Gross Primary Production Products Research Articles

Modeling GPP in the Nordic forest landscape with MODIS time series data—Comparison with the MODIS GPP product

Satellite sensor-derived data are suitable for regional estimations of several important biophysical variables. Data with a finer spatial resolution should improve regional estimations of GPP (gross primary productivity), since they better capture the variation in a heterogeneous landscape. The main objective of this study was to investigate if MODIS 500m reflectance data can be used to drive empirical models for regional estimations of GPP in Nordic forests. The performance of the proposed models was compared with the MODIS 1km GPP product. Linear regression analyses were made on 8-day averages of eddy covariance GPP from three deciduous and ten coniferous sites in relation to MODIS 8-day composite data and 8-day averages of modeled incoming PPFD (photosynthetic photon flux density). Time series of EVI2 (two-band enhanced vegetation index) were calculated from MODIS 500m reflectance data and smoothed by a curve fitting procedure. For most sites, GPP was fairly strongly to strongly related to the product of EVI2 and PPFD (Deciduous: R2=0.45–0.86, Coniferous: R2=0.49–0.90). Similar strengths were found between GPP and the product of EVI2 and MODIS 1km daytime LST (land surface temperature) (R2=0.55–0.81, 0.57–0.77) and between GPP and EVI2, PPFD and daytime LST in multiple linear regressions (R2=0.73–0.89, 0.65–0.93). One year of data was collected from all coniferous sites to derive a general empirical model for GPP versus (1) the product of EVI2 and PPFD (R2=0.70), (2) the product of EVI2 and daytime LST (R2=0.62) and (3) EVI2, PPFD and daytime LST (R2=0.72). These three models were then validated at six sites for the remaining years by linearly relating eddy covariance GPP to modeled GPP, which resulted in fairly strong to strong relationships for most sites (R2=0.49–0.91, RMSE=0.63–1.22gCm−2day−1, R2=0.53–0.73, RMSE=0.90–1.43gCm−2day−1, R2=0.56–0.87, RMSE=0.79–1.11gCm−2day−1). In comparison, similar validation strengths were found for the latest collection 5.1 of the MODIS 1km GPP product (R2=0.59–0.88, RMSE=0.80–1.16gCm−2day−1). The main conclusion is that the suggested empirical models driven by MODIS 500m reflectance data can be used for regional estimations of Nordic forest GPP, while preserving a finer resolution than the MODIS 1km GPP product.

Remote Sensing Evaluation of CLM4 GPP for the Period 2000–09*

Abstract Remote sensing can provide long-term and large-scale products helpful for ecosystem model evaluation. The authors compare monthly gross primary production (GPP) simulated by the Community Land Model, version 4 (CLM4) at a half-degree resolution with satellite estimates of GPP from the Moderate Resolution Imaging Spectroradiometer (MODIS) GPP product (MOD17) for the 10-yr period January 2000–December 2009. The assessment is presented in terms of long-term mean carbon assimilation, seasonal mean distributions, amplitude and phase of the annual cycle, and intraannual and interannual GPP variability and their responses to climate variables. For the long-term annual and seasonal means, major GPP patterns are clearly demonstrated by both products. Compared to the MODIS product, CLM4 overestimates the magnitude of GPP for tropical evergreen forests. CLM4 has a longer carbon uptake period than MODIS for most plant functional types (PFTs) with an earlier onset of GPP in spring and a later decline of GPP in autumn. Empirical orthogonal function analysis of the monthly GPP changes indicates that, on the intraannual scale, both CLM4 and MODIS display similar spatial representations and temporal patterns for most terrestrial ecosystems except in northeast Russia and in the very dry region of central Australia. For 2000–09, CLM4 simulated increases in annual averaged GPP over both hemispheres; however, estimates from MODIS suggest a reduction in the Southern Hemisphere (−0.2173 PgC yr−1), balancing the significant increase over the Northern Hemisphere (0.2157 PgC yr−1). The evaluations highlight strengths and weaknesses of the CLM4 primary production and illuminate potential improvements and developments.

