Conference Agenda

Observation and modelling of the coupled energy and water balance is the key to understand hydrospheric and cryospheric processes at high elevation. In the Qinghai – Tibet Plateau (QTP) in – situ observations of liquid and solid precipitation are very sparse and studies on the mass balance of glaciers and the water balance of catchments are hampered by this gap. We are exploring the potential of using model forecasts of precipitation at high spatial resolution to replace or complement in-situ observations. A first experiment on applying WRF to model an extensive snowfall event on the entire QTP was completed and the results are very encouraging. In this study in – situ observations of air temperature, snow depth and snow water equivalent were used to evaluate model performance and particularly alternate model configurations. Our experiments did show that the WRF configurations with advanced land surface physics schemes captured better the spatial distribution of the snow event, but overestimated the intensity and extent of SD and SWE. Next, we focused on areas at lower elevation to carry out experiments with a coupled energy and water balance model of a catchment using again model output on precipitation.

A second set of experiments with WRF targeted the evaluation of model precipitation and other at-surface fields, e.g. air temperature and wind speed, for individual glaciers. This approach can potentially overcome a major challenge in energy and mass balance of glaciers, i.e. the lack of spatially distributed forcing data at high spatial resolution.

The energy and mass balance of glaciers was also analysed using a suite of earth observation data. The trend in glacier thickness at very high spatial resolution was determined for several glaciers using multi – temporal DEM-s generated with ZY – 3 stereo-image data. This study determined changes in glacier surface elevation separately for the accumulation and ablation zone. For the same glaciers, accurate surface velocity fields were retrieved by staking L-TM, L8-OLI and S2-MSI images.

Quantitative information on water losses is important to understanding the global terrestrial water cycle and land – atmosphere interactions. However, land surface water loss (evapotranspiraiton, ET) estiamted by land surface models usually neglects the sub-grid heterogeneity of landatmosphere parameters, and it will cause aggregation biases in spatially-averaged ET estimates, considering the nonlinear dependences of ET on the heterogeneous land-atmosphere parameters. One frequently adopted strategy clusters the heterogeneous surface within a model grid into several tiles, assumed to be homogeneous, usually based on high-resolution land cover data. While the differences in bulk-averaged parameters between different tiles are considered, the heterogeneity within each tile is neglected. To evaluatethe aggregation bias, a numerical analysis was conducted to compare the aggregation bias was calculated by comparing ET estimates based on bulk-averaged SSM and LAI with the one obtained by aggregation of the flux estimates based on the Probability Distribution Function (PDF), which complies with energy conservation. Four types of PDF were used to simulate different scenarios on the heterogeneity (within a tile) of SSM and LAI, i.e., from water scarcity to wet, and from sparse to dense vegetation covered surfaces.Overall, the numerical experiments indicate that impacts on tile ET related to LAI are smaller than the ones related to SSM. Different meteorological conditions combined with the nonlinear dependence of ET on SSM/LAI may lead to large changes in the aggregation bias, even from underestimates to overestimates or conversely. In climate conditions with larger atmospheric water demand, enhancing evaporation, underestimation is more likely, and vice versa. Neglecting the actual spatial variability of both SSM and LAI within tiles can lead to both large relative error (> 20%) and absolute error (> 1 mm/day) in the estimated ET in semi-arid areas. A negative bias is expected at low ET / ET0 and a positive bias is expected at large ET / ET0, regardless of climate conditions (i.e., ET0). The relation between aggregation bias and meteorology found in this study has the potential to identify or even as a starting point to correct the possible serious underestimations and overestimations in applications.

