Many national and regional programs exist aimed at improving livestock productivity while maintaining the condition of natural resources. The agricultural components of these programs have mostly focused on improving the condition of the soil asset by encouraging farmers to adopt land management practices that reduce the risk of soil loss through wind and water erosion, manage soil acidification, and improve the carbon content of their soil. Are these programs having an impact on the condition of the soil resource? This has been a difficult question to answer despite significant investment, in some cases, in research to improve monitoring and reporting. Recent advancements in satellite remote sensing have led to the production of datasets depicting vegetation and soil cover and forage productivity which can be used to monitor and report on wind and hillslope water erosion risk and carrying capacity.

The Rangeland and Pasture Productivity (RaPP) Map (available in http://map.geo-rapp.org) is a web-based tool that delivers public online access to datasets of vegetation and soil cover and forage productivity. We provide global monthly coverage of these data (2001-present) from the MODIS satellite with 500 metres spatial resolution. Specifically, one dataset estimates three cover types: photosynthetic vegetation (PV), non-photosynthetic vegetation (NPV) and bare soil (BS). These three fractions are referred to as “fractional cover”. The sum of PV and NPV equals to Total Vegetation Cover (TVC). In areas with sparse or no tree cover TVC is equivalent to the ground cover (i.e. the vegetation cover in contact with the soil surface). The second dataset provides gross and net primary productivity (GPP and NPP) for the herbaceous and woody component of vegetation. In selected regions (e.g. in Australia) some of these datasets are also regularly produced with higher spatial resolution from the Landsat and Sentinel 2 sensors (i.e. with 30 and 20 metres respectively). The RaPP Map tool also includes ancillary global and country-specific layers which can help interpreting the time-series remote sensing data, including land use and land cover, soil and physiography, livestock statistics among others.

RaPP Map helps regional natural resource management organisations: 1) view and report on ground cover change, 2) plot time-series of average mean ground cover and forage productivity for an area, 3) understand the decile ranking of the ground cover level for the month by comparing it to the same month in other years, and 4) understand how many percentage points the ground cover level is above or below the long term mean for each pixel and 4) understand if the ground cover level is normal. The tool can also be used by industries and policy makers to inform them of ground cover change. In Australia, it will provide the basis for Department of Agriculture and Water Resources’ reporting on improvements in resource condition at the national level.

Future developments proposed include: training (in Australia) for natural resource management organisations in the use of the tool, improvements in the tool to facilitate ground cover target setting, exploration of whether the effects of antecedent (sum of rainfall over previous months) rainfall on vegetation cover levels can assist in the separation between rainfall and anthropogenic (land management) effects on vegetation cover. We are also working in improving the validation of the remote sensing datasets by collating existing field measurements in rangeland systems.

Rice is the staple food for more than half of the world population and it is essential to enhance the sustainability of rice production systems for global food security. Paddy rice cropland distributions, management intensity, and climate change impacts will undergo changes over the coming decades. In particular in Asia, where more than 90% of rice is produced and consumed, it is important to have consistent methods to monitor the predicted changes in rice production, e.g. by Earth Observation.
With the launch of Sentinel-1 which acquires data systematically at least every 12 days up to 6 days, time series of C-band data are available for all rice grown regions. The systematic and open data are now subject to development of robust and generalized methods for their long-term exploitation applicable for continuous monitoring at regional coverage.

In the ESA GeoRice project, works have been undertaken in the Mekong delta to provide mapping of rice grown area, rice phenological stage, and rice cropping density. To fully assess the utility of such rice monitoring system to assist decision-makers for more informed decisions that will mitigate the impact of variability in climate and in rice production systems, it is important to upscale the proof-of-concept for rice monitoring in the Mekong delta to a regional scale demonstration, covering several countries.
This is the objective of the GeoRice-extension project, which is to demonstrate large scale rice monitoring in five countries in South-East Asia (Vietnam, Lao, Cambodia, Thailand, Myanmar). The target products include seasonal rice crop maps, cropping density (for 1, 2, or 3 rice crop seasons per year), crop annual anomalies, and dynamic rice growth maps (every 12 days) depicting rice growth status.
To deal with the large volume of data, the open source GeoRice processing system is updated to run in a high performance environment, with implementation on Data Cube and DIAS being considered for large scale processing.
The demonstration will be conducted with national mandated entities and international stakeholders such as GEOGLAM/AMIS or the Mekong River Commission. The project benefits from collaboration with partners in Asia-Rice/GEOGLAM initiative, with cooperation on the VNSC DataCube.

The paper will present the updated rice monitoring algorithms, discuss challenges to upscale and transfer the GeoRice monitoring approach as well as the first results obtained on the large scale mapping of rice areas in South East Asia.