Use of a vegetation index model to estimate gross primary production in open grassland

Observations from satellite remote sensing provide important tools for regional and global monitoring of terrestrial ecosystems. This article evaluated a recently developed vegeta- tion index (VI) model that estimates gross primary production (GPP) in an open grassland using flux measurements and 250-m moderate-resolution imaging spectroradiometer (MODIS) images. Two VIs, the two-band enhanced vegetation index (EVI2) and the normalized difference vegetation index (NDVI), were tested as indicators of both fraction of absorbed photosynthetic active radiation (f APAR) and light use efficiency (LUE). The standard MODIS GPP product was moderately correlated with flux-measured GPP as indicated by the coefficients of determination (R 2 ) of 0.56 and a root mean square error (RMSE) of 15.4 gCm −2 8d −1 . In comparison, results suggested that the VI model can provide improved estimates of GPP, with R 2 of 0.80, 0.82, and 0.81 for EVI2 × EVI2 × photosynthetically active radiation (PAR), NDVI × EVI2 × PAR, and NDVI × NDVI × PAR, respectively. The underlying reason was that both EVI2 and NDVI were potential proxies of f APAR (R 2 of 0.43 and 0.53, respectively) and LUE (R 2 of 0.52 and 0.46, respectively). The primary analysis will be helpful for spatial improvements of future GPP models. © 2012 Society of Photo-Optical Instrumentation Engineers (SPIE). (DOI: 10.1117/1.JRS.6

Exploring Simple Algorithms for Estimating Gross Primary Production in Forested Areas from Satellite Data

Algorithms that use remotely-sensed vegetation indices to estimate gross primary production (GPP), a key component of the global carbon cycle, have gained a lot of popularity in the past decade. Yet despite the amount of research on the topic, the most appropriate approach is still under debate. As an attempt to address this question, we compared the performance of different vegetation indices from the Moderate Resolution Imaging Spectroradiometer (MODIS) in capturing the seasonal and the annual variability of GPP estimates from an optimal network of 21 FLUXNET forest towers sites. The tested indices include the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Leaf Area Index (LAI), and Fraction of Photosynthetically Active Radiation absorbed by plant canopies (FPAR). Our results indicated that single vegetation indices captured 50–80% of the variability of tower-estimated GPP, but no one index performed universally well in all situations. In particular, EVI outperformed the other MODIS products in tracking seasonal variations in tower-estimated GPP, but annual mean MODIS LAI was the best estimator of the spatial distribution of annual flux-tower GPP (GPP = 615 × LAI − 376, where GPP is in g C/m2/year). This simple algorithm rehabilitated earlier approaches linking ground measurements of LAI to flux-tower estimates of GPP and produced annual GPP estimates comparable to the MODIS 17 GPP product. As such, remote sensing-based estimates of GPP continue to offer a useful alternative to estimates from biophysical models, and the choice of the most appropriate approach depends on whether the estimates are required at annual or sub-annual temporal resolution.

Remote sensing of canopy light use efficiency in temperate and boreal forests of North America using MODIS imagery

Light use efficiency (LUE) is an important variable characterizing plant eco-physiological functions and refers to the efficiency at which absorbed solar radiation is converted into photosynthates. The estimation of LUE at regional to global scales would be a significant advantage for global carbon cycle research. Traditional methods for canopy level LUE determination require meteorological inputs which cannot be easily obtained by remote sensing. Here we propose a new algorithm that incorporates the enhanced vegetation index (EVI) and a modified form of land surface temperature (Tm) for the estimation of monthly forest LUE based on Moderate Resolution Imaging Spectroradiometer (MODIS) imagery. Results demonstrate that a model based on EVI×Tm parameterized from ten forest sites can provide reasonable estimates of monthly LUE for temperate and boreal forest ecosystems in North America with an R2 of 0.51 (p<0.001) for the overall dataset. The regression coefficients (a, b) of the LUE–EVI×Tm correlation for these ten sites have been found to be closely correlated with the average EVI (EVI_ave, R2=0.68, p=0.003) and the minimum land surface temperature (LST_min, R2=0.81, p=0.009), providing a possible approach for model calibration. The calibrated model shows comparably good estimates of LUE for another ten independent forest ecosystems with an overall root mean square error (RMSE) of 0.055gC per mol photosynthetically active radiation. These results are especially important for the evergreen species due to their limited variability in canopy greenness. The usefulness of this new LUE algorithm is further validated for the estimation of gross primary production (GPP) at these sites with an RMSE of 37.6gCm−2month−1 for all observations, which reflects a 28% improvement over the standard MODIS GPP products. These analyses should be helpful in the further development of ecosystem remote sensing methods and improving our understanding of the responses of various ecosystems to climate change.