Meanwhile, to achieve a better water loss products, the ETMonitor algorithm was further improved following the former study of last year, to take advantage of thermal remote sensing. In the improved scheme, the evaporation fraction was first obtained by land surface temperature - vegetation index triangle method, which was used to estimate ET in the clear days. The soil moisture stress index (SMSI) was defined to express the constrain of soil moisture on ET, and clear sky SMSI was retrieved according to the estimated clear sky ET. Clear sky SMSI was then interpolated to cloudy days to obtain the SMSI for all sky conditions. Finally, time-series ET at daily resolution was achieved using the interpolated spatio-temporal continuous SMSI. Good agreements were found between the estimated daily ET and flux tower observations with root mean square error ranging between 1.08 and 1.58 mm d-1, which showed better accuracy than the former ET products, especially for the forest sites. The improved algorithm was further applied based on ESA-CCI (European Space Agency - Climate Change Initiative) soil moisture data product, and ET products in the northeastern Thailand from 2001 to 2015 was achieved and analyzed.

This project aims at developing new algorithms and finding new synergies between different remote sensing products in order to better monitor water resources in the Red River basin and in the Luan River basin, namely by combining water level, soil moisture (SM) and runoff products. Over the Red River, water balance and water management can prove quite challenging, for a number of different reasons: it has a complex topography, with a high drop of 2574 m, and since it is a transboundary river, there is a lack of information on reservoir management. The Luan River is characterized by steep hills and deep ravines at its upper reaches, leading to it overflowing during the rainy season. On the contrary, the water flow is much reduced in winter, with the river being icebound for some months. All these features amount to difficulties in getting the information on time for flood or drought early warning systems. Nevertheless, by using remote sensing data, the aim is to tackle these difficulties and obtain a better estimation of water resources, leading to a better management.

In this respect, SM products can be a powerful asset. Currently the spatial resolution of satellite SM products is quite coarse, ranging from 36 km for SMAP (Soil Moisture Active Passive) to 40 km for SMOS (Soil Moisture and Ocean Salinity). However, in some cases, higher resolutions are required. In this respect, DISPATCH (DISaggregation based on a Physical And Theoretical scale CHange) is an algorithm that downscales the SMOS and/or SMAP SM data by using MODIS (Moderate Resolution Imaging Spectroradiometer) or Sentinel-3 land surface temperature (LST) and vegetation cover data, along with a self-calibrated evaporation model. The algorithm estimates the SM variability at a 1km resolution within a low resolution pixel by relying on the self-calibrated evaporation model. More specifically, it derives a term, called soil evaporative efficiency (SEE), defined as the ratio of actual to potential evaporation, from LST and vegetation cover data. By taking into account the instantaneous spatial link between SEE and SM, it then distributes the high resolution SM around the low resolution observed mean values. Previous results obtained over the Red River basin and derived from SMOS needed further investigation due to Radio Frequency Interference (RFI) detected over the area. Since then, work has been undergone to filter the RFI from the SMOS-derived 1 km SM products. For this study, 1 km SM products have been produced over the entire Red River basin and the Luan River basin for the 2015-2018 time period, derived from both SMOS and SMAP. Preliminary results show to be promising, with an improvement with respect to the SMOS-derived products, thanks to the RFI filtering. SMAP-derived SM products have also shown promising results over the area.

By combining these enhanced 1 km soil moisture products over the Red River basin and the Luan River with water level products, the hydrological model estimations can be further improved (Li et al. 2019).

Land surface temperature (LST) plays an important role in the processes controlling energy and water exchanges at the surface-atmosphere boundary. It has been widely used in studies such as hydrology, ecology, meteorology, and climatology. Satellite remote sensing makes it possible to retrieve LST at relatively dense and regular spatial sampling intervals over large areas. Over the past three decades, satellite thermal infrared (TIR) remote sensing has become one of the most important approaches to estimate LST.

