Gaolong Zhu

GOST: A Geometric-Optical Model for Sloping Terrains

GOST is a geometric-optical (GO) model for sloping terrains developed in this study based on the four-scale GO model, which simulates the bidirectional reflectance distribution function (BRDF) of forest canopies on flat surfaces. The four-scale GO model considers four scales of canopy architecture: tree groups, tree crowns, branches, and shoots. In order to make this model suitable for sloping terrains, the mathematical description for the projection of tree crowns on the ground has been modified to consider the fact that trees grow vertically rather than perpendicularly to sloping grounds. The simulated canopy gap fraction and the area ratios of the four scene components (sunlit foliage, sunlit background, shaded foliage, and shaded background) by GOST compare well with those simulated by 3-D virtual canopy computer modeling techniques for a hypothetical forest. GOST simulations show that the differences in area ratios of the four scene components between flat and sloping terrains can reach up to 50%-60% in the principal plane and about 30% in the perpendicular plane. Two case studies are conducted to compare modeled canopy reflectance with observations. One comparison is made against Landsat-5 Thematic Mapper (TM) reflectance, demonstrating the ability of GOST to model canopy reflectance variations with slope and aspect of the terrain. Another comparison is made against MODIS surface reflectance, showing that GOST with topographic consideration outperforms that without topographic consideration. These comparisons confirm the ability of GOST to model canopy reflectance on sloping terrains over a large range of view angles.

Foliage Clumping Index Over China's Landmass Retrieved From the MODIS BRDF Parameters Product

The three-dimentional plant canopy architecture is often characterized using the foliage clumping index useful for ecological and land surface modeling. In this paper, an algorithm is developed to retrieve the foliage clumping index with the Moderate Resolution Imaging Spectroradiometer bidirectional reflectance distribution function (BRDF) parameter product (MCD43A1), which is generated using the RossThick-LiSparse Reciprocal (Ross-Li) model. First, the Ross-Li model is modified to improve the simulation of the reflectance at hotspot using the Polarization and Directionality of Earth Reflectance measurements as benchmarks to determine BRDF parameters. Then, the modified model (Ross-Li-H) is used to simulate the reflectance at hotspot and darkspot, which is used to calculate the normalized difference between hotspot and darkspot (NDHD). With the relationship between clumping index and NDHD simulated by the 4-Scale geometrical model, the clumping index over Chinas landmass at 500-m resolution is retrieved every 8 days during the period from 2003 to 2008. Finally, The effect of topography on the retrieved clumping index is corrected using a topographic compensation function calculated from the digital elevation model at 90-m resolution. The topographically corrected clumping index values correlate well with field measurements at five sites over China, indicating the feasibility of the algorithm for retrieving the clumping index from the MCD43A1 product.

A novel moisture adjusted vegetation index (MAVI) to reduce background reflectance and topographical effects on LAI retrieval.

A new moisture adjusted vegetation index (MAVI) is proposed using the red, near infrared, and shortwave infrared (SWIR) reflectance in band-ratio form in this paper. The effectiveness of MAVI in retrieving leaf area index (LAI) is investigated using Landsat-5 data and field LAI measurements in two forest and two grassland areas. The ability of MAVI to retrieve forest LAI under different background conditions is further evaluated using canopy reflectance of Jack Pine and Black Spruce forests simulated by the 4-Scale model. Compared with several commonly used two-band vegetation index, such as normalized difference vegetation index, soil adjusted vegetation index, modified soil adjusted vegetation index, optimized soil adjusted vegetation index, MAVI is a better predictor of LAI, on average, which can explain 70% of variations of LAI in the four study areas. Similar to other SWIR-related three-band vegetation index, such as modified normalized difference vegetation index (MNDVI) and reduced simple ratio (RSR), MAVI is able to reduce the background reflectance effects on forest canopy LAI retrieval. MAVI is more suitable for retrieving LAI than RSR and MNDVI, because it avoids the difficulty in properly determining the maximum and minimum SWIR values required in RSR and MNDVI, which improves the robustness of MAVI in retrieving LAI of different land cover types. Moreover, MAVI is expressed as ratios between different spectral bands, greatly reducing the noise caused by topographical variations, which makes it more suitable for applications in mountainous area.