A new approach for estimating northern peatland gross primary productivity using a satellite-sensor-derived chlorophyll index

Carbon flux models that are largely driven by remotely sensed data can be used to estimate gross primary productivity (GPP) over large areas, but despite the importance of peatland ecosystems in the global carbon cycle, relatively little attention has been given to determining their success in these ecosystems. This paper is the first to explore the potential of chlorophyll-based VI models for estimating peatland GPP from satellite data. Using several years of carbon flux data from contrasting peatlands, we explored the relationships between the MERIS Terrestrial Chlorophyll Index (MTCI) and GPP, and determined whether the inclusion of environmental variables such as PAR and temperature, thought to be important determinants of peatland carbon flux, improved upon direct relationships. To place our results in context, we compared the newly developed GPP models with the MODIS (Moderate Resolution Imaging Spectrometer) GPP product. Our results show that simple MTCI-based models can be used for estimates of inter- and intra annual variability in peatland GPP. The MTCI is a good indicator of GPP and compares favourably with more complex products derived from the MODIS sensor on a site specific basis. The incorporation of MTCI into a light use efficiency type model, by means of partitioning the fraction of photosynthetic material within a plant canopy, shows most promise for peatland GPP estimation, outperforming all other models. Our results show great promise and demonstrate that satellite data specifically related to vegetation chlorophyll content, may ultimately facilitate improved quantification of peatland carbon flux dynamics

Site-level evaluation of MODIS-based primary production in an old-growth forest in Northeast China

To improve the accuracy of the moderate resolution imaging spectroradiometer (MODIS) gross primary production (GPP) algorithm, it is critical to evaluate MODIS GPP production for different land cover types using ground-based measurements. In this paper the MODIS primary production products (MOD17) is evaluated by using site-specific input parameters to the algorithm and compared to the results of eddy covariance measurements over an old-growth forest. Direct comparisons suggest that 8-day GPP predicted by the standard MODIS algorithm was highly correlated with flux tower based GPP (r2 = 0.77, P<0.001), while with an average underestimation of 39%. The difference is substantial in magnitude mainly because the inputs of underestimated MODIS biome-specific parameters, maximum light use efficiency emax and MODIS derived fraction of photosynthetically active radiation. The modified MODIS algorithm GPP calculated with site-specific input parameters compares favorably with ground flux tower observations (r2 = 0.92, relative error = 7%). These results suggest that the MODIS GPP production is most likely underpredicted in forest sites with high primary production, and site-specific input parameters could help to improve the accuracy of MODIS GPP algorithm.

Comparison of multiple models for estimating gross primary production using MODIS and eddy covariance data in Harvard Forest

Gross primary production (GPP) defined as the overall rate of fixation of carbon through the process of vegetation photosynthesis is important for carbon cycle and climate change research. Three models, the Vegetation Photosynthesis Model (VPM), the Temperature and Greenness (TG) model and the Vegetation Index (VI) model have been compared for the estimation of GPP in Harvard Forest from 2003 to 2006 using climate variables acquired by eddy covariance (EC) measurements and Moderate Resolution Imaging Spectroradiometer (MODIS) satellite images. All these models provide more reliable estimates of GPP than that of MODIS GPP product. High Pearsons correlation coefficients r equal to 0.94, 0.92 and 0.90 are observed for the VPM, the TG and the VI model, respectively. Relationships between GPP and land surface temperature (LST, R 2 = 0.72), and vapor pressure deficit (VPD, R 2 = 0.45) indicate that climate variables are important for GPP estimation. Due to proper characterization of temperature, water stress and leaf age by three scalars, VPM best follows the seasonal variations of GPP. By incorporation of the MODIS surface reflectance and LST product, the TG model is the most suitable choice for areas without prior knowledge as it is based entirely on remote sensing observations. Results from the VI model demonstrate the possibility of using a single vegetation index for light use efficiency (LUE) estimation in deciduous forest that is of high spatial heterogeneity. The validation and comparison of models will be helpful in development of future GPP models using combinations of climate variables and/or remote sensing observations.

