ASSESSMENT OF DIFFERENT WATER STRESS INDICATORS BASED ON EUMETSAT LSA SAF PRODUCTS FOR DROUGHT MONITORING IN EUROPE G. Sepulcre Canto, A. Singleton, J. Vogt European Commission, DG Joint Research Centre, Institute for Environment and Sustainability, Via E. Fermi 2749, I 21027 Ispra (VA), Italy. Email: guadalupe.sepulcre@jrc.ec.europa.eu ABSTACT This study presents the first steps in evaluating the potential of a selection of EUMETSAT Satellite Application Facility on Land Surface Analysis (LSA SAF) products for drought assessment and monitoring in Europe. The products tested are Evapotranspiration (ET), Land Surface Temperature (LST) and fraction of Absorbed Photosynthetic Active Radiation (fapar). To assess the potential of these products, LSA SAF fapar was compared to the MERIS fapar product already used in the European Drought Observatory (EDO). LST was assessed, as a first step, using anomalies of temperature derived from MODIS imagery. Finally, the LSA SAF ET was used as input for the ratio of ET to potential evapotranspiration (ET 0 ), the latter estimated from the ECMWF interim reanalysis. The results of the comparisons between LSA SAF and MERIS fapar and the potential of LSA SAF fapar for the EDO system, is discussed. Also, the temperature anomalies and the ET/ET 0 results for the drought period of spring/summer 2011 are shown and compared to Standardized Precipitation Index (SPI) maps produced routinely by EDO. INTRODUCTION The European Drought Observatory (EDO) (http://edo.jrc.ec.europa.eu) is an initiative of the European Commission s Joint Research Centre that aims to integrate drought information from the European Member States in order to provide a drought monitoring tool that encompasses continental, national, regional and local scales. Apart from specific indicators provided by different institutions that reflect local conditions, a set of standardized indicators are produced to enable an homogeneous analysis over the entire European continent. These standardized indicators currently include the anomalies of fapar obtained from the MERIS sensor on board ENVISAT (Gobron et al., 2004). Droughts affect the vegetation canopy through plant water stress. fapar is considered a good indicator to assess drought impacts on vegetation because of its sensitivity to water stress (Gobron et al. 2005) that affects the capacity of vegetation to intercept solar radiation and therefore the vegetation growth rate. Water stress is further reflected by an increase of the vegetation canopy temperature due to the reduction of evaporative cooling. This effect is the basis of several water stress indicators using Land Surface Temperature (LST) or Evapotranspiration (ET) providing pre visual information potentially useable for water and agricultural management. 1
The EUMETSAT Satellite Application Facility on Land Surface Analysis (LSA SAF) has developed a series of products including fapar, LST and ET of high value for drought monitoring. The benefits of using LSA SAF products are their spatial coverage (Europe, Africa and South America) and the fact that they are produced operationally in near real time (daily in the case of fapar, every 30 min in the case of ET and every 15 min in the case of LST), which is fundamental for drought early warning. The main purposed of this study is to test these products for drought monitoring in Europe within the EDO framework. METHODOLOGY Three different strategies were followed to test each of the three products. LSA SAF fapar mean values for a window covering Europe were compared with the mean values of MERIS fapar. The comparison was done on a 10 daily basis from 2007 to 2011. The MERIS fapar has previously demonstrated its usefulness to detect drought effects in vegetation (see for example: EDO, 2011). The comparison was also done spatially for the second 10 day period of May 2011 as an example. Due to LSA SAF webpage dissemination problems, LST was evaluated using MODIS brightness temperature (Tb) images as a first step, using the LSA SAF product only to assess some spatial comparisons. MODIS Tb from 2006 to 2011 was used to obtain monthly anomalies using the z score: z = X t σ X where X t is the mean temperature for a month of the current year, X is the long term mean and σ the standard deviation for the same month over the available time series. Finally, the LSA SAF ET product was used, 10 daily accumulated, for obtaining the ratio ET/ET 0 that represents the plant water demand limited by the soil water availability. ET 0 was obtained by applying the FAO Penman Monteith equation to ECMWF ERA Interim air temperature, humidity, radiation, and wind speed data. Images of ET/ ET 0 were obtained for the spring/summer 2011 when central Europe suffered serious drought conditions (EDO, 2011) and also for spring 2012, when some of the countries that suffered drought conditions in 2011, received normal or above normal precipitation amounts. Figure 1 shows images of the meteorological drought index SPI 3 for spring/summer 2011. SPI (McKee et al., 1993) is one of the most widely used drought indicators, being a statistical indicator that compares the total precipitation received at a particular location during a period, in this case 3 months, with the long term precipitation distribution for the same period of time at that location. In EDO a reference period of 30 years (1981 2010) is used. In this case, SPI is obtained from the interpolation of observed meteorological point data. An SPI between 1 and 1 indicates near normal conditions, from 1 to 1.5 moderate drought, from 1.5 to 2 severe drought and if it is lower than 2, extreme drought. 2
Figure1: SPI 3 images of Europe from March to August 2011 RESULTS AND DISCUSSION The temporal comparison of LSA SAF fapar and MERIS fapar can be seen in Figure 2. Mean spatial values of fapar for Europe, obtained from both sensors using different algorithms, were coherent although showed some bias, especially for the higher values. The correlation coefficient obtained for this temporal comparison was 0.8. The same correlation coefficient was obtained when comparing spatially both fapar images for the second 10 day period of May, as an example. These results show the potential of LSA SAF fapar as a basis to calculate fapar anomalies for drought analysis; however the temporal series available is still too short to adequately characterize normal conditions. Figure2: Temporal evolution of mean values of fapar for the Europe window. 3
This is also the case for the Land Surface Temperature product. Nevertheless, Figure 3 shows, as an example, the good reflection of drought conditions in the brightness temperature anomalies obtained from MODIS imagery for the spring/summer 2011 period. The figure shows qualitatively the clear correspondence between the temperature anomaly values and the SPI values (cf Figure 1) for the same period. When comparing with fapar anomalies for the same period (EDO, 2011) brightness temperature anomalies also respond more quickly to drought conditions. The result of comparing spatially LSA SAF Land Surface Temperature (LST) images with MODIS Brightness Temperature (Tb) images demonstrated their spatial coherence even if the physical meaning of the products is different (data not shown here). Future research using the LSA SAF LST will explore the diurnal cycle of LST. Figure 3: Anomalies of brightness temperature over Europe, obtained from MODIS imagery from March to August 2011 Despite the usefulness of fapar and LST for drought monitoring, the fact that drought and water stress are not the only factors that can cause a decrease of fapar or an increase of LST values/anomalies must be taken into account. In the case of LST, general meteorological conditions (air temperature, wind speed, solar radiation.) or changes in land cover can have significant effects. In contrast, ET is a parameter that integrates environmental factors like meteorological conditions and soil moisture status. The results of obtaining the ratio ET/ET o, using the LSA SAF ET product for Europe from April to the end of July 2011 are shown in Figure 4. These images show good correspondence between the areas where SPI was lower than 1 or 2 indicating moderate or extreme drought and the areas with a value of the ratio lower than 0.6 or 0.4. A particular example for France can be seen in Figure 5, where a comparison with images of the same periods in 2011 and 2012 is shown. During 2012 the country received a normal quantity of precipitation. This example shows clearly how the ratio ET/ET o reflects the effects of the drought conditions in 2011. 4
Figure 4: ET/ET o images obtained from April to July 2011. Despite this good correspondence between ET/ET o and drought conditions in areas where there is a high percentage of vegetation cover, the answer obtained for areas where the vegetation cover is low or not in the growing period is different. This can be seen in Figure 4, for example in Spain, where the values of ratio ET/ET o obtained from June were lower than 0.4 even though the country received normal rain quantities (Figure 1). Due to the absence of time series of data long enough to characterize the normal conditions, a solution to avoid confusion in these areas could be filtering by taking into account the beginning and the end of the growing period or by using the percentage of vegetation cover as a basis. 5
(a) (b) Figure 5: ET/ET o images of France, obtained: a) from the second 10 days of May to the first 10 days of June 2011, b) from the second 10 days of May to the first 10 days of June 2012. CONCLUSIONS Three products available in LSA SAF were tested in the framework of the European Drought Observatory in order to show their potential for drought monitoring. However, drought characterization (in particular, the calculation of anomalies) is dependent on the normal conditions of the area under study, which require a homogeneous time series of data, long enough to adequately characterize natural variability of the area where the index is calculated. This time series is not yet ready in LSA SAF. In the absence of time series of data long enough to characterize the normal conditions of the area, ET/ETo gives relevant information for drought monitoring except in areas of low vegetation cover. Filtering to take into account the beginning and the end of the growing period or the percentage of vegetation cover is being investigated as a possible solution. Future research will also take into account the LST diurnal cycle and other EUMETSAT products such as H SAF products of soil moisture and precipitation. REFERENCES European Drought Observatory (EDO), Drought News in Europe: Situation in May 2011, available at: http://edo.jrc.ec.europa.eu/documents/news/edodroughtnews201105.pdf, last access: 17 September 2012. Gobron, N., Aussedat, O., Pinty, B., Taberner, M., and Verstraete, M. M.: Medium Resolution Imaging Spectrometer (MERIS) An optimized FAPAR Algorithm Theoretical Basis Document, EUR Report No. 21386, 2004. 6
Gobron, N., Pinty, B., M elin, F., Taberner, M., Verstraete, M. M., Belward, A., Lavergne, T., and Widlowski J. L.: The state of vegetation in Europe following the 2003 drought, Int. J. Remote Sens., 26, 2013 2020, 2005. McKee, T. B., Doesken, N. J., and Kleist, J: The relationship of drought frequency and duration to time scales. Proceedings of the 8th Conference of Applied Climatology, Anaheim, CA, Am. Meterol. Soc., 179 184, 1993. 7