Validation of MODIS gross primary productivity for a subtropical coniferous plantation in Southern China

Terrestrial carbon cycle plays an important role in global climate change. As a key component of terrestrial carbon cycle, gross primary productivity (GPP) is a major determinant of the exchange of carbon between the atmosphere and terrestrial ecosystems. 8-day global GPP estimated from ground meteorological data and remotely sensed fraction of photosynthetic active radiation (fPAR) by MODIS using the light use efficiency approach is currently provided as MOD17 product. Previous studies indicated that MODIS GPP has large uncertainties in some ecosystems. In this study, GPP of a subtropical coniferous plantation at Qianyanzhou Experimental Station in southern China was firstly calculated using the MODIS GPP algorithm (MOD17 algorithm) driven by MODIS fPAR and measured meteorological data. Calculated GPP was validated using GPP measured during 2003 and 2004 with the eddy covariance technique. Then the potential to better MODIS GPP was investigated through comparing GPP calculated using the MOD17 algorithm and improved fPAR or/and maximum light use efficiency (εmax) calibrated with measured GPP. The results indicated that the MODIS GPP product significantly underestimated measured GPP at this planted forest. The R2 of MODIS GPP with measured GPP was 0.72 and 0.67 in 2003 and 2004, respectively. And the calculated annual GPP was 33% and 47% lower than measured values in these two years. The improvement on fPAR through using LAI data estimated with photosynthetic active radiation (PAR) measured above and below canopy can definitely remedy underestimation of annual GPP. The application of εmax determined through model calibration improved annual GPP more significantly, indicating that the errors in MODIS GPP at this site can be mainly attributed to the underestimation of fPAR and εmax. When the improved fPAR and εmax were used, the agreement between calculated and measured 8-day GPP improved significantly, with R2 equals to 0.78 and 0.85 for years 2003 and 2004, respectively. And the calculated annual GPP was only 3.5% lower and 1.3% higher than measured values in these two years. Through this study, it can be concluded that accurate εmax and LAI from which fPAR is calculated are required for reliably calculating regional/global GPP with the MOD17 algorithm. The fusion of flux data with remote sensing data can provide the accurate estimate of εmax andhas a great potential to control uncertainties in calculated regional/global GPP.

Evaluation on wetland classification in Yancheng natural reserve, China using HJ-1 data

Remote sensing of wetlands is one of important aspect in application and research of remote sensing. This paper studied the HJ-1 CCD data (30m/pixel) applied to wetland resource analysis of the Yancheng national natural reserve. And we gave evaluation on ability of the HJ-1 CCD data in wetland classification. Firstly, we gave the spectral analysis of the typical wetland types. The extraction experiments were carried out by supervised and unsupervised classification methods. Finally, we compared the classification results with those gained from Landsat-5 TM data (June 2007). It indicates that the HJ-1 CCD data have good recognition performance.

The comparison of different methods to measure leaf area index of forests in Maoershan Mountain, Northeastern China

Different optical instruments are currently available for measuring LAI such as LAI 2000 Plant Canopy Analyser (LAI-2000), Tracing Radiation and Architecture of Canopies (TRAC) and Digital Hemispherical Photography (DHP). Their applicability varies in different ecosystems. This study was devoted to compare LAI measured using four different methods (LAI measured by DHP, LAI measured by TRAC, LAI calculated using effective LAI measured by LAI-2000 and clumping index measured by DHP, and LAI calculated using effective LAI measured by LAI-2000 and clumping index measured by TRAC) in the Maoershan experimental forest farm of Northeast Forestry University located in Shangzhi city of Heilongjiang province. Methods used to measure LAI have considerable effects on observed LAI. The means of LAI measured by four different methods are 3.15, 4.73, 3.97, and 4.24 and corresponding standard deviations are 1.54, 2.39, 1.82, and 1.75, respectively. According to previous studies, the combination of LAI-2000 with TRAC can give the most reliable measurements of LAI. Therefore, DHP tends to underestimate LAI at this area, especially for dense canopies while TRAC tends to overestimate slightly LAI for dense canopies. The fitting of LAI measured using four different methods with normalized difference vegetation index (NDVI) and reduced simple ratio (RSR) calculated from TM data acquired on June 24, 2009 indicated that RSR is a better predictor of LAI than NDVI in this study area. The agreements between measured and estimated LAI using the best fitted models are 58%, 70%, 57% and 68% for these four different methods. Corresponding root mean square errors (RMSE) are 0.80, 0.85, 0.88, and 0.75, respectively. The regional means of LAI retrieved using the empirical models established on the basis of RSR and LAI measured with four different methods are 3.47, 5.26, 4.31, and 4.68, respectively, indicating that if DHP is used as a surrogate of TRAC and LAI-2000, LAI might be underestimated by about 25.7% in this area.

