Lebanon currently hosts the highest number of refugees per capita in the world. Groundwater levels in some regions of the country have decreased by more than 20 meters in the last decade. This has been attributed to a rapid population increase, expansion of agricultural activities, and a drying climate. Water use estimates at the regional and the field scale in the country are crucial for enhancing resilience and for managing competing sectorial water demands. In this research, we derive a novel time series of field-scale actual evapotranspiration for the country of Lebanon using an energy balance model for the period 1984-2017 utilizing all available original Level 1 Landsat thermal imagery and Level 2 surface reflectance products including Landsat 4, 5, 7, and 8 and local weather data. An automated hot and cold pixel identification procedure run in batch mode is implemented via pySEBAL, a new version of SEBAL. Within-Landsat scene interpolations of ET maps between 2015 and 2017 were improved by utilizing NASA’s newly developed harmonized Landsat 8- Sentinel -2 product. The product was used to improve values of daily evaporative fraction (fRET) by correlating it with at-satellite fraction of vegetation cover and deriving fRET maps for Sentinel-2 overpasses. Analysis of 1200 cloud-masked and gap-filled Landsat-derived ET products shows that ET in the country is a major consumer of water. For example, agriculture in the heavily cultivated Bekaa valley of Lebanon consumes more than 900 mcm/year. Most (64%) of the ET is consumed from blue water in the dry season of May-October. ET in the wet season of November – April consumes more than 70% of the precipitation in the Bekaa valley. We analyze changes in storage in several major basins including the Litani and the trans-boundary Orontes. A water balance analysis over the upper Litani Basin for example shows that groundwater storage in the Litani basin is decreasing at a rate of 330 ± 50 mm/yr. The outcome of this research also offers unique opportunities for estimating past and future trends of agricultural activities in the country and it provides new estimates of water balance for all watersheds. Results show the importance of field-scale automated energy balance models for estimating previously uncertain water balance components such as groundwater storage and recharge/deficit.

Area B - Board 157

Evapotranspiration monitoring in near real time is required in several applications linked to agriculture, water management, river management and power production. EUMETSAT Satellite Application Facility on Land Surface Analysis has been providing products related to instantaneous and daily evapotranspiration since 2005 based on the observations from the EUMETSAT MSG mission. An improved version of the evapotranspiration algorithm has recently been implemented allowing the retrieval of the associated surface heat fluxes. In this contribution, the retrieval algorithm is briefly presented, and the new products are evaluated against other data sources not necessarily based on satellite observations, such as in-situ observations, short-term forecasts from the European Center for Medium-range Forecasts (ECMWF) and the Global Land Data Assimilation System (GLDAS). The results show an increased capability of the new algorithm towards the near real time monitoring of the evapotranspiration at continental scale, with less than 20% of error for most climate and vegetation conditions, especially thanks to the assimilation of the biophysical parameters and the land surface temperature derived from MSG and the improvement of the model parameterization. In average, the correlation coefficient found with yearly time series recorded by around 50 eddy flux ground stations over the period 2007-2011 is in average equal to 0.78 and 0.73 for the daily and instantaneous products respectively, a root mean square difference equal to 0.73 mm per day and 0.08 mm per hour (and 54 W.m-² for instantaneous latent heat flux) and a mean bias of -0.11 mm per day and -0.02 mm per hour (and -16 W.m-²). The comparison with the 3-hourly one quarter-degree resolution ECMWF and GLDAS datasets shows the added-value of using the MSG EO mission for semi-arid regions, irrigated and low lands, while some tropical and equatorial humid regions show un-resolved large discrepancies between the 3 systems, that should lead to further research efforts at international level.

Area B - Board 158

These days, the environmental problems such as global warming are often taken up and discussed worldwide, and observations from satellites play an important role to clarify changes in the global environment. The GCOM (Global Change Observation Mission) project was founded to observe the global changes, and was configured by GCOM-W (- Water) and GCOM-C (- Climate), both of which are the JAXA (Japan Aerospace Exploration Agency) satellites. GCOM-W was launched in March 2012 and the purpose is to solve the water circulation. After the GCOM-W started observation, GCOM-C was launched on December 23rd, 2017, the purpose of which is to reveal the global-scale climate changes which are caused by human activities and/or by nature. GCOM-C started its observations on January 1st, 2018, and released the products, after one-year calibrations and validations, in December 2018. We will focus on GCOM-C in the presentation.