However, a major shortcomingof satellite TIR remote sensing for LST estimation is its extremely low tolerance to clouds. Clouds not only reduce the spatial coverage of the TIR LST but also decreases the actual temporal resolution. The evidence has been reported for current satellite TIR LST products over different areas in previous studies. Therefore, the performance of current satellite TIR LST product is greatly limited in many applications, especially for those requiring LST with both high temporal resolution and dense spatial coverage. In contrast, passive microwave (MW) remote sensing is insensitive to clouds: thus, it is an important independent source for LST estimation complementing the available TIR LST. Merging TIR and MW observations is able to overcome shortcomings of single-source remote sensing to derive such a LST, in which how to efficiently improve the spatial resolution of MW LST to the same level as the TIR LST is a crucial link. However, in current merging methods, models adopted for downscaling MW LST fails to quantify the effect of temporal variation of LST. Thus, the accuracy and the image quality of the merged LST can be deteriorated and therefore remain a major impediment for these methods to be generalized over large areas.

In this context, this study proposes a practical method to merge TIR and MW observations from a perspective of decomposition of LST in temporal dimension. The physical basis of the method is decomposing LST into three temporal components: the annual temperature cycle component (ATC), diurnal temperature cycle component (ΔDTC) as prescribed by solar geometry and weather temperature component (WTC) driven by weather change. For each component, a dedicated algorithm was applied to improve its spatial resolution or optimize its accuracy according to its thermal properties. The merged LST can be obtained by combining the improved components together,

The method was applied to MODIS and AMSR-E/AMSR2 data to generate an 11-year record of 1-km all-weather LST over northeast China: the resulting merged LST have a standard deviation of error (STD) of 1.29-2.71 K compared to the 1-km MODIS LST (MYD11A1) and successfully fill missing pixels due to clouds. Validated against in-situLST from three ground sites with diverse land cover types, the merged LST have a root mean square of error (RMSE) of 1.20-2.75 K, which is comparable to MODIS LST; besides, no obvious differences in the accuracy of the merged LST were found between daytime and nighttime, or under clear sky and unclear sky conditions.

The generated all-weather LST was also compared with the 1-km AATSR LST from the European GlobTemperature project. Good agreement between these two products was also found: the mean bias error (MBE) was from -0.04 to 0.14 K and the STD was from1.02 to 1.61 K.

Compared to a 1-km all-weather LST from a previous method, the merged LST derived from this study performs better in both accuracy and image quality, indicating the proposed method has an improved capability to generate 1-km all-weather LST data. The method was further applied to generate the daily all-weather LST during 2003-2017 for the Tibetan Plateau and its surrounding areas. This dataset is now being utilized in the modelling of water cycle over the Tibetan Plateau.

In this study, a time series ground-based active and passive microwave experiment for snow is presented. The experiment is carried out in 2017-2018 and 2018-2019 winter in Xinjiang, China. In the experiment of 2017-2018 winter, ground based SAR and microwave radiometers are used to measure the multiple frequency and multiple polarization backscattering coefficient and brightness temperature of snow covered soil. In 2018-2019 winter experiment, precise phase measurement is emphasized in the SAR observations to study the phase change due to snow accumulation and co-polar phase difference of terrestrial snow. Different microwave scattering and emission models of snow are used to study the measurement results, and the microwave signature of snow are studied by model simulations. The application of backscattering coefficient, brightness temperature, phase change and co-polar phase difference in snow water equivalent retrieval will be discussed.

Mountain glaciers can directly reflect local climate change and play a crucial role in the terrestrial water cycle and food security of local people. Nyainqentanglha Mountains (NM) have about 9600 km² glaciers, which account for 18.47% of the Tibetan Plateau (TP) and are the major water resources of rivers, lakes and human activities as well. The field observation is difficult to implement because of the high altitude and risk, therefore, many different experimental remote sensing techniques have been applied to estimate the glacier mass balance by several authors. Although the spaceborne optical photogrammetry is one of the promising ways to capture the glacier mass balance, the High Resolution TLA stereo images have been used less frequently. And the glacier mass balance patterns in the EM need to be further explored.

