006 to 2007 seasons and found that an average error of estimation is about 15% for the study region. It can be concluded that VIs derived from remote sensing can be an effective tool for early estimation of potato yield.

A new data fusion model for high spatial- and temporal-resolution mapping of forest disturbance based on Landsat and MODIS

Investigating the temporal and spatial pattern of landscape disturbances is an important requirement for modeling ecosystem characteristics, including understanding changes in the terrestrial carbon cycle or mapping the quality and abundance of wildlife habitats. Data from the Landsat series of satellites have been successfully applied to map a range of biophysical vegetation parameters at a 30 m spatial resolution; the Landsat 16 day revisit cycle, however, which is often extended due to cloud cover, can be a major obstacle for monitoring short term disturbances and changes in vegetation characteristics through time. The development of data fusion techniques has helped to improve the temporal resolution of fine spatial resolution data by blending observations from sensors with differing spatial and temporal characteristics. This study introduces a new data fusion model for producing synthetic imagery and the detection of changes termed Spatial Temporal Adaptive Algorithm for mapping Reflectance Change (STAARCH). The algorithm is designed to detect changes in reflectance, denoting disturbance, using Tasseled Cap transformations of both Landsat TM/ETM and MODIS reflectance data. The algorithm has been tested over a 185 × 185 km study area in west-central Alberta, Canada. Results show that STAARCH was able to identify spatial and temporal changes in the landscape with a high level of detail. The spatial accuracy of the disturbed area was 93% when compared to the validation data set, while temporal changes in the landscape were correctly estimated for 87% to 89% of instances for the total disturbed area. The change sequence derived from STAARCH was also used to produce synthetic Landsat images for the study period for each available date of MODIS imagery. Comparison to existing Landsat observations showed that the change sequence derived from STAARCH helped to improve the prediction results when compared to previously published data fusion techniques.

Investigating modelled and observed Terra/MODIS 500-m reflectance data for viewing and illumination effects

MODIS data are needed for the development of a light-use efficiency model over Swedish forests. Sweden stretches between Lat. 55N. and Lat. 69N. and MODIS reflectance data are recorded at vastly different sun and viewing angles. We have investigated modelled and observed red and near-infrared (NIR) MODIS data, and the normalised difference vegetation index (NDVI), in order to enable the development of a processing scheme that can be used for correcting MODIS data for directional effects. Directional reflectances have been simulated with the Kuusk–Nilson forest reflectance model, based on data from the Swedish National Forest Inventory. Modelled BRDF data are stored in a look-up table (LUT) and the possibility of finding a match with the observed data is tested. The analysis shows that matching may be based on normalised data rather than absolute values. Directional effects in the modelled data differ between sparse, medium and dense forest stands. The matching of observed MODIS data to the model data stored in the LUT is severely hampered by the scatter in the MODIS data. Reducing this scatter, e.g., by an improved cloud screening and improved atmospheric correction, will be necessary for the development of a reliable correction scheme.

Comparisons of land cover and LAI estimates derived from ETM+ and MODIS for four sites in North America: a quality assessment of 2000/2001 provisional MODIS products

The MODIS land science team produces a number of standard products, including land cover and leaf area index (LAI). Critical to the success of MODIS and other sensor products is an independent evaluation of product quality. In that context, we describe a study using field data and Landsat ETM+ to map land cover and LAI at four 49-km 2 sites in North America containing agricultural cropland (AGRO), prairie grassland (KONZ), boreal needleleaf forest, and temperate mixed forest. The purpose was to: (1) develop accurate maps of land cover, based on the MODIS IGBP (International Geosphere–Biosphere Programme) land cover classification scheme; (2) derive continuous surfaces of LAI that capture the mean and variability of the LAI field measurements; and (3) conduct initial MODIS validation exercises to assess the quality of early (i.e., provisional) MODIS products. ETM+ land cover maps varied in overall accuracy from 81% to 95%. The boreal forest was the most spatially complex, had the greatest number of classes, and the lowest accuracy. The intensive agricultural cropland had the simplest spatial structure, the least number of classes, and the highest overall accuracy. At each site, mapped LAI patterns generally followed patterns of land cover across the site. Predicted versus observed LAI indicated a high degree of correspondence between field-based measures and ETM+ predictions of LAI. Direct comparisons of ETM+ land cover maps with Collection 3 MODIS cover maps revealed several important distinctions and similarities. One obvious difference was associated with image/map resolution. ETM+ captured much of the spatial complexity of land cover at the sites. In contrast, the relatively coarse resolution of MODIS did not allow for that level of spatial detail. Over the extent of all sites, the greatest difference was an overprediction by MODIS of evergreen needleleaf forest cover at the boreal forest site, which consisted largely of open shrubland, woody savanna, and savanna. At the agricultural, temperate mixed forest, and prairie grassland sites, ETM+ and MODIS cover estimates were similar. Collection 3 MODIS-based LAI estimates were considerably higher (up to 4 m 2 m −2) than those based on ETM+ LAI at each site. There are numerous probable reasons for this, the most important being the algorithms' sensitivity to MODIS reflectance calibration, its use of a prelaunch AVHRR-based land cover map, and its apparent reliance on mainly red and near-IR reflectance. Samples of Collection 4 LAI products were examined and found to consist of significantly improved LAI predictions for KONZ, and to some extent for AGRO, but not for the other two sites. In this study, we demonstrate that MODIS reflectance data are highly correlated with LAI across three study sites, with relationships increasing in strength from 500 to 1000 m spatial resolution, when shortwave-infrared bands are included.