Medium Spatial Resolution Satellite Imagery to Estimate Gross Primary Production in an Urban Area

Remote sensing data with medium spatial resolution can provide useful information about Gross Primary Production (GPP), especially on the scale of urban areas. Most models of ecosystem carbon exchange that are based on remote sensing use some form of the light use efficiency (LUE) model. The aim of this work is to analyze the distribution of annual GPP in the urban area of Denpasar, Bali. Additional analysis using two types of satellite data (ALOS/AVNIR-2 and Aster) addresses the impact of spatial resolution on the detection of various ecosystem processes in Denpasar. Annual GPP estimated using ALOS/AVNIR-2 varied from 0.13 gC m−2 yr−1 to 2,586.18 gC m−2 yr−1. Meanwhile, the Aster estimate varied from 0.14 gC m−2 yr−1 to 2,595.26 gC m−2 yr−1. GPP as measured by ALOS/AVNIR-2 was lower than that from Aster because ALOS/AVNIR-2 has medium spatial resolution and a smaller spectral range than Aster. Variations in land use may influence the measured value of GPP via differences in vegetation type, distribution, and photosynthetic pathway type. The medium spatial resolution of the remote sensing data is crucial for discriminating different land cover types in heterogeneous urban areas. Given the heterogeneity of land cover over Denpasar, ALOS/AVNIR-2 detects a smaller maximum value of GPP than Aster, but the annual mean GPP from ALOS/AVNIR-2 is higher than that from Aster. Based on comparisons with previous work, we find that ALOS/AVNIR-2 and Aster satellite data provided more accurate estimates of maximum GPP in Denpasar and in the tropical Kalimantan-Indonesia and Amazon forest than estimates derived from the MODIS GPP product (MOD17).

Evaluating evapotranspiration and water-use efficiency of terrestrial ecosystems in the conterminous United States using MODIS and AmeriFlux data

In this study, we used the remotely-sensed data from the Moderate Resolution Imaging Spectrometer (MODIS), meteorological and eddy flux data and an artificial neural networks (ANNs) technique to develop a daily evapotranspiration (ET) product for the period of 2004–2005 for the conterminous U.S. We then estimated and analyzed the regional water-use efficiency (WUE) based on the developed ET and MODIS gross primary production (GPP) for the region. We first trained the ANNs to predict evapotranspiration fraction (EF) based on the data at 28 AmeriFlux sites between 2003 and 2005. Five remotely-sensed variables including land surface temperature (LST), normalized difference vegetation index (NDVI), normalized difference water index (NDWI), leaf area index (LAI) and photosynthetically active radiation (PAR) and ground-measured air temperature and wind velocity were used. The daily ET was calculated by multiplying net radiation flux derived from remote sensing products with EF. We then evaluated the model performance by comparing modeled ET with the data at 24 AmeriFlux sites in 2006. We found that the ANNs predicted daily ET well ( R 2 = 0.52–0.86). The ANNs were applied to predict the spatial and temporal distributions of daily ET for the conterminous U.S. in 2004 and 2005. The ecosystem WUE for the conterminous U.S. from 2004 to 2005 was calculated using MODIS GPP products (MOD17) and the estimated ET. We found that all ecosystems' WUE-drought relationships showed a two-stage pattern. Specifically, WUE increased when the intensity of drought was moderate; WUE tended to decrease under severe drought. These findings are consistent with the observations that WUE does not monotonously increase in response to water stress. Our study suggests a new water-use efficiency mechanism should be considered in ecosystem modeling. In addition, this study provides a high spatial and temporal resolution ET dataset, an important product for climate change and hydrological cycling studies for the MODIS era.