In this study, we used Zi Yuan-3 (ZY-3) Three-Line-Array (TLA) stereo images to extract the glacier mass balance in two study sites during 2000–2017. One is located in the western of the NM (WNM), a moderately complex terrain. The other one lies in the eastern end of the NM in the southeastern TP (ENM), where the topography is more complex than in the WNM. The glaciers in the WNM and ENM are of a subcontinental and maritime type, which provides an opportunity to compare and analyze the glacier mass balances of different glacier types during a decade.

The results showed that the glaciers in the WNM and ENM experienced mass loss in the 2000-2017, and the glacier thinning rates in the ablation regions were apparently larger than in the accumulation regions. In the WNM, the mean glacier elevation change and mass balance were -0.31 m a^-1 and -0.26±0.18 m w.e. a^-1, while the glaciers in the ENM obviously melted faster than the WNM, and these two values became -0.92 m a^-1 and -0.78±0.12 m w.e. a^-1, respectively. In the WNM and ENM, the glacier mass balances in the ablation zones were -0.57±0.18 m w.e. a^-1 and -1.02±0.12 m w.e. a^-1, while both values in the accumulation zones were 0.16±0.02 m w.e. a^-1 and -0.08±0.12 m w.e. a^-1.

Multi-source remote sensing data, from visible to thermal infrared are used for forcing, calibration, validation and data assimilation of/into basin scale hydrological models.

FEST-EWB model is run for the whole Heihe River basin at spatial resolution of 0.05° and temporal resolution of 1 hour. Results are provided in terms of hourly evapotranspiration, soil moisture and land surface temperature maps for the 2012.

FEST-EWB model algorithm solves the system of energy and mass balances in terms of a representative equilibrium temperature (RET) that is the land surface temperature that closes the energy balance equation and so governs the fluxes of energy and mass over the basin domain. This equilibrium surface temperature, which is a critical model state variable, is comparable to LST as retrieved from operational remote sensing data (MODIS, LANDSAT) which is used for the calibration of soil and vegetation parameters at pixel scale.

Evapotranspiration estimates are then compared at local scale with two eddy covariance data showing an overall agreement between the estimated and measured data, as certified by numerous statistical indexes.

Then, at basin scale the modelled Evapotranspiration has been compared with a number of global products: the Chinese ETMonitor, and with global reanalysis products MOD16 ET, MERRA2, ERA-INTERIM, GLDAS-2 and GLEAM. At basin scale, the agreement between the model and the ET data is consistent but presents some irregularities, as a consequence of each ET product’s own foundational hypotheses and algorithms.

The objective of this study is the evaluation of the potential of two Land Surface Temperature downscaling algorithms with respect to high resolution LST from LANDSAT 7 ETM+ and resampled MODIS LST (MOD11A1).

Four different LST sources have been compared over the Heihe river basin, an endorheic basin in China, characterized by a wide variety of ecosystems, from desert oases to irrigated croplands and wooded mountain ridges.

The first two LST sources are measurements provided by the ETM+ instruments (aboard satellite Landsat-7) at 30m spatial resolution every 16 days, and MODIS (aboard Terra at the lower resolution of 1000m. The other two sources are products of downscaling algorithms. The STARFM (Spatial and Temporal Adaptive Reflectance Fusion Model) algorithm merges ETM+ and MODIS images to obtain LST data with the spatial resolution of the former and the temporal frequency of the latter, involving neighbouring pixels in the process. On the other hand, the DisTrad (Disaggregation of radiometric Temperature) algorithm, establishes a link between LST and NDVI, in order to revert to the former anytime vegetation data is the only data available.

Globally, resampled MODIS and DisTrad perform better, not quite reaching the accuracy of ETM+ but all the same yielding an accurate approximation. On the other hand, STARFM struggles with the variety of land cover types, offering an acceptable performance in the desertic area, which is the most uniform of all. Categorizing the pixels according to their land cover type, it is found that vegetated areas, especially croplands, are the most difficult to interpret for the LST sources, with low performance in the early summer during the peak of the maize season. Furthermore, grouping pixels by their lighting condition (whether they are in light or in shadow) does not offer major results: data quality for shadowed pixels is quite worse than for lighted pixels, but the number of the formers is so low that the impact on the overall result is close to negligible.