Decrease of net primary productivity in China's terrestrial ecosystems caused by severe droughts in 2009

In 2009, terrestrial ecosystems in China were hit by a series of droughts in different seasons. However, the degree at which net primary productivity (NPP) of terrestrial ecosystems was affected in China is not clear yet. In this study, the remote sensing driven process-based Boreal Ecosystem Productivity Simulator (BEPS) model was used to estimate NPP decrease in Chinas terrestrial ecosystems caused by the abnormal droughts in 2009. The results show that the BEPS model is able to estimate gross primary productivity (GPP) and NPP of Chinas terrestrial ecosystems reliably. Estimated GPP and NPP show similar spatial patterns, decreasing from east to west and from south to north. In 2009, annual NPP was lower than the averages over the period from 2000 to 2010 in most regions of China, especially in areas of southern China. The decrease of annual NPP in 2009 over southeast Tibet and southeast coastal areas was even more than 100 g C m−2 yr−1. The annual total NPP of Hunan and Yunnan provinces, Guangxi and Tibet autonomous regions in 2009 decreased by 4% to 6% of multi-year means, owing to the impact of consecutive drought from summer to winter in these areas. The national total of NPP in this year decreased about 35.5 Tg C yr−1, approximately equivalent to 1% of annual total NPP in Chinas terrestrial ecosystems averaged during the period from 2000 to 2010.

Spatial distribution of soil erosion in a black soil region of Northeast China studied using remote sensing and GIS techniques

The soil erosion intensity and its spatial distribution of Mashezihe watershed located at Bin county in the typical gentle hilly black soil region of Northeast China were studied using the Surface Erosion Grading Indicator (SEGI) method and Universal Soil Loss Equation (USLE) model. Land cover types and vegetation cover fraction derived from the Advanced Land Observing Satellite (ALOS) remotely sensed data were employed in conjunction with other spatial datasets (DEM and soil texture) for assessing soil erosion in the ArcGIS platform. Results show that with the support of remotely sensed parameters by ALOS, the USLE model is able to identify more detailed information on soil erosion than the SEGI method. The total areas of very slight and slight erosion identified by the two methods are very similar. However, the USLE model produced larger areas moderately, highly, and very highly eroded than the SEGI method. It also classified severely eroded areas of 52.25hm2. According to grading regulation of soil loss tolerance issued by Ministry of Water Resources of China, the areas of very slight erosion (namely no obvious erosion phenomenon) and slight erosion identified by the USLE model are 21298.38 hm2 and 22919.19 hm2, accounting for 45.49% and 48.95% of the entire watershed, respectively, mainly in areas with slopes smaller than 5° and elevations lower than 200 m. Areas moderately, highly and very highly eroded are 1776.75, 486.56 and 286.88 hm2, equivalent to 3.79%, 1.04% and 0.61% of the study area, mainly in areas with slopes ranging from 8° and 15° and elevations ranging from 200 to 300 m. The severely eroded areas are sporadically distributed in areas with slopes above 15° and elevations higher than 300 m.

Comparison of forest Leaf Area Index retrieval based on simple ratio and reduced simple ratio

Leaf Area Index (LAI) is an essential parameter for process-based ecological and climate models. Spectral vegetation indices calculated from remote sensing data are widely used for LAI retrieval at large scales. The applicability of two vegetation indices, namely Simple Ratio (SR) and Reduced Simple Ration (RSR), for retrieving LAI at Maoershan mountain in Heilongjiang province of China was investigated through analyzing the correlations of SR and RSR calculated from Landsat-5 TM data acquired on 24 June, 2009 with LAI measured at 23 typical plots with Li-Cor LAI-2000 during 12 to 20 July, 2009. The fitted model with SR as the predator captured 54.7% of variations of LAI among these 23 plots while the variations of LAI explained by the fitted model with RSR as the predictor increased to 75.4%, indicating the better performance of RSR over SR in retrieving forest LAI in the study area owing to the ability of RSR to reduce the influence of soil background reflectance through the incorporation of the reflectance of shortwave infrared wavelength into SR. LAI derived from models based on SR and RSR was well correlated (R2=0.7219, N=180932). The mean of LAI estimated using the SR-based model was 0.37 larger than that estimated using the RSR-based model for the entire study area. LAI estimated using the former model was smaller than that estimated by the latter model when LAI estimated by the latter was larger than 5.6, indicating that RSR saturates slower than SR under the condition of high LAI. However, RSR is more sensitive to the influence of topography and shadows of clouds than SR.