To observe the global climate change, GCOM-C has SGLI (Second Generation Global Imager) sensor which is the multi-wavelength optics sensor and consists of 2 types of sub-sensors. One is the SGLI-VNR (Visible, Near Infrared Radiometer) sensor which can observe visible and near-infrared band, and the other is SGLI-IRS (InfraRed Scanning radiometer) sensor which can observe from short-wavelength infrared to thermal infrared bands. SGLI-VNR has 2 polarization and multi-direction (POL) channels and 11 non-polarization (NP) ones, out of total 13 channels. Also, SGLI-IRS has 4 of short-wavelength infrared (SWI) channels and 2 of thermal infrared (TIR) ones, out of total 6 channels. All the channels except POL ones can observe data at 250m spatial resolution. By using these multi-bands, high-resolution sensors, JAXA provides 28 physical products such as aerosols produced by human activities, sea surface temperature and amount of chlorophyll which is related with human activities at ocean, etc., as shown in table 1. Fifteen of these products are regarded as Essential Climate Variables (ECVs), which are defined as physical, chemical and biological parameters or a group of those considered as indicators of global change covering atmosphere, terrestrial, ocean and cryosphere.

The GCOM-C products can be downloaded for free from JAXA’s data dissemination system, G-Portal (https://gportal.jaxa.jp/gpr/). G-Portal provides not only the GCOM-C products but also other products of JAXA satellites. Associated information is also available on GCOM-C web page (https:// suzaku.eorc.jaxa.jp/GCOM_C/). The products can be discovered through GEOSS portal(http://www.geoportal.org/) developed by GEO (Group on Earth Observations), and IDN (Internet Directory Network, https://idn.ceos.org/) developed by CEOS (Committee on Earth Observation Satellites). We hope that many researchers and business persons use our products. In the presentation, we introduce details of the satellite and products, how to download and use the products. Also we talk about the early outcome.

Area B - Board 159

This paper presents a cross-product framework to assess the uncertainty of an ensemble of atmospheric radiative forcing estimates due to the direct effect of fire-driven surface albedo change. Radiative forcings are modeled using Earth Observation land surface albedo and burnt area products, at several spatial resolutions, over the 2005-2011 period. Nine combinations of products between the MODIS Collection 6, the Copernicus Global Land Service (C-GLS) products, and the Fire-disturbance Climate Change Initiative (CCI) burnt area and GlobAlbedo products, were used and results were summarized by vegetation cover type.

Results show that depending on the combined products and spatial resolution, estimates can differ significantly. In general higher estimates result at coarser resolutions and variation between product combinations can differ between 26% to 46%, depending on the type of vegetation.

This study presents an example of inter-ECV evaluation at different spatial scale for which there is a known cause-and-effect relation. In the context of an increasing number, and coverage, of Earth Observation satellite programs, the results highlights the need for fitness for purpose, not only in context of product accuracy but also in terms of product compatibility for the study of the changing climate and environment.

Area B - Board 160

Surface albedo is the fraction of incoming solar radiation reflected hemispherically by the surface. It is one of the essential climate variables (ECV). It serves as an indicator of climate change and change of the albedo will also affect the climate. The Satellite Application Facility project on Climate Monitoring, CM SAF, which is financially supported by European Organization for the Exploitation of Meteorological Satellites (EUMETSAT), has recently developed a surface albedo data record used in this study (CLARA-A2 SAL) covering the years 1982-2015, which is based on homogenized AVHRR data. The CLARA-A2 SAL product is given as monthly means and 5-day (pentad) means and the retrieved albedo is defined to wavelength range 0.25 -2.5 µm and the observations are averaged to 0.25° grid, which is also the resolution of the final product. Cloud cover constitutes a major challenge for the surface albedo estimation using AVHRR data. CLARA-A2 SAL was based on the PPS binary cloud mask. However, the demand of pixelwise high accuracy cloud masking based on only single AVHRR images without any augmenting information is not realistic in all possible conditions of cloud fraction and cloud type on any land cover type, especially for the past satellites lacking or having low quality SWIR channels. Another approach to tackle cloudy conditions for the next surface albedo product release, CLARA-A3 SAL, is presented in this study. It is based on cloud fraction probabilities and albedo distributions.