Rice is the most important staple crop in the Philippines. Like many other rice producing countries in Asia, it is cultivated in all regions of the country. A diversity of rice varieties, management practices and environments lead to a large variation in seasonality (when rice is grown) and cropping intensity (how often rice is grown per year in the same plot of land, once, twice or even three times per year). Seasonality and intensity also change over time depending on climatic, environmental and economic factors. Detecting where and when these changes occur can provides information on where and how rapidly practices have changed over time and why. Remote sensing offers a unique opportunity to detect these changes and trends with methods that can account for spatial and temporal variability.
We applied the PhenoRice algorithm to 14 years of moderate resolution remote sensing (MODIS) data (utilizing 250m resolution 16 days composites from Terra and Aqua) over the Philippines to detect rice growing areas, Start and End dates (SoS and EoS) and Length (LoS) of each season. The detected intensity and seasonality were compared to published data – static provincial level crop calendars - to check that the PhenoRice detections were valid. Then, to generate statistically robust estimates of trends in seasonality we averaged the SoS and EoS of each detected rice season within a 10km hexagonal tessellation over the entire country and mapped the spatial patterns of trends in SoS, EoS and LoS.
We observed a general trend towards shorter LoS in both wet and dry seasons, with seasons reducing by as much as one day per year over the 14 years period. The trends in SoS and EoS were much more variable over the country, with some regions showing trends of earlier SoS and EoS and others later. The strong spatial clustering of these trends suggests that they are influenced by local factors such as adoption of new management practices, changes in irrigation water release dates and weather. We discuss the spatial patterns in relation to these factors and describe how this approach could be scaled up to detect trends in rice seasonality and intensity across all of rice growing Asia.

Accurate and efficient estimation of the cultivated area of main staple crops is a crucial piece of information for humanitarian assistance, and one that is extremely difficult to obtain in an efficient and timely way. This problem is particularly serious in the context of conflict-driven food security crisis such as those in South Sudan, Central African Republic or NE Nigeria, where large scale population displacement has led to significant changes in cultivation patterns and extent. In a recovery from conflict phase it is also of crucial importance to estimate how far agricultural livelihoods have been restored.
The World Food Programme runs major large scale humanitarian assistance operations in these areas and therefore has major interest in estimating the amount of cropland in particular staple crop specific areas. A common feature in these regions is that there is no organized system for collection of agricultural area and production data and a virtual lack of primary data on location and extent of crop data.

ESA’s Sen2Agri system offers users the possibility to obtain crop type maps and associated estimates of harvested area per main crop. For this to be possible, users need to acquire ground data information on crop types across the target area extent, accounting for the existence of different farming systems (rainfed subsistence and commercial, irrigated) and agro-climatic zones.

Since mid 2017, WFP has invested in the development of an operational system to produce crop area estimates in problem areas: 1) it adapted the Sen2Agri system to run on a cloud computing and storage service (Earth Observation Data Centre, in Vienna, Austria), since the areas under consideration are fairly large. 2) it developed or adapted smartphone based tools for the capture of crop type data in the field. 3) Developed field protocols for data capture, integrating the collection of crop type data in the logistics of food security household surveys.

To date a number of experiments of crop type mapping have been carried out. Two feasibility pilots were carried out in mid-2017: the first in Karamoja (NE Uganda) and a second in a restricted area in South Sudan (North Bahr-el-Ghazal and Warrap) with dedicated data collection missions. The feasibility of data collection integrated in household surveys was tested in late 2017 in South Sudan. Full operational deployments were carried out in mid-2018 in these same areas, both with data collection integrated in the household surveys. Summary results from an experiment using drones for ground data capture, coupled with data captured with smartphones will be presented – drones may offer a much more cost-efficient way to capture ground data than what can be achieved with enumerators alone.

Results from these campaigns are presented including crop type maps and performance statistics. An analysis of the constraints, limiting factors and unresolved issues are presented with a discussion of the way forward and the role of future improvements, e.g. inclusion of Sentinel-1 data.

Effective agricultural water management requires accurate and timely information on the availability and use of irrigation water. However, most existing information on irrigation water use (IWU) lacks the objectivity and spatiotemporal representativeness needed for operational water management. Although optical remote sensing data have been used to identify areas affected by irrigation, they are not directly sensitive to actual irrigation water amounts. On the other hand, microwave observations, which are commonly used to retrieve the moisture content in the topsoil layer provide a potential, yet to date unexploited source for estimating IWU. In particular, we hypothesize that the quantitative estimation of IWU is feasible when high-quality soil moisture measurements are available.

To this end, we developed a method that combines surface soil moisture retrievals from the spaceborne SMAP, AMSR2, and ASCAT microwave sensors with modeled soil moisture from the MERRA-2 reanalysis to derive monthly IWU dynamics over the contiguous United States (CONUS) for the period 2013-2016 (Zaussinger et al., 2018). The methodology relies on the assumption that the MERRA-2 land surface hydrology formulation does not account for irrigation, while the remotely sensed soil moisture retrievals do contain an irrigation signal. For many CONUS irrigation hot spots, the estimated spatial irrigation patterns show good agreement with an irrigated-area reference map. Moreover, in intensively irrigated areas, the estimated temporal dynamics of IWU agree well with reference data on local irrigation practices. State-aggregated mean IWU volumes derived from the combination of SMAP and MERRA-2 soil moisture show a good correlation with statistically reported state-level irrigation water withdrawals.