Scaling and assessment of GPP from MODIS using a combination of airborne lidar and eddy covariance measurements over jack pine forests

Understanding the influence of within-pixel land cover heterogeneity is essential for the extrapolation of measured and modeled CO 2 fluxes from the canopy to regional scales using remote sensing. Airborne light detection and ranging (lidar) was used to estimate spatial and temporal variations of gross primary production (GPP) across a jack pine chronosequence of four sites in Saskatchewan, Canada for comparison with the Moderate Resolution Imaging Spectroradiometer (MODIS) GPP product. This study utilizes high resolution canopy structural information obtained from airborne lidar to bridge gaps in spatial representation between plot, eddy covariance (EC), and MODIS estimates of vegetation GPP. First we investigate linkages between canopy structure obtained from measurements and light response curves at a jack pine chronosequence during the growing season of 2004. Second, we use the measured canopy height and foliage cover inputs to create a structure-based GPP model (GPP Landsberg) which was tested in 2005. The GPP model is then run using lidar data (GPP Lidar) and compared with eight-day cumulative MODIS GPP (GPP MODIS) and EC observations (GPP EC). Finally, we apply the lidar GPP model at spatial resolutions of 1 m to 1000 m to examine the influence of within-pixel heterogeneity and scaling (or pixel aggregation) on GPP Lidar. When compared over eight-day cumulative periods throughout the 2005 growing season, the standard deviation of differences between GPP lidar and GPP MODIS were less than differences between either of them and GPP EC at all sites. As might be expected, the differences between pixel aggregated GPP estimates are most pronounced at sites with the highest levels of spatial canopy heterogeneity. The results of this study demonstrate one method for using lidar to scale between eddy covariance flux towers and coarse resolution remote sensing pixels using a structure-based Landsberg light curve model.

Evaluating drought effect on MODIS Gross Primary Production (GPP) with an eco‐hydrological model in the mountainous forest, East Asia

AbstractSurface soil moisture dynamics is a key link between climate fluctuation and vegetation dynamics in space and time. In East Asia, precipitation is concentrated in the short monsoon season, which reduces plants water availability in the dry season. Furthermore, most forests are located in mountainous areas because of high demand for agricultural land, which results in increased lateral water flux and uneven distribution of plant available water. These climatic and topographic features of the forests make them more vulnerable to drought conditions. In this study, the eco‐hydrological model (Regional Hydro‐Ecological Simulation System) is validated with various water and carbon flux measurements in a small catchment in Korea. The model is then extended to the regional scale with fine‐resolution remote sensing data to evaluate the Moderate Resolution Imaging Radiometer (MODIS) leaf area index and gross primary productivity (GPP) products. Long‐term model runs simulated severe drought effect in 2001 well, which is clearly shown in the ring increment data. However, MODIS GPP does not capture this drought effect in 2001, which might be from a simplified treatment of water stress in the MODIS GPP algorithm. This study shows that the MODIS GPP products can potentially overestimate carbon uptake specifically during drought conditions driven by soil water stress.

PREDICTING TREE DIVERSITY ACROSS THE UNITED STATES AS A FUNCTION OF MODELED GROSS PRIMARY PRODUCTION

At the regional and continental scale, ecologists have theorized that spatial variation in biodiversity can be interpreted as a response to differences in climate. To test this theory we assumed that ecological constraints associated with current climatic conditions (2000-2004) might best be correlated with tree richness if expressed through satellite-derived measures of gross primary production (GPP), rather than the more commonly used, but less consistently derived, net primary production. To evaluate current patterns in tree diversity across the contiguous United States we acquired information on tree composition from the USDA Forest Service's Forest Inventory and Analysis program that represented more than 17,4000 survey plots. We selected 2693 cells of 1000 km2 within which a sufficient number of plots were available to estimate tree richness per hectare. Our estimates of forest productivity varied from simple vegetation indices indicative of the fraction of light intercepted by canopies at 16-d intervals, a product from the MODIS (Moderate Resolution Imaging Spectro-radiometer), to 8- and 10-d GPP products derived with minimal climatic data (MODIS) and SPOT-Vegetation (Systeme Pour l'Observation de la Terre), to 3-PGS (Physiological Principles Predicting Growth with Satellites), which requires both climate and soil data. Across the contiguous United States, modeled predictions of gross productivity accounted for between 51% and 77% of the recorded spatial variation in tree diversity, which ranged from 2 to 67 species per hectare. When the analyses were concentrated within nine broadly defined ecoregions, predictive relations largely disappeared. Only 3-PGS predictions fit a theorized unimodal function by being able to distinguish highly productive forests in the Pacific Northwest that support lower than expected tree diversity. Other models predicted a continuous steep rise in tree diversity with increasing productivity, and did so with generally better or nearly equal precision with fewer data requirements.