The surface albedo is first determined for every pixel without taking possible cloudiness into account. Then monthly distributions of the calculated albedo values are generated for each pixel. For typical snow-free land cover types the cloudy pixels constitute an essentially exponentially decaying tail of higher values for the surface albedo distribution. In addition, every pixel is provided with the information of the cloud probability and cloud fraction. Since the surface albedo of a snow-free pixel does not typically vary in a large range, whereas the cloud cover albedo does, the cloud-free pixels constitute the peak of the distribution. Using the half and ¾ height points of the peak it is possible to estimate the surface albedo applying a simple physically motivated formula. For snow-covered terrain the cloudy pixels constitute an essentially exponentially decaying tail of smaller albedo values, as snow is typically higher reflecting than clouds. For old snow the cloud and snow albedo distributions may coincide. For snow-free land cover the albedo distribution alone is mostly sufficient for albedo estimation, but for snow covered terrain the availability of pixelwise cloud fraction and cloud probability values improves the discrimination of the land (snow) from cloud cover. For monthly means the number of individual observations is sufficient. For pentad means only those pixels are used that belong in the monthly analysis to the cloudfree part of the albedo distribution.

Area B - Board 161

Surface albedo, defined as the ratio of the outgoing to incoming solar radiation at the Earth’s surface, is an essential climate variable. Surface albedo is used to characterize the surface radiation budget and can affect Earth’s climate system by modulating energy balance and is, therefore, used widely in climate models. However, climate models usually parameterize surface albedo following simplified assumptions and are unable to be calibrated with ground measurements on a global basis due to scale differences and limited spatial/temporal coverage. Multiple long-term global surface albedo datasets have recently been derived from satellite observations and extensively validated and were used in this study for comparisons with climate models. In this study, we provide the first synoptic estimates of surface albedo on a global basis from multiple long-term satellite datasets and CMIP5 model simulations. The global annual surface albedos from observations are estimated to be 0.237, 0.081, and 0.124 over land, ocean, and global surfaces, respectively. The corresponding results from the CMIP5 models are 0.251, 0.080, and 0.130, respectively. Compared with satellite datasets, the CMIP5 models tend to overestimate surface albedo in general, carry larger inter-model spread, and have limited capability of replicating interannual surface albedo variations. Declining surface albedo is found in most of the satellite datasets and CMIP5 models on a global scale, especially in the period after 2000, the magnitude of which still carries large uncertainties. The albedo uncertainties at high latitudes and/or in winter are likely related to parameterizations over snow/ice covered regions in the CMIP5 models. Meanwhile, satellite datasets also bear with large uncertainties in high latitude winters, which requires more in-depth evaluation of satellite datasets and improvement of retrieval algorithms. In the end, we recommend using the ensemble mean values from the satellite datasets to calibrate and validate surface albedo from climate models for future model refinement.

Area B - Board 162

Afforestation is proposed as an effective tool to protect environment and increase ecosystem services. It contributes to climate mitigation through carbon sequestration, and also impacts local climate via altering energy budgets. The net warming or cooling effect of afforestation in temperate regions is still uncertain, depending on the competing contributions of albedo and evapotranspiration as well as background climate. Since the launch of the ‘Gain for Green’ program (GPP) by China’s central Government in 1999 to increase forest cover and control soil erosion, parts of agriculture land have converted to forests and grassland in Loess Plateau, a pilot region of this program with arid and semi-arid climate.

The objective of this study is to investigate how surface temperature and precipitation response to afforestation due to government policy in Loess Plateau, using multiple long time series satellite products from 1982 to 2016 combined with a coupled atmospheric-land model (WRF/Noah-MP). To highlight the effect of GPP and avoid sensors calibration issue, the whole temporal range is divided into two periods: from 1982 to 1999 and from 2000 to 2016. Forest, crops and grassland fractions derived from 30m China’s Land-Use/cover Datasets are translated into percentages of corresponding Plant Function Types as input for Noah-MP land surface model. Spatial patterns of 2-m air temperature and precipitation are simulated by coupled WRF/Noah-MP model for different periods, constrained with prescribed long time series satellite products.

It is found that afforestation leads to a cooling effect in Loess Plateau from satellite observations since the launch of GPP. Increasing precipitation trend is also observed from PERSIANN CDR and ground rain gauge stations from 2000 to 2016, contrasting with the decreasing trend from 1982 to 1999. Simulations from coupled WRF/Noah-MP model aid to explore the underlying mechanism driving temperature and precipitation response on afforestation with land-atmosphere interactions into consideration.

Area B - Board 163

The Earth Polychromatic Imaging Camera (EPIC) onboard Deep Space Climate Observatory (DSCOVR) provides 10 narrowband spectral images of the sunlit side of the Earth. The blue (443 nm), green (551 nm), and red (680 nm) channels are used to derive EPIC broadband radiances based upon narrowband-to-broadband regressions developed using collocated MODIS equivalent channels and CERES broadband measurements. Comparisons between the broadband radiances derived using these regressions and CERES observations yield relative biases of 0.6% and 3.4% for ocean and non-ocean surfaces, while the relative RMS errors are 5.8% and 6.6%, respectively.
The pixel-level EPIC broadband radiances are averaged to provide global daytime means at all applicable EPIC times. They are converted to global daytime mean shortwave (SW) fluxes by accounting for the anisotropy characteristics using a cloud property composite based on lower Earth orbiting satellite imager retrievals and the CERES angular distribution models (ADMs). EPIC composite which includes all information needed for angular distribution model selections was developed to accurately characterize the anisotropy for each EPIC pixel. EPIC SW fluxes are derived at all applicable EPIC image times. Global daytime mean SW fluxes show strong diurnal variations with daily maximum-minimum differences as great as 20 Wm-2 depending on the conditions of the sunlit portion of the Earth. The EPIC SW fluxes are compared against the CERES SYN1deg hourly SW fluxes. The global monthly mean differences (EPIC-SYN) between them range from 0.1 Wm-2 in July to -4.1 Wm-2 in January, and the RMS errors range from 3.2 Wm-2 to 5.2 Wm-2. Daily mean EPIC and SYN fluxes calculated using concurrent hours agree with each other to within 2% and both show a strong annual cycle. The SW flux agreement is within the calibration and algorithm uncertainties, which indicates that the method developed to calculate the global anisotropic factors from the CERES ADMs is robust and that the CERES ADMs accurately account for the Earth's anisotropy in the near-backscatter direction.

Area B - Board 164

The FLUXCOM initiative (www.fluxcom.org) generated a large data-driven ensemble of global land-atmosphere energy fluxes based on integrating flux tower measurements of energy fluxes with machine learning methods, and global satellite and climate data. This archive includes in total 147 gridded products of latent and sensible heat fluxes, as well as net radiation using different machine learning methods, driving data, and three ways to account for systematic measurement bias. We distinguish two setups: (1) An “RS” setup with fluxes estimated from MODIS satellite data at an 8-daily temporal and 0.0833° spatial resolution for the period 2001-2015 with eight machine learning methods. (2) An “RS+METEO” setup with fluxes estimated from meteorological data, and mean seasonal cycles of satellite data with daily temporal and 0.5° spatial resolution with three machine learning methods and 4 global climate forcing data sets. Our ensembles are suitable to quantify and analyse land-atmosphere energy fluxes and in particular its uncertainties which is of particular relevance when using our data-driven products for benchmarking global land surface model simulations or for assessing energy budgets over land.
This contribution will outline the experimental design, methods, results, and evaluation of results against independent data sets. We present global and continental budgets of land energy fluxes with quantified uncertainties. We will show that the two experimental setups yield consistent energy budgets within their respective uncertainties. We will further show how uncertainties of the energy balance closure gap at flux tower sites propagates to uncertainties of global and continental energy budgets and analyse which energy flux is subject to the largest relative uncertainty. We assess the individual advantages and disadvantages of the RS and RS+METEO setups, and outline conceptual and practical limitations of the data-driven approach with respective implications for interpretations in applications ranging from energy budget assessments, land surface model benchmarking, and the analysis of spatial-temporal changes in land-atmosphere energy fluxes.