However, our methodology also systematically underestimates the official estimates. We argue that this can be mainly attributed to the inability of the employed coarse resolution satellite soil moisture retrievals to resolve small-scale irrigation practices, which jointly can account for significant water use at larger scales. To analyze irrigation dynamics closer to field-scale, higher resolution soil moisture (e.g., from Sentinel-1) and ancillary data (e.g., dynamic crop maps from Sentinel-2) will be exploited to further enhance the accuracy of IWU mapping.
---
Zaussinger, F., Dorigo, W., Gruber, A., Tarpanelli, A., Filippucci, P., and Brocca, L.: Estimating irrigation water use over the contiguous United States by combining satellite and reanalysis soil moisture data, Hydrol. Earth Syst. Sci. Discuss., https://doi.org/10.5194/hess-2018-388, in review, 2018.

Monitoring irrigated areas has received considerable attention given the importance for food security, modeling the water exchange between land and atmosphere, analyzing the impact of climate change and weather variability on crop water requirements, as well as the management of water resources that affect global food security. However, the current extent of irrigated areas over continental to global scale is still uncertain. Existing maps, especially covering large areas, are mainly derived from country-level statistics.
Remote sensing has proven to be a valuable tool to map agricultural production areas. This can range from mapping the areal coverage of croplands, the productivity of these areas, as well as obtain information on the management practices. Satellite remote sensing offers an immense potential for monitoring irrigation, given the scale and period for which these type of data are available. However, studies and operational products for mapping irrigated areas remain limited. A first aspect is the difficulty of mapping agricultural areas in general, given their highly dynamic nature and their complexity. This in consequence also impacts the task of identifying irrigated areas as a subclass of cultivated areas. A second aspect is the difficulty to make the step from land cover (cropland) to land use (irrigation practice), which requires additional data sources other than the optical satellite data. In addition, detecting irrigated fields has proven to be difficult in humid areas when considering only optical data given the high overlap between crop development and resulting spectral profiles, as well as the high cloud coverage which reduces data availability.

We propose a generic irrigation mapping method which provides a solution to these limitations, by combining information on (i) the land cover, (ii) the phenology as obtained from time series analysis, (iii) actual Evapotranspiration (ETa) and (iv) precipitation. An important note is that the current focus is on mapping areas which receive full application of water to overcome periods of rainfall deficits. Supplementary irrigation mapping is the focus of future developments, given the important role it can play on the water balance. Additionally, the basis of the irrigation mapping is a labeling methodology, meaning that the delineation of the land cover cropland areas and the land use irrigated/rainfed are two separate and independent steps. Although the accuracy of the mapped irrigated croplands is to a large extent determined by the accuracy of the delineated cropland areas, this will not be discussed here. Focus will be on the irrigation labeling methodology.

This mapping effort is part of the FAO portal to monitor Water Productivity through Open access of Remotely sensed derived data (WaPOR - https://wapor.apps.fao.org), where all the data used in this study are available. On this portal, information is freely available on cropland areas, such as land cover mapping, phenology, biomass production, evaporation, transpiration, and precipitation. The described methodology has been used to map the irrigated cropland areas at 250m (MODIS-based) and 100m (Proba-V based).

The general principle behind the irrigation labeling method is the calculation of the Water Deficit Index (WDI). It compares within a growing season the available water inputs from precipitation, and the water outputs from the transpiration and evaporation. Thus, three inputs are needed to calculate the WDI on a per-pixel basis: (i) start and end of the growing season from the NDVI profiles, (ii) ETa on a ten-daily basis, and (iii) daily precipitation (here based on the CHIRP and CHIRPS datasets, available at roughly 5km resolution but resampled to the target resolution). Based on the phenology-information, the total seasonal ETa and precipitation are calculated. The WDI is then defined as the ratio of the seasonal precipitation (Ps) and the seasonal Eta (ETas). For Ps, the accumulation starts one month before the start of the growing season to account for water already available in the soil at the beginning of the growing season. A WDI of 1 or larger means that water made available through precipitation sufficed to sustain the water that was consumed by the crops (ETas). A WDI<1 means a larger water consumption was observed in relation to the available water from precipitation over the growing season.

Given the initial aim of mapping only areas which receive full application of water, a threshold of WDI<0.9 sufficed. When supplementary irrigation would also be of interest, a more detailed analysis of the WDI with respect to rainfall distribution and actual biomass production would be needed. In the figure in attachment, for an area around the Sous river in Morocco (a), it is shown how an existing land cover map ((b) - Copernicus 100m land cover of Africa) was irrigation-labeled (d) based on the WDI-threshold of 0.9 (c).

These maps are made on a seasonal basis, i.e. for each growing season in the period 2009-2017 the WDI is calculated and used to label the detected seasons as rainfed or irrigated. In addition, annual land cover maps are made. In this step inconsistencies in irrigation labelling between the years are taken into account, and a pixel is labeled as irrigated if (i) one or more growing seasons in that year are labeled as irrigated, or (ii) the majority of the years in a five-year window are labeled as irrigated. This way, one-off misclassifications are taken into account.

Future developments focus on more consistent mapping of the areas over the years, as well as fine-tuning the WDI to local conditions to also map supplementary irrigation.