The leaf area index (LAI) serves as a significant vegetation growth indicator and plays an essential role in vegetation's feedback to climate change. Currently, artificial intelligence (e.g., deep learning) algorithms possess strong capabilities in constructing complex relationships and demonstrate successful integration with remote sensing for LAI inversion. Among these algorithms, the long short-term memory (LSTM) network excels in handling sequence data and features a multi-layer nonlinear structure that effectively captures complex nonlinear relationships between vegetation canopy reflectance and LAI. However, previous researches mainly relied on the strong learning capabilities of LSTM without incorporating essential remote sensing knowledge, which led to the lack of process information guidance in the training stage. Consequently, the performance of the trained model may be significantly limited. In this letter, we proposed a process-guided LSTM (LSTM-PG) deep learning method for LAI estimation by integrating radiative transfer models. The constrained training dataset was generated using the Soil-Leaf-Canopy (SLC) model. We separately utilized the loss function of mean squared error (MSE) and a process-guided loss function to generate LSTM models for LAI predictions from the simulated SLC datasets. Subsequently, we validated the accuracy of the LAI retrieval models using field measurements from the ImagineS project. Our results indicated that the proposed process-guided (PG) method (R² = 0.79, RMSE = 0.87) performed better than the LSTM-MSE estimations (R² = 0.79, RMSE = 0.93). Additionally, statistical analyses across various scenarios demonstrated significant advantages of the proposed method, and the LSTM-PG predictions showed good spatial consistency with the LAI reference maps.

Topography distorts the angular distribution of the canopy gap fraction (GF). Path length (PL) correction is a simple and effective method to harmonize this distortion and improve canopy reflectance modelling andin situleaf area index measurements for vegetation, including both continuous (e.g., grass and crop) and discrete (e.g., forests) canopies, over sloping terrains. The rigorously theoretical derivation of PL correction for continuous canopies has been implemented. However, for discrete canopies, the PL show a serious heterogeneity, making it nearly impossible to be calculated. In this regard, there is still a need to develop theoretical derivation to evaluate and improve PL correction for forests over sloping terrains. In this study, (1) PL correction is proven to be equivalent to the correction of the canopy GF over sloping terrains, and our strategy concerns the canopy GF as a proxy of PL. (2) PL correction is first proven to be completely valid for forests with the Poisson trees distribution; yet it may produce uncertainty in certain directions for forests with tree distribution deviating from the Poisson model, especially for forests with regular tree distribution. (3) An improved model based on a Nilson and Peterson’s GF model for correcting PL for forests is given in this study. The results show that error produced by the PL correction for some forests can be effectively decreased by the improved model. The variation of directional tree distribution parameter cB(θ) with slope is the main cause of error produced by PL correction for forests. The study is of importance for better understanding and more accurate application of PL theory in topographic corrections andin situleaf area index measurements for forest canopies over sloping terrains.

Multiple satellite data can provide rich information for time series leaf area index (LAI) inversion. Long Short-Term Memory (LSTM) has been widely applied to capture long time dependencies from sequential data and solve the problem of vanishing (or exploding) gradients. It's feasible to retrieve time series LAI based on the LSTM algorithm. In this study, an LSTM model was built based on MODIS reflectance and a fused LAI from three satellite LAI products, including GLASS, MODIS and GEOV2 LAI. Then, time series LAI was obtained from the LSTM model coupled with reflectance data. Finally, the retrieval results were validated against fine resolution LAI reference maps and compared with the three LAI products. The accuracy of the proposed LSTMfusion LAI was best (R2=0.85, RMSE=0.69) among all the LAI estimations. In addition, the comparison among the temporal profiles of LSTMfusion LAI and the three global LAI products illustrated that the proposed model was able to efficiently generate time series LAI, which was more continuous and smoother than MODIS and VIIRS LAI.

Time series leaf area index (LAI) is essential to studying vegetation dynamics and climate changes. The LAI at current status can be regarded as the accumulative consequence of the counterpart at prior times. Although the deep learning algorithm - Long short-term memory (LSTM) can capture long-time dependencies from sequential satellite data for time series LAI estimation, it only uses the information at prior statuses, and neglects the backward propagation of current vegetation change information. Thus, the LSTM-based LAI quality might be limited. In this letter, the bidirectional LSTM (Bi-LSTM) approach was proposed to integrate the information of multiple satellite products from both the past and future for temporal LAI retrieval. The fused values from GLASS, MODIS, and VIIRS LAI products, as well as MODIS reflectance in 2014-2015, serve as the output response and input for the Bi-LSTM training. Then, we compared the Bi-LSTM predictions with the counterparts from the LSTM, the fused LAI and three products using independent validation datasets in 2016. Results illustrated that our proposed Bi-LSTM method achieved better performance with higher accuracy (R2=0.84, RMSE=0.76) when compared to the LSTM estimation (R2=0.83, RMSE=0.82) and LAI products (R2<0.68, RMSE>1). Furthermore, our proposed method provided smoother and more continuous temporal profiles of LAI than other retrieval approaches.