Area B - Board 165

Since the pioneering Global Positioning Satellite (GPS) radio occultation (RO) mission GPS/Met in the mid-1990s, and in particular since the start of the continuous RO dataflow by 2001, an RO data record of almost two decades has been accrued that is clearly set to continue also over the next decades to come. RO probes the fundamental state of the atmosphere, the thermodynamic state of the gas as expressed by air density, pressure, temperature, and tropospheric water vapor. These are the fundamental thermodynamic variables for tracking climate variability and change, including in terms of atmospheric heat content (AHC), both for sensible heat (SH) and latent heat (LH).

The RO remote sensing technique provides exceptional utility for the purpose due to its unique combination of high accuracy, high vertical resolution, long-term stability, and nearly all-weather global coverage in its monitoring of the troposphere and stratosphere. A data record from select high-quality (Vaisalä) radio sondes (RS) complements these data since the mid-1990s as do re-analyses (RA) of increasing quality and long-term stability such as the recent European Reanalysis ERA5.

We use the RO data record 2001-2018, complemented by (from 1993 onwards) and cross-evaluated with RS and RA data records, to compute and analyze latitude-resolved monthly-mean AHC and its SH and LH components over the last 25 years. Based on understanding the level of long-term stability and homogeneity in the records, and related limitations, we investigate SH, LH, and total AHC changes and heat energy exchanges along different domain separations. These include heat energy exchanges among the main atmospheric layers as delineated by the altitude-resolved RO data (boundary layer, free lower troposphere, upper troposphere, and stratospheric layers), as well as heat content trends and interannual variability (e.g., from El Nino/La Nina) in interaction with upper ocean heat content (OHC) changes (from mixed layer to a 400-700 m depth layer).

We show results from this work based on our Wegener Center Occultation Processing System OPSv5.6 RO data record and related Vaisalä RS and ERA-Interim RA records. A significant long-term heat accumulation is visible, in particular in the tropical troposphere, as are the imprints of El Nino/La Nina and other variability modes. In addition, we discuss the status of the SH/LH/AHC monitoring based on our new Reference Occultation Processing System (rOPS), which provides SI-traceable RO data complemented by ERA5 RA data and includes integrated propagation of estimated systematic and random uncertainties from the raw RO data down to the final SH/LH/AHC data.

Area B - Board 166

It is essential to measure Earth Energy Imbalance (EEI) in order to monitor and understand the perturbation of the water and energy flows in the climate system in response to external forcing and in particular to anthropogenic forcing. Over 93% of the excess energy that is gained in response to the EEI is stored in the ocean. For this reason the global ocean heat content (OHC) places a strong constraint on the EEI and the ability to determine the global ocean heat storage change is essential to assess the changes in the global water energy cycle. Two different approaches are currently available to measure the OHC changes from space. The first approach is to measure the net ocean surface fluxes with satellite-based surface radiative flux, and turbulent fluxes. The second approach is to estimate the thermal expansion of the ocean which is linearly related to the OHC changes. Measuring the thermal expansion of the ocean is possible
from space by combining satellite altimetry and space gravimetry observations. Both approaches have advantages and disadvantages. The first approach provides all components of surface fluxes, which enables to compute the error in the net regional flux based on the error in each component. For this reason, the error is traceable. The disadvantage of the first approach is that the net flux computed by summing up all components over the ocean differs significantly from the EEI. The second approach is intrinsically global and enables to estimate the EEI and its variations at annual time scales with an accuracy of a few tenth of W.m-2. But it does not allow for the moment to get accurate net regional estimates of the changes in OHC. In this presentation we propose to to review the recent progresses achieved in both approaches to estimate EEI. We also propose a way forward in which both approaches are combined in a variational framework.

Area B - Board 167