Real-time precipitation retrieval is crucial for timely warnings of extreme weather events such as heavy rain or floods. Geostationary satellite observations combined with machine learning methods provide an effective way to achieve real-time precipitation retrieval, yet are hampered by two issues. The traditional stable model, pre-trained with historical data, lacks precipitation information from adjacent time periods and its accuracy declines over time. Geostationary satellites can only capture cloud top information, leading to inaccurate localization of precipitation areas in estimations based solely on such data. To solve these problems, a novel Dynamic Two-step Real-time Precipitation Retrieval Algorithm (D-PRA) has been developed. It conducts real-time retrieval in two steps through dynamic data and variables selection: identifying precipitation and then retrieving precipitation intensity. It further incorporates atmospheric profile information to enhance the details of in-cloud and under-cloud conditions. D-PRA was applied to Himwari-8 observations and was validated with the rain gauge observations in Chinese regions at the hourly scale. The results showed that, in comparison with GSMaP_NOW, D-PRA exhibits significant enhancements. The probability of detection attained 0.72, representing a nearly fivefold increment, and the root mean square error was 0.99 mm, a 25.5% reduction. Moreover, D-PRA is stable across seasons and time periods, demonstrating its good reliability and robustness. D-PRA has great potential to improve the accuracy of precipitation retrieval for extreme event monitoring and disaster management.

Accurate and timely precipitation data plays a crucial role in flood monitoring, forecasting and emergency management. The primary challenge with current real-time precipitation products is their limited accuracy in identifying precipitation. To improve the accuracy of precipitation retrieval, a dynamic two-step precipitation retrieval method was developed, i.e., first identifying precipitation and then estimating its intensity. The geostationary satellite infrared brightness temperature data, atmospheric profile data, and rain gauge observations were used to construct the precipitation identification model and precipitation intensity retrieval model based on a random forest machine learning method. This method was applied to precipitation retrieval in China and surrounding areas using the Himawari-8 observations and the ERA5 atmospheric profile data. The results of hourly precipitation identification showed good agreement with rain gauge observations, with a Probability of Detection of 0.725 and a False Alarm Ratio of 0.426. The estimated hourly precipitation intensity agrees well with the rain gauge data and shows a lower error than the GSMaP_NOW real-time precipitation product. The good performance of the developed method in this study for real-time precipitation monitoring was also confirmed in different seasons.

In the Tibetan Plateau (TP) region, the foreseeable increase in air temperature may have profound and complex effects on the local hydrological cycle, and is likely to increase water loss from the land surface to the atmosphere through evapotranspiration (ET). Quantifying ET and its regulatory mechanisms are major challenges for understanding the water cycle and land-atmosphere interactions in the TP region. We evaluated the performance of several Earth observation-based ET datasets in the TP region, and explored the spatiotemporal variation of ET in the same region. The accuracy of different global ET datasets was evaluated, and ETMonitor and PML-V2 provide the best accuracy with overall high correlation, low bias, and low root mean square error. ETMonitor ET is also the only product with both high spatial (~1 km) and temporal (daily) resolution. ETMonitor ET may reflect the effect of mountain topography on ET better than other global products, i.e., ET values are higher in the humid valleys with denser vegetation cover and higher soil moisture, and ET values are lower on the mountain slopes at higher elevations with less vegetation cover and colder climate. Other ET products failed to capture the spatial patterns of ET in the mountainous regions, and this suggests that the spatial resolution is not the only dominant factor leading to the poorer performance of these ET products in the mountain regions of the TP. The results show that multi-year average ET is 339 mm/yr in the TP region during 2000-2021, which accounts for about 51% of the total precipitation in the TP region. From 2000 to 2021, ET over the Tibetan Plateau shows an overall increasing trend with large spatial variability.

Evapotranspiration (ET) plays a crucial role in the global energy and water cycles. Clarifying the time-series characteristics of ET is essential for accurately estimating ET and comprehending the ET change. This study utilized the latest ETMonitor product to analyze ET time-series characteristics in China from 2001 to 2021. The standard deviation and coefficient of variation were used to measure the absolute and relative variability of ET. A monthly ET time series was decomposed using an additive decomposition method to analyze its components. The results showed that China's ET exhibited an increasing trend, especially in the middle regions of the Yellow River to northeastern China, at a rate exceeding 5 mm/year. Northwestern China showed larger variation coefficients due to lower ET. Seasonal and irregular components were found to dominate the monthly ET time series, with significant seasonal components observed in eastern China and irregular components dominating the western region. Moreover, the seasonal characteristic of ET was more evident in north and northeast China compared to southern and northwestern regions. Additionally, relative variation in ET during winter was more significant than in summer, primarily in northern river basins.

Terrestrial surface soil heat flux is an important component of surface energy balance, uncertainty in surface soil heat flux may cause errors in evapotranspiration estimation when using energy balance-based algorithms at large scale using remote sensing observations. This study assessed the performance of different equations for daily averaged soil heat flux estimation for large-scale application using parameters derived from remote sensing observations, including five soil heat flux equations used in different evapotranspiration models and three machine learning-based soil heat flux algorithms. These algorithms, taking soil heat flux as a constant or varying fraction of net radiation, generally overestimate daily soil heat flux with root mean square error (RMSE) of 14.90 ~ 23.12 W m-2. The machine learning methods could achieve better accuracy of daily soil heat flux estimation, with the lowest RMSE of 3.85 W m-2 by random forest algorithm based on in situ observations at flux tower sites. The random forest algorithm was further applied for global soil heat flux estimation, and the obtained soil heat flux could capture the global patterns of soil heat flux well with RMSE of 4.77 W m-2. The results indicate that soil heat flux could be well estimated by the machine learning method, and it is promising to improve regional evapotranspiration estimation based on remote sensing observations.

This study quantitatively evaluated the performance of six global drought indices (the self-calibrating Palmer Drought Severity Index -scPDSI, the Standardized Precipitation-Evapotranspiration Index - SPEI, the Global Precipitation Climatology Centre Drought Index - GPCC_DI, the Multivariate Standardized Drought Index - MSDI, the Standardized Soil Moisture Index - SSI, and the Standardized Precipitation Index - SPI) over China using the drought records from the Emergency Events Database (EM-DAT) by developing two indicators of the monitored drought area percentage (MDAP) and the monitored drought period percentage (MDPP). The results showed that scPDSI, SPEI, GPCC_DI, and MSDI can capture drought events in China from 1980 to 2015 better than SPI and SSI, with MDAP of ~80% and MDPP of ~70% at optimal timescales, among which SPEI and MSDI is slightly better than scPDSI and GPCC_DI.

Thailand is characterized by typical tropical monsoon climate, and is suffering from serious water related problems, including seasonal drought and flooding. It is critical to study the spatiotemporal pattern of evapotranspiration (ET) in Thailand to support the local water resource management. In this study, daily ET was estimated over Thailand by ETMonitor, a process based model, based on mainly satellite earth observation datasets. The original ETMonitor is improved by combing land surface temperature -based method to calibrate the algorithm regionally by parameterizing the constrain of soil moisture. Good agreements were found between the estimated daily ET and flux tower observation with root mean square error from 1.06 to 1.21 mm d-1. The Chi and Mun river basins, located in the Northeast Thailand, were selected to analyze the ET spatiotemporal pattern. The results indicate that the ET had large seasonal variability in these two basins, which is largely influenced by the monsoon climate.

The meteorological drought, caused by a shortage of precipitation over a certain period, is the root of a causal chain in drought series in agriculture, ecology and hydrology. Different precipitation datasets produced inconsistent conclusions in the characteristics and trends of drought events. This study analyzed the characteristics and trends of meteorological drought over China using the standardized precipitation index (SPI) derived from TRMM and CMORPH precipitation data. The annual precipitation over China was increased from 1998 to 2014, indicating a wetting trend over China in the past near twenty years. CMORPH data showed more significant wetting trend than TRMM data. The wetting trend mainly occurred in northern China, while southern China has experienced a drying trend. The drying trend over China in spring and summer is more obvious than in autumn and winter. However, all analysis showed that the drought area over China as a whole was decreasing from 1998 to 2014.

The state of the land surface and the water cycle over the South and East Asia can be determined by space observation. New or significantly improved algorithms have been developed and evaluated against ground measurements. Variables retrieved include land surface properties, i.e. NDVI, LAI, FPAR, albedo, soil moisture, glacier and lake levels. Based on these biophysical parameters derived from microwave and optical remote sensing observations, a hybrid remotely sensed evapotranspiration (ET) estimation model named ETMonitor was developed and applied to estimate the daily actual ET of the Southeast Asia at a spatial resolution of 1 km. The changes in glaciers and lakes on the Tibetan Plateau, and the drainage links between glaciers and lakes are determined in this climate-sensitive region.

Radiometric measurements taken at a distance from evaporating bodies have been used for about 50 years to obtain a measure of water use. The basic physics of determining water use is the principle of energy conservation at the evaporating surface, but energy flux densities need to be parameterized using the variables which can be actually captured by radiometric measurements. Both observations of surface temperature and of soil water content are used for this purpose. Models of energy and water exchange vary from semi-empirical relationships between latent heat flux density and surface temperature to fully three-dimensional models of the soil-foliage-atmosphere system. Several ET satellite data products are available and evaluations against ground measurements have been published. Results suggest a dependence of accuracy on climate, soil and vegetation. The review of documented performance is combined with the sensitivity analysis of widely used ET algorithms, e.g. SEBI, SEBS, SEBAL and ET Monitor.

Global rainfall interception loss information is essential to understand the dynamics of the water cycle. And it remains large uncertainty in its global variation, since its ground observation are scattered. The development of satellite earth observation provides good opportunity for global rainfall interception loss estimation to reveal its spatiotemporal variation. In current study, an analytical model, Gash model, was revised and applied for global rainfall interception estimation based mainly on satellite based remote sensing products, e.g. precipitation, leaf area index, canopy height. The ratio of interception loss to precipitation showed good agreement with in situ observations, with correlation factor and RMSE of 0.65 and 6.17%. Large spatial variation of rainfall interception loss were found with high value in tropical rainfall forest region and high latitude forest regions. However, the temporal variation was much less.

Evapotranspiration (ET) is an important ecohydrological process especially in arid and semi-arid regions. In current study, a process-based model named ETMonitor was developed to estimate the ET, based mainly on the biophysical and hydrological parameters retrieved from satellite earth observations. And global daily ET from 2008 to 2012 with a spatial resolution of 1 km was estimated based on multi-source earth observations datasets. The estimated ET agreed well with the in situ observations at field scale, with R2= 0.74, Bias = -0.05 mm d-1, RMSE = 0.87 mm d-1. The spatial patterns of estimated ET also agree well with the current available global ET products such as MOD16 and GLEAM. The ET products provide critical information on global terrestrial water and energy cycles and environmental change.