Fine spatial details of vegetation growth are usually lost in leaf area index (LAI) products obtained from coarse spatial resolution satellite sensors. This may bring uncertainties in ecosystem process models, which usually require LAI products with fine spatiotemporal resolutions. Successful downscaling of LAI dynamics to fine spatial resolution is very important for meeting the demands of these models. Hence, a multiscale multisensor approach using the ensemble Kalman smoother (EnKS) technique is proposed in this paper. The LAI dynamics at a coarser spatial resolution are incorporated as prior information into the remotely sensed observations for time series LAI estimation at a finer spatial resolution. Downscaled LAI dynamics are evaluated based on spatial distribution and temporal trajectory. The results indicate the assimilated LAI to be in good agreement with the reference values at the different spatial scales. For example, the coefficient of determination (R2) between the reference values and fine-resolution LAI results retrieved by the proposed approach is 0.71 with a root-mean-square-error (RMSE) value of 0.65 on Julian day 185 at the Agro site. The method has proved to be effective for downscaling LAI dynamics, which improves the spatiotemporal patterns of fine-resolution LAI retrievals with respect to earlier methods.

Ecosystems is survived in the suitable environment which provide appropriately abiotic resources for organism and IUCN have applied abiotic degradation as Criterion C to assess the risk of ecosystems. However, the origin and collapse status of the criterion is vague for assessors, and the results are inconsistent as the response of ecosystems to environment are different. Therefore, the ecological amplitude of ecosystem to environment stress was introduced in the ecosystems risk assessment with the relationship between criterion and ecological amplitude. To put this concept into practice, remote sensing was applied to acquire the optimum and tolerance of each ecosystem. The natural forest in xishuangbanna was assessed by this proposed method to identify the stress of temperature. The result show that the status of natural forest in the past is least concern (LC), and vulnerable (VU) in the feature. With temperature in the future significantly increasing, natural forest may be suffered with heat stress. The proposed method describe the risk derived from degradation of environment in mechanism greatly improved the consistency and feasibility of Criterion C in the ecosystems risk assessment.

Sloping terrain induces distortion of canopy reflectance (CR), and the retrieval of biophysical variables from remote sensing data needs to account for topographic effects. We developed a 1-D model (the path length correction (PLC)based model) for simulating CR over sloping terrain. The effects of sloping terrain on single-order and diffuse scatterings are accounted for by PLC and modification of the fraction of incoming diffuse irradiance, respectively. The PLC model was validated via both Monte Carlo and remote sensing image simulations. The comparison with the Monte Carlo simulation revealed that the PLC model can capture the pattern of slopeinduced reflectance distortion with high accuracy (red band: R2= 0.88; root-mean-square error (RMSE) = 0.0045; relative RMSE (RRMSE) = 15%; near infrared response (NIR) band: R2= 0.79; RMSE = 0.041; RRMSE = 16%). The comparison of the PLC-simulated results with remote sensing observations acquired by the Landsat8-OLI sensor revealed an accuracy similar to that with the Monte Carlo simulation (red band: R2= 0.83; RMSE = 0.0053; RRMSE = 13%; NIR band: R2= 0.77; RMSE = 0.023; RRMSE = 8%). To further validate the PLC model, we used it to implement topographic normalization; the results showed a large reduction in topographic effects after normalization, which implied that the PLC model captures reflectance variations caused by terrain. The PLC model provides a promising tool to improve the simulation of CR and the retrieval of biophysical variables over mountainous regions.

Surface soil moisture (SSM) is an important parameter at the land-atmosphere interface. In past decades, passive microwave remote sensing offers a good opportunity for obtaining SSM on a global scale, and many downscaling methods have been proposed using the triangle-based empirical soil moisture relationship models to overcome the limitation of coarse spatial resolution of its SSM products for regional applications. This paper aimed to examine and compare the effectiveness of five typical triangle-based empirical soil moisture relationship models for estimating SSM with Landsat-5 data and in situ measurements from the Maqu network on the northeastern part of the Tibetan Plateau for nine cloud-free days. The results showed that the model that treats the SSM as a second-order polynomial with land surface temperature, vegetation indices (VIs), and surface albedo as inputs exhibited the best performance compared with the results of other models. The VI comparison indicated that the use of the normalized difference VI or the fractional vegetation cover in this model outperformed other VIs, with the root-mean-square deviation of approximately 0.055 m3/m3and the coefficient of determination ($\text{R}^{2}$ ) above 0.78 at the nine-day average level. In addition, a significant spatial scale effect of the model was also found through analyzing the model fitting results at different window sizes. The study provides important insight into the best empirical relationship models for capturing soil moisture dynamics. These models can support the passive microwave soil moisture data spatial downscaling and validation applications in future studies.

How to accurately detect cloud and snow in the remote sensing imagery is an open problem for the remote sensing application. For only visible and near infrared band in HJ-1A/B CCD images, the cloud detection algorithm using the shortwave infrared and thermal infrared band is restricted by the band-lacking problem. Based on the multi-temporal information of the HJ-1A/B CCD images, a new algorithm is proposed in this paper. Using available images in one month, a cloud-free reference image was firstly composed. Then, the cloud and snow pixel are separated through the difference of the blue band between the reference and each date. Subsequently, the regional covariance matrix is computed further to eliminate the non-cloud pixels. The test result shows that, the overall accuracy is about 85.96% to 93%. It indicates that the proposed method can integrate the temporal and texture information to improve detection accuracy for the cloud and snow.

The leaf area index is one of key parameters for ecosystem monitoring, global carbon circulation and climate change. At present leaf area index is routinely available from Earth Observation (EO) instruments such as MODIS. However MODIS-derived estimates of LAI require validation before they are utilised by the ecosystem modelling community. The paper presents a validation of the MODIS LAI collection 5 product over forested terrain in Xishuangbanna, southwest China, based on field measurements which are upscaled using high resolution HJ image. Results suggest that the MODIS LAI product has an accuracy with R2of 0.35 and RMSE of 0.61 m2/m2, and overestimates values in the high LAI broadleaf forest (where LAI>3.5 m2/m2). Further validation efforts must be carried out over the study area for an assessment of the MODIS LAI in order to improve its accuracy.

Based on the HANTS and dynamic threshold method, the spatio-temporal changes of the alpine grassland phenology in the Zoigê plateau was analyzed by using MODIS EVI data from 2001 to 2013. The results were found as follows: (1) The spatial distribution of the average vegetation phenology from 2001 to 2013 is closely related to the water and heat conditions. Accompanying the deterioration in heat and water conditions from low altitude to high altitude and south to north, SOG(start of growing season) was delayed gradually, EOG(end of growing season) advanced slowly, and LOG(long of growing season) shortened gradually. Water played an important role in the regional differentiation of phenology (2) From 2001 to 2013, SOG came earlier by 0.6d/a, EOG was late by 0.2d/a, and LOG lengthened by 0.8d/a. The inter-annual phenology changes of the vegetation exhibited significant differences at different elevations and water condition. (3) Heat and moisture is the main ecological factor influencing the growth of plant. Temperature responses of phenology significantly became stronger with increasing cumulative preseason precipitation.

The leaf area index is one of key parameters for ecosystem monitoring, global carbon circulation and climate change. The remote sensing data from different satellites have become the primary data source for estimating leaf area index from regional to global scale. In this study, we assess the potential use of Landsat TM, HJ-1 CCD, and SPOT HRVIR sensors for leaf area index estimation in a mountainous area. Results suggest that three sensors behave similarly for LAI inversion over complicated terrain. The maximum correlation coefficients are in the order of broadleaf forest (0.82) > shrub/grass (0.78) > all plots (0.61) > needleleaf forest (0.53), which are derived from the field LAI-SPOT SWVI, TM SR, SPOT SWVI, and SPOT RSR relationships for all plots, needleleaf forest, broadleaf forest, and shrub/grass, respectively.

The new data assimilation algorithm is developed to estimate LAI from time-series MODIS reflectance data (MOD09A1). The canopy radiative transfer model (ACRM) is coupled with an empirical LAI dynamic model, and the ensemble Kalman smoother (EnKS) is used to estimate the parameters of the coupled model from MOD09A1 data. The preliminary analysis using MODIS surface reflectance data at some AmeriFlux network sites was performed to validate this method. The results show that the algorithm is helpful to produce the temporally continuous LAI estimation of cropland efficiently. By comparing with the field measured LAI, the retrieved LAI has been significantly improved and shown more smooth in time series than the MODIS LAI product.