A new model of gross primary productivity for North American ecosystems based solely on the enhanced vegetation index and land surface temperature from MODIS

Many current models of ecosystem carbon exchange based on remote sensing, such as the MODIS product termed MOD17, still require considerable input from ground based meteorological measurements and look up tables based on vegetation type. Since these data are often not available at the same spatial scale as the remote sensing imagery, they can introduce substantial errors into the carbon exchange estimates. Here we present further development of a gross primary production (GPP) model based entirely on remote sensing data. In contrast to an earlier model based only on the enhanced vegetation index (EVI), this model, termed the Temperature and Greenness (TG) model, also includes the land surface temperature (LST) product from MODIS. In addition to its obvious relationship to vegetation temperature, LST was correlated with vapor pressure deficit and photosynthetically active radiation. Combination of EVI and LST in the model substantially improved the correlation between predicted and measured GPP at 11 eddy correlation flux towers in a wide range of vegetation types across North America. In many cases, the TG model provided substantially better predictions of GPP than did the MODIS GPP product. However, both models resulted in poor predictions for sparse shrub habitats where solar angle effects on remote sensing indices were large. Although it may be possible to improve the MODIS GPP product through improved parameterization, our results suggest that simpler models based entirely on remote sensing can provide equally good predictions of GPP.

Incorporation of a soil water modifier into MODIS predictions of temperate Douglas-fir gross primary productivity: Initial model development

The moderate resolution imaging spectroradiometer (MODIS) is being used to monitor gross primary production (GPP), both spatially and temporally, routinely from space. However, estimates of GPP at various flux stations indicate that the MODIS algorithm may (i) over-predict GPP at sites where limitation to growth by low-soil water content is not adequately captured by the reduction in stomatal conductance by vapor pressure and (ii) under-predict GPP in highly productive, evergreen, needle leaf forests, due to a reduced radiation-use-efficiency term. The objective of this paper is to determine if any systematic bias exists in the MODIS algorithm relative to eddy covariance (EC) estimates of GPP made over an evergreen, needle leaf temperate rainforest on Vancouver Island, Canada, which is routinely water-stressed in summer months. Results indicate that 8-day GPP as predicted by the standard MODIS algorithm, with appropriate parameters for evergreen needle leaf forest, was highly correlated to EC-measured GPP ( r 2 = 0.89, p < 0.001, S.E. = 0.9 g C m −2 day −1), however with significant bias, under predicting GPP by as much as 30%. Increasing the radiation-use-efficiency term ɛ max (g C MJ −1) from the MODIS lookup value to the maximum observed at the site resulted in a reduced bias in the predicted GPP, however estimates were 8% higher than EC measurements. To account for soil water stress on plant growth, we implemented a soil water modifier initially proposed by Leuning et al. [Leuning, R., Cleugh, H., Zegelin, S., Hughes, D., 2005. Carbon and water fluxes over a temperate Eucalyptus forest and a tropical wet/dry savanna in Australia: measurements and comparison with MODIS remote sensing estimates. Agric. For. Meteorol. 129, 151–173] that accounts for rainfall and potential evaporation in the antecedent 3 months, a surrogate for soil water availability. Results confirm that field observations of relative available soil water content in the 0–60 cm layer matched the proposed soil water modifier closely with the relationship between the modified MODIS algorithm GPP and the EC-measured GPP remaining highly significant ( r 2 = 0.91, p < 0.001, S.E. = 1.1 g C m −2 day −1) with no significant bias. Whilst broad scale implementation of such a soil water modifier into the MODIS algorithm is still limited due to lack of rainfall data, at least in the short-term, the modifier does provide an alternative for researchers and land mangers, interested in applying the MODIS GPP products over regional areas, but who may have, or are observing, over-estimated production estimates due to the lack of inclusion of soil water modification to growth.

입력자료 개선에 의한 MODIS 총일차생산성의 신뢰도 향상

현재 1 km 해상도의 Moderate Resolution Imaging Spectroradiometer(MODIS) 총일차생산성(GPP) 영상이 8일 간격으로 제공되고 있다. 본 연구에서 MODIS GPP 산출에 사용되는 입력기상(DAO)자료와 광합성유효복사흡수율(FPAR) 자료의 오차를 정량화 하였고, 이들 오차가 MODIS GPP의 불확실성에 미치는 영향을 분석하였다. 입력자료의 평가를 위해 지상기상관측소의 기상자료를 사용하였고, 구름효과 등을 저감한 FPAR의 시계열을 복원하였다. 평가 결과 입력자료의 오차는 MODIS GPP에서 17% 정도 과대평가되었다. 두 오차 중에서 기상자료의 오차가 주요 원인이었으며, FPAR의 오차는 부차적인 것으로 판명되었다. 다양한 토지피복 중에서 혼효림의 MODIS GPP 오차가 약 20%로 가장 크고, 농경지는 약 11%의 오차를 보였다. 입력자료에 의한 MODIS GPP의 오차는 GPP의 계절변화뿐만 아니라 연간 GPP 변화에도 상이한 결과를 초래하였다. 따라서 MODIS GPP에 내재한 오차는 상당하다고 판단되며, 향후 GPP 모니터링에 응용하기 위해선 상기 기술한 오차 요인들을 저감한 입력자료에 의거해 MODIS GPP를 재가공할 필요가 있다. The Moderate Resolution Imaging Spectroradiometer (MODIS) regularly provides the eight-day gross primary productivity (GPP) at 1 km resolution. In this study, we evaluated the uncertainties of MODIS GPP caused by errors associated with the Data Assimilation Office (DAO) meteorology and a biophysical variable (fraction of absorbed photosynthetically active radiation, FPAR). In order to recalculate the improved GPP estimate, we employed ground weather station data and reconstructed cloud-free FPAR. The official MODIS GPP was evaluated as +17% higher than the improved GPP. The error associated with DAO meteorology was identified as the primary and the error from the cloud-contaminated FPAR as the secondary constituent in the integrative uncertainty. Among various biome types, the highest relative error of the official MODIS GPP to the improved GPP was found in the mixed forest biome with RE of 20% and the smallest errors were shown in crop land cover at 11%. Our results indicated that the uncertainty embedded in the official MODIS GPP product was considerable, indicating that the MODIS GPP needs to be reconstructed with the improved input data of daily surface meteorology and cloud-free FPAR in order to accurately monitor vegetation productivity in Korea.

Developing a continental-scale measure of gross primary production by combining MODIS and AmeriFlux data through Support Vector Machine approach

Remote sensing is a potentially powerful technology with which to extrapolate eddy covariance-based gross primary production (GPP) to continental scales. In support of this concept, we used meteorological and flux data from the AmeriFlux network and Support Vector Machine (SVM), an inductive machine learning technique, to develop and apply a predictive GPP model for the conterminous U.S. In the following four-step process, we first trained the SVM to predict flux-based GPP from 33 AmeriFlux sites between 2000 and 2003 using three remotely-sensed variables (land surface temperature, enhanced vegetation index (EVI), and land cover) and one ground-measured variable (incident shortwave radiation). Second, we evaluated model performance by predicting GPP for 24 available AmeriFlux sites in 2004. In this independent evaluation, the SVM predicted GPP with a root mean squared error (RMSE) of 1.87 gC/m 2/day and an R 2 of 0.71. Based on annual total GPP at 15 AmeriFlux sites for which the number of 8-day averages in 2004 was no less than 67% (30 out of a possible 45), annual SVM GPP prediction error was 32.1% for non-forest ecosystems and 22.2% for forest ecosystems, while the standard Moderate Resolution Imaging Spectroradiometer GPP product (MOD17) had an error of 50.3% for non-forest ecosystems and 21.5% for forest ecosystems, suggesting that the regionally tuned SVM performed better than the standard global MOD17 GPP for non-forest ecosystems but had similar performance for forest ecosystems. The most important explanatory factor for GPP prediction was EVI, removal of which increased GPP RMSE by 0.85 gC/m 2/day in a cross-validation experiment. Third, using the SVM driven by remote sensing data including incident shortwave radiation, we predicted 2004 conterminous U.S. GPP and found that results were consistent with expected spatial and temporal patterns. Finally, as an illustration of SVM GPP for ecological applications, we estimated maximum light use efficiency ( e max), one of the most important factors for standard light use efficiency models, for the conterminous U.S. by integrating the 2004 SVM GPP with the MOD17 GPP algorithm. We found that e max varied from ∼ 0.86 gC/MJ in grasslands to ∼ 1.56 gC/MJ in deciduous forests, while MOD17 e max was 0.68 gC/MJ for grasslands and 1.16 gC/MJ for deciduous forests, suggesting that refinements of MOD17 e max may be beneficial.