Overall, MODIS and DisTrad are the best candidates to compensate for the low temporal frequency of ETM+ without losing too much accuracy. Furthermore, DisTrad application requires more input data and computing effort than re-sampling MODIS.

Land Surface Temperature (LST) is an indicator for the exchange of energy in the process of atmosphere-ground interaction. It is an important parameter indicator for global resource and environmental dynamic analysis. Sentinel-3A satellite was jointly developed by theEuropean Space Agency (ESA) and the European Meteorological Satellite Organization (EUMETSAT),and was successfully launched in February 2016. One of its main payloads is the Sea and Land Surface Temperature Radiometer (SLSTR), which has three channels (i.e. S7-S9) in the thermal infrared range with a 1000 m resolution in the nadir view mode. The central wavelengths of S8 and S9 are 10.85 μm and 12 μm, respectively. Thus, images of these two channels can satisfy the requirement of LST estimation.

The objective of this study is (i) to explore the applicability of the classical split window algorithms(SWA)for estimating LST from the SLSTR data acquired by S8 and S9 channels and (ii) to analyze the possible sources of error. Nine SWAswidely accepted by the scientific communities were selected as the candidate algorithms, including PR1984, BL-WD, VI1991, UL1994, WA2014, ULW 1994, SR2000, BL 1995, and GA2008. These SWAs were also used to develop the algorithm for the Chinese GLASS LST product.

The aforementioned SWAs were trained globally based on simulation datasets from the forward radiative transfer simulation. The SeeBor atmospheric profile database was used as the basis in the forward simulation. For each profile, 10 LST and near-surface air temperature differences were defined, i.e. from -16 K to 20 K in increments of 4 K; spectral emissivities of 48 materials were used; the view zenith angles were defined as 0 to 55°in increments of 5°. MODTRAN 5 model was employed to conduct the forward simulation. For each trained SWA, the NDVI threshold method was used to determine the land surface emissivity of each pixel. The European Mesoscale Weather Forecasting Center (ECMWF) data were used to determine the atmospheric water vapor content and near-surface air temperature of each pixel.

Results show that all the selected SWAs have accuracies better than 2 K in training. Ground measured LST at four ground sites of HiWATER with good spatial representativeness were used to validate the estimated LST from the actual SLSTR data during November 2016 to December 2017. Validation show that the accuracies of the nine SWAs are approximately 2-4 K, better than the official SLSTR LST product. The estimated LST is affected by many factors, such as the land surface emissivity, air humidity, land surface type, air temperature, and atmospheric water vapor content. The study would be beneficial for improving the SLSTR LST product.

The Lancang-Mekong river, known as the Lancang river in China, the Mekong reiver outside China. The Lancang-Mekong Basin is a trans-boundary river with an area of 795,000 km², including territorial parts of six countries: namely China, Myanmar, Laos, Thailand, Cambodia, and Vietnam. With a total length of over 4350 km, the Lancang-Mekong River is the longest river in Southeast Asia. It originates from the glacier melting of Qinghai Tibet plateau at the elevation of 5200 m, and eventually flows into the South China Sea at Mekong Delta in Vietnam. More than 72 million people benefit from this river. Consequently, the Lancang-Mekong basin has an outstanding practical significance for the ecological and economy development of alongshore area. However, the current land use/cover in the Lancang-Mekong river basin is in a very critical situation. Large patches of primary and secondary forests have been destroyed in Laos, Myanmar and Cambodia. Crop rotation is replaced by single cropping of rubber, cashew, sugar cane, and eucalyptus etc. Social and economic transformation, urbanization and interregional cooperation brought by increasing human activities also play an important role in land use/cover change in the Lancang-Mekong river basin. The land use/cover change have influenced climate, precipitation, the energy balance, carbon budget, and hydrological cycle in the basin.

