Niu Zheng

The carbon flux estimation of China terrestrial ecosystems based on NOAA/AVHRR data

Net ecosystem productive (NEP) is defined as the net carbon dioxide flux to or from an ecosystem without natural or human disturbance, and integrates all ecosystems carbon sources and sinks: NEE=GPP-Ra-Rh. Here NPP was calculated with the light use efficiency model based on NOAA/AVHRR, and the soil respiration of the corresponding ecosystem came from references. On the whole, the NEP shows the terrestrial ecosystem in China is a carbon sink though there are great uncertainties. The geographic gradient of the NEP clearly shows more correlation with temperature on latitude gradient, and with precipitation on longitude. Three main sources of uncertainties were analyzed: (1) land cover classification based on remote sensing; (2) NPP modelling; (3) soil respiration modelling. The observation and modelling integrated with remote sensing will be a very important solution to the carbon source/sink at large spatial sale.

Retrieval forest stock volume of large plantation in South China using RADARSAT-SAR

Estimation of forest stem volume in large area using remotely sensed data has remains difficult due to poor understanding relationships of forest stand parameters and remote-sensing spectral response. Recent studies have show that SAR can be used to investigate forest attribute, especially in tropic forest where there is so much cloud that good quality optic remote sensing is not easily available. The goal of this research was to assess the efficiency of SAR data for forest plantation management and the sensitivity of stem volume to RADARSAT-SAR data. Based on field survey data of the large managed plantation and GIS, the ability of RADARSAT-SAR data in extracting forest stock volume was analyzed in this paper. Results show that RADARSAT-SAR Backscatter coefficient has a relatively good logarithmic correlation with forest stock volume, tree height and diameter at breast height for the six species. The investigation of response of tree height and diameter at breast height to backscatter also shows that backscatter coefficient is affected by tree height much more than by diameter at breast height for Eucalypt, but for pine trees, backscatter coefficient is affected by diameter at breast height much more than by tree height. Inversion results of stem volume show that stem volume can be estimated for such plantation but had limitations. The estimation results obtained are not accurate enough for forest management purposes at the forest stand level. However, the approximate volume estimates derived by the method can be useful in areas where no other forest information is available.

Application of neural network in retrieving chlorophyll a concentration from SeaWiFS in the sea area near Dalian

Chlorophyll a concentration is very important for assessment of marine environmental pollution and estimation of oceanic primary productivity. Only remote sensing can provide continuous, real-time and large-scale chlorophyll a concentration information in oceans. Traditional empirical chlorophyll algorithms don’t perform well especially in coastal waters partly for the nonlinear nature of transfer function. However, the neural networks (NN) are able to model any nonlinear function flexibly. On basis of SeaWiFS imagery and in situ chlorophyll a concentration measured in the sea area near Dalian in China, Neural Network(NN) with the architecture of 2 inputs, 6 nodes in a single hidden layer and 1 output was employed to retrieve chlorophyll a concentration. Regression analysis was also conducted. The R of NN and that of regression equation were 0.976, 0.572 respectively; RMSE (root mean square error) were 0.112, 0.387 respectively. Distribution map of chlorophyll a retrieved from NN is consistent with previous research. NN is more accurate better than regression method for retrieval of chlorophyll a concentration and it is a promising method for research of oceanic constituents in Case 2 waters. Keywords-Neural Network; chlorophyll a concentration, SeaWiFS; retrieval

Polynomial analysis of canopy spectra and biochemical component content inversion

A polynomial expression model was developed in this paper to describe directional canopy spectra, and the decomposition of the polynomial expression was used as a tool for retrieving biochemical component content from canopy multi-angle spectra. First, the basic formula of the polynomial expression was introduced and the physical meaning of its terms and coefficients was discussed. Based on this analysis, a complete polynomial expression model and its decomposition method were given. By decomposing the canopy spectra simulated with SAILH model, it shows that the polynomial expression can not only fit well the canopy spectra, but also show the contribution of every order scattering to the whole reflectance. Taking the first scattering coefficients a10 and a01 for example, the test results show that the polynomial coefficients reflect very well the hot spot phenomenon and the effects of viewing angles, LAI and leaf inclination angle on canopy spectra. By coupling the polynomial expression with leaf model PROSPECT, a canopy biochemical component content inversion model was given. In the simulated test, the canopy multi-angle spectra were simulated by two different models, SAILH and 4-SCALE respectively, then the biochemical component content was retrieved by inverting the coupled polynomial expression + PROSPECT model. Results of the simulated test are promising, and when applying the algorithm to measured corn canopy multi-angle spectra, we also get relatively accurate chlorophyll content. It shows that the polynomial analysis provides a new method to get biochemical component content independent of any specific canopy model.

Retrieving the crop coefficient spatial distribution for cotton under different growth status with Landsat ETM+ image

Crop water requirement was important in the irrigation scheduling. The crop coefficient is a parameter for estimating crop water requirement by multiplying with the reference crop evapotranspiration. Crop coefficient used to be approximated by crop developing days. The method must have some defect because the crop coefficient is a parameter related to crop status, climate condition and surface albedo. All the factors relating to the crop coefficient are spatially diverse and remote sensing has advantages in obtaining the distributing parameters for vegetation and climate factor. Based on the Penman-Monteith equation, the reference crop evapotranspiration and potential evapotranspiration for cotton under different growth status was estimated with measured meteorological data, then the crop coefficient for cotton was retrieved from a Landsat ETM+ image. And the sensitivity of crop coefficient to the influence factors were analysed. The results showed that the crop coefficient retrieved from the ETM+ image was greater than those suggested by FAO and the crop coefficient was influenced and decided by NDVI that represents crop growth status, while surface albedo that has a very larger variance for the sparse vegetation cover has scarcely any effect on crop coefficient and the climate factors has litter influence on crop coefficient too; with the vegetation cover fraction developing, the climate factor has a much more positive effect on the crop coefficient.