Light absorption by leaf chlorophyll and maximum light use efficiency

The MODIS Land Science Team has produced a standard product of gross primary production (GPP) for the global land biosphere, using a satellite-based Production Efficiency Model. In this special issue for the validation studies, two papers have examined the accuracy, error, and uncertainty of the GPP product. Here, we present a user's perspective on accuracy assessment of the data products and discuss two large sources of uncertainty in the context of model variables and parameters: 1) light absorption by chlorophyll versus light absorption by canopy and 2) maximum light use efficiency

Assessing interannual variation in MODIS-based estimates of gross primary production

Global estimates of terrestrial gross primary production (GPP) are now operationally produced from Moderate Resolution Imaging Spectrometer (MODIS) imagery at the 1-km spatial resolution and eight-day temporal resolution. In this study, MODIS GPP products were compared with ground-based GPP estimates over multiple years at three sites-a boreal conifer forest, a temperate deciduous forest, and a desert grassland. The ground-based estimates relied on measurements at eddy covariance flux towers, fine resolution remote sensing, and modeling. The MODIS GPP showed seasonal variation that was generally consistent with the in situ observations. The sign and magnitude of year-to-year variation in the MODIS products agreed with that of the ground observations at two of the three sites. Examination of the inputs to the MODIS GPP algorithm-notably the fraction of photosynthetically active radiation (FPAR) that is absorbed by the canopy), minimum temperature scalar, and vapor pressure deficit scalar-provided explanations for cases of disagreement between the MODIS and ground-based GPP estimates. Continued evaluation of interannual variation in MODIS products and related climate variables will aid in assessing potential biospheric feedbacks to climate change

Evaluation of remote sensing based terrestrial productivity from MODIS using regional tower eddy flux network observations

The Moderate Resolution Spectroradiometer (MODIS) sensor has provided near real-time estimates of gross primary production (GPP) since March 2000. We compare four years (2000 to 2003) of satellite-based calculations of GPP with tower eddy CO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> flux-based estimates across diverse land cover types and climate regimes. We examine the potential error contributions from meteorology, leaf area index (LAI)/fPAR, and land cover. The error between annual GPP computed from NASA's Data Assimilation Office's (DAO) and tower-based meteorology is 28%, indicating that NASA's DAO global meteorology plays an important role in the accuracy of the GPP algorithm. Approximately 62% of MOD15-based estimates of LAI were within the estimates based on field optical measurements, although remaining values overestimated site values. Land cover presented the fewest errors, with most errors within the forest classes, reducing potential error. Tower-based and MODIS estimates of annual GPP compare favorably for most biomes, although MODIS GPP overestimates tower-based calculations by 20%-30%. Seasonally, summer estimates of MODIS GPP are closest to tower data, and spring estimates are the worst, most likely the result of the relatively rapid onset of leaf-out. The results of this study indicate, however, that the current MODIS GPP algorithm shows reasonable spatial patterns and temporal variability across a diverse range of biomes and climate regimes. So, while continued efforts are needed to isolate particular problems in specific biomes, we are optimistic about the general quality of these data, and continuation of the MOD17 GPP product will likely provide a key component of global terrestrial ecosystem analysis, providing continuous weekly measurements of global vegetation production