eStation+methods

toc =1. Background information=

1.1. Describing local climate
Long term environmental information are provided by the eStation to assess environmental trends and changes. Monthly rainfall and temperature ranges can generated for each protected area to provide an overview of the local climate.

Further analyses can provide details on the start, peak and end of all detected growing seasons in the PA for 1983 to today. This data can be used to "assess deviations in the vegetation calendar, as an indicator of interannual variation of vegetation status, to forecast the development of climate driven diseases, and to support long term analysis in terms of land cover change - such as a change in vegetation type associated with permanent change in the vegetation calendar." (Phenology Product Sheet) 1.2. Feedback on long term observations include component="comments" page="eStation methods" limit="20" =2. Near real time data acquisition and processing= The data processed by the eStation include mainly rainfall, active fires, small water body presence, a water index (NDWI), and a vegetation index (NDVI

2.1.1. The Vegetation products
The VEGETATION instruments onboard the SPOT satellites allow the production of 10-day global coverage composites at 1 km resolution. In the framework of the FP6 Specific Support Action VGT4Africa, VEGETATION data covering Africa are used to generate a series of products related to the monitoring of vegetation and surface water. The products are broadcasted to the EUMETCast receiving stations installed in Africa, making African land surface monitoring available automatically at no charge for the users. The time series of NDVI data allows identification of changes in vegetation vigor and density in response to bio-physical conditions (including plant type, weather and soil) and human activities. NDWI is rather related to the vegetation water content and presents some advantages in the usually cloudy equatorial areas due to the low sensitivity of the index to atmospheric conditions in comparison to the NDVI.

2.1.2. Feedback on the vegetation products
include component="comments" page="eStation methods" limit="10"

**2.2.1. The Small water bodies products**
The same VEGETATION instrument allows the detection of surface water, thanks to its short wave infrared channel, at the resolution of 1 pixel (1 km). The continuity of surface observation allows the seasonal assessment of water availability. Monitoring the surface water availability means interpreting the temporal sequence of availability of ephemeral and temporary water surfaces detected by the SPOTVEGETATION system. A long time series of data (from 1999 to 2007) has been used to characterize the type of water bodies, i.e. permanent or seasonal as well as their recurrence and the maximal extent of surface water. The seasonal integration characterizes the annual water availability and the date of availability. The small water bodies product consists in the detection of the surface water, at full satellite resolution (1 km) every 10 days, in three classes: free water, humid area and a mixture of both. The free surface water detection has been successfully validated on western Africa with a commission error lower than 2% for free water detection. Former analysis have shown that the system is able to detect temporary water bodies that are often not reference in other data bases, such as GLWD. In addition to the mapping of water pixels in water, the product has also time components, namely the start of replenishment of the small water bodies in the season and the end of availability of these water bodies and the duration of their detection Monitoring surface water scattered in small water bodies provides useful information for several applications, including human activities, cattle management, epidemiology, biodiversity (incl. migrating birds). In addition long term time series of water occurrence in semi arid regions is an interesting indicator of the impact of climate variation. The products should not be considered highly accurate for water body mapping because of resolution. Nevertheless indication about seasonal behaviour of few water bodies in a given region should allow extrapolation of conditions to smaller water bodies.Fire activity and burned areas: mainly monitored in sub-Saharan protected areas, the information is derived from the MODIS fire products available through the NASA funded Fire Information for Resource Management System (FIRMS). Information on fire occurrence is derived four times a day by the MODIS sensors onboard the TERRA and AQUA satellites, at a 1km spatial resolution, while the burned area information has a 500x500 m resolution. From all the datasets, active fires and burned areas, we derive statistics with a 10-day time step (decade) for each protected area in sub-Saharan Africa over a time series which spans from year 2000 to present. The fire-related data and information, updated continuously as soon as new satellite data become available, are provided in support to park managers as well as to researchers exploring land-use and climatic changes. Data and fire statistics for protected areas are accessible through the DOPA.

**2.2.1. Feedback on the Small water bodies products**
include component="comments" page="eStation methods" limit="10"

**2.3.1. The Small water bodies products**
The Land Surface Temperature (LST) is the radiative skin temperature over land. It plays an important role in the physics of land surface as it is involved in the processes of energy and water exchange with the atmosphere. LST values are of special interest for meteorology, hydrology, agrometeorology, climatology and environmental studies. The LST product is distributed at the original MSG repeat cycle (15 minutes), and averaged over 1 day and 10 days periods routinely computed.

**2.3.2. Feedback on the Land Surface Temperature products**
include component="comments" page="eStation methods" limit="10"

**2.4.1. Rainfall products**
Rainfall is one of the most important part of the water resource. Due to financial and infrastructural constraints, rain-gauge and precipitation radar networks are currently extremely sparse over most of Africa. There is a particular shortage of rain-gauge data that is available in real time. For such areas the satellite rainfall estimate is a good alternative to overcome the shortcomings of measurements. Three source of satellite rainfall estimate are used in the framework of this project: EWSNET Rain Fall Estimate (RFE), TAMSAT RFE and the near real-time 15 minutes Multi-sensor Precipitation Estimate from EUMETSAT.Environmental trends and detection of anomalies

**2.4.2. Feedback on the rainfall products**
include component="comments" page="eStation methods" limit="10" =3. Detection of anomalies=

3.1. Detecting changes
Environmental anomalies can be detected in selected areas, like protected areas, by contrasting every 10 days (dekadal) environmental data against historical records. These anomalies can be characterised by their strength, their duration and their deviation from their expected occurrence in time, something typical of seasonal changes. Environmental factors for the current year are compared against the long term average. For each factor, the dekadal values for the previous year are plotted and the graph is updated every 10 days as data for the current year become available. Figure 2: // time series of eStation data and detection of anomalies. The dark gray line on each graph is the average for each dekad based on the available timeseries, and the light grey areas indicate the 95% confidence limits around this average. //

3.2. Characterizing anomalies Every 10 days we collect environmental data for les rainfall, active fires, small water body presence, a water index (NDWI), and a vegetation index (NDVI). These data are further analyzed to detect any significant differences in the current 10 day period compared to same 10 days in the historical record. Any significant differences - which we term 'environmental anomalies' - are flagged as an 'alert'. Alerts are based on three characteristics of the anomalies;
 * The **strength** of the anomaly represents how different it is from the expected value. The bigger the difference, the greater the strength.
 * The **deviation** of the alert tries to determine if the alert may simply be an early or late event such as an early start to the rainy season or a late start to the fire season. We search through the historical data to see if a similar event has ever occurred within a few dekads either side of this date. The larger the time window with no such similar historical event the greater the deviation.
 * The **duration** is simply the number of dekads (cycle of 10 days) over which the anomaly has been observed; 1 dekad, 2 dekads, 3 or more dekads.

3.3. Feedback on detecting anomalies include component="comments" page="eStation methods" limit="10"