Remote sensing has long been recognized as an effective tool for broad-scale(such as global scale, regional scale and basin scale) land use /cover mapping. At present, eight land cover thematic datasets( such as USGS with 1km, UMD with 1km, BU with 1km, GLC2000 with 1km, Globcover 2005 with 300m, GlobCover 2009 with 300m, GlobCover 2010 with 250m, Globeland30 with 30m) at a global scale have been developed with resolution ranging from 30m to 1km. In recently years, remote sensing scientists are interested in land use /cover change, climate variation, and urbanization in the Lancang-Mekong river basin. Remote-sensing technology has the potential to monitor the environmental changes in basin scale. However, the Lancang-Mekong Basin is very large, and covers numerous climate zones and eco-regions, and needs seven MODIS tiles, or over 50 Landsat frames to cover the complete north–south-trending basin. Furthermore, the almost persistent cloud cover over the Lancang-Mekong Basin for large parts of the year, and optical remote sensing images are unavailabe in part of regions. The landscapes have complex spectral and textural characterization in the Lancang-Mekong river Basin. Because of these factors, there is reported about high resolution(≤10m) land use/cover products in the Lancang-Mekong river Basin in recent years. As new satellites and sensors become avaiable. the Sentinel-2A/B are optical satellites, which respectively launched in 2015 and 2017. The Sentinel-2 has multi-spectral data with 13 bands in the visible, near infrared and shortwave infared with respectively spatial resolution of 10m, 20m, and 60m. Multi-source remotely sensed images with high temporal, spatial and spectral resolution in the Lancang-Mekong river basin have obtianed by the ESA Sentinels satellites in recently years.

Lots of research has shown that the land use /cover information play a role in climate variation, ecology and environment destroy, and naural hazards. The land use/ cover is fundamental information for natural resource management, environmental change studies, urban planning to sustainable developemnt, and many other societal benefits in the Lancang-Mekong river basin. Bacause of region area large, complex topography, cloudy and rainy environment in the Lancang-Mekong river basin, so far, the high resolution land use/cover products have not presented. The relation research has became urgent. More specifically, the whole research may include the following some parts. (1) employing Sentinel Application Platform (SNAP) toolboxes to pre-process the data sets over the Lancang-Mekong river basin (2) Developing object-oriented random forest (OORF) Classifcation algorithm for mapping Land use/cover in basin scale . (3) Mapping high resolution land use/cover in the basin using the presented alogrithms in the basin.

Glaciers in the Tibetan Plateau have significantly influenced the local ecology and economy as a water resource. Southeast Tibetan Plateau is a typical region of debris-free and debris-covered glaciers in China. Automatic glacier mapping utilizing remote sensing data is challenging due to the spectral similarity of supraglacial debris and the adjacent bedrock. Therefore, the knowledge of the changes of debris-free and debris-covered glaciers in the southeastern Tibetan Plateau is still limited. This study investigated spatial patterns and area changes at decadal scales of debris-free and debris-covered glaciers in the southeastern Tibetan Plateau by utilizing a machine-learning algorithm based on multi-temporal satellite images. Specifically, Random Forest method was applied based on Landsat and ASTER GDEM V2 data for 3 target years over 20 years (1995, 2005, and 2015). Glacier area changes were analyzed in terms of glacier characteristics (size, elevation and debris coverage) over the period of 1995 – 2015. The results demonstrated that this region has experienced a significant deglaciation of 29.86% (2842.08 km²) over a period of 20 years. The glacier size greatly influenced the change of glacier area and the shift of glacier retreat to higher elevations was notable. The melt rate of absolute area of the debris-free glaciers was faster than that of debris-covered glaciers and glaciers with varying supraglacial debris coverage responded differently. Meteorological data suggested that increasing temperature since 1995 probably represented the primary controlling factor for glacier variations in this region.