Update on Water Scarcity and Droughts indicator development

Status box Title: Update on Water Scarcity and Droughts indicator development Authors: XXX, DG ENV Background: Within the 2010-2012 CIS period, an updated mandate on water scarcity and d

Status box

Title: Update on Water Scarcity and Droughts indicator development

Author(s): XXX, DG ENV

Background:

Within the 2010-2012 CIS period, an updated mandate on water scarcity and droughts

was approved, requesting to deliver a set of common indicators for both water scarcity (influenced by human activity) and drought (natural).

At the Water Directors meeting in May 2011 a set of 7 awareness raising indicators was

agreed for testing These indicators provide, in combination, an overview of thedevelopments as regards water scarcity and droughts and will allow distinguishingbetween the natural and man-made phenomena

At the Water Directors meeting in November 2011, two agreed indicators (SPI andFAPAR) were endorsed to illustrate drought events as elements of the future waterscarcity and drought indicator system Further technical drafting and testing of theremaining drought indicators as well as in particular of the water scarcity indicator (WEI+)was encouraged

The testing exercise and further conceptual developments have led to the agreement bythe Expert Network to include the Water Scarcity indicator WEI+ in the indicator system

The three indicators agreed so far can extensively be calculated on the basis of pan-Europeaninformation, either already existing or under development (water accounts)

The three awareness-raising indicators agreed so far can extensively be calculated on the basis ofpan-European information, either already existing or under development (e.g water accounts),though their testing and reality-check with MS is recommended in order to increasingly improvetheir robustness

In parallel, specific management already exist in several MS, which are taking into account thelocal climate, water systems and existing monitoring more fully

2. FULLY DEVELOPED DROUGHT INDICATORS

The following three indicators have been agreed so far by the Expert Group (revised factsheets areattached):

1

 The Standardized Precipitation Index (SPI) is an indicator to detect and quantify

meteorological drought situations by comparing the current situation to historical records Itwill be hosted by the European Drought Observatory (EDO) This indicator can producedifferent time-related outputs, so meteorological drought evidence and evolution can beshown for the past month(s), season(s) and/or year(s), facilitating the establishment of links

to other drought indicators

 The selected Vegetation Response indicator is the fraction of Absorbed Photosynthetically Active Solar Radiation (fAPAR) It represents the fraction of the solar

energy which is absorbed by the vegetation canopy and is a biophysical variable directlycorrelated with the primary productivity of the vegetation Its anomalies (the deviation fromthe long-term mean for a certain period of time) are considered a good indicator to detectand assess drought impacts on vegetation canopies

 The Water Exploitation Index Plus (WEI+) of a particular area is the total consumption of

water divided by the renewable freshwater resources' It provides an indication of thepressure on the water resources of a certain territory as a consequence of waterwithdrawals Hence, it also identifies areas most prone to suffer recurrent or permanentsituations of water scarcity

In relation to meeting the goal of awareness raising on situations of drought and water scarcity andthe establishment of effective mechanisms for its detection and analysis as a starting point for theimplementation of response measures against these situations, it should be noted that:

- Drought indicators provide essentially information on climate, and indirectly on theavailability of resources Negative results or anomalies may indicate different potentialimpacts and reveal the occurrence of dry periods of varying intensity and duration whichmay require the implementation of actions in the framework of the Drought riskmanagement plans

- Moreover, the observation of the recurrent presence of such anomalies over time in a giventerritory would indicate the need to design appropriate policy measures in response to thefrequency of the phenomenon observed

- Water scarcity indicators provide information about the pressure on freshwater resources in

a given area and its vulnerability to the appearance of water-scarce periods and/orpotential increases in water use that may endanger the sustainability of the water supplysystems as well as maintaining the environmental conditions required for the maintenance

of water dependent ecosystems This can happen at an annual or monthly/seasonal level

- As in the previous case, there may be punctual situations of scarcity that require the launch

of management actions laid down in the Drought risk management plans, but also theobservation of negative developments that indicate the need to anticipate policy responsemeasures

3 NEXT STEPS

Further work might also be useful for other water scarcity indicators, e.g a Water Demand Index orthe comparison of observed outflows with natural outflows, but this initiative falls outside the currentscope of this EG Mandate 2010-2012 and should be evaluated after a thorough assessment of theindicator set and its data availability, reliability and sustainability

A number of further indicators have been discussed but not been agreed so far by the ExpertGroup1: Snowpack, Standardised Run-off Index (SRI), Groundwater and Soil Moisture Inparticular:

1 The SRI, Groundwater and Soil Moisture indicators are also influenced by water uses, and hence not pure drought indicators

2

- Regarding the snowpack indicator, FI will continue leading its development, in particularregarding the quantification of the water equivalence of a certain snowpack, with the aim todevelop and test the indicator for the next EG meeting.

- Regarding the streamflow indicator (SRI), ES will lead a testing exercise and integratecomments and experiences into the factsheet from those MS that have tested or will testthe indicator This development is expected to conclude by the next EG meeting

- Regarding the groundwater indicator, FR will lead testing and improvement of the indicatorfactsheet, involving in particular those countries that are already using groundwaterevolution as an indicator for awareness raising This activity is planned to be developed forthe next EG meeting

- Regarding soil moisture, it is still considered as a useful indicator, but further developmentsand clarifications have to be checked for the validity of the data An update is expected forthe next EG Meeting, lead by ES and JRC

- Regarding the WEI+, the EG is looking forward to validate the data from the wateraccounts project as well as the results of comparing the data with the current WEIthresholds with a view to updating them for use of the WEI+

- As previously agreed the EG will be sharing best practices as regards managementindicators in the remaining part of the period covered by the mandate

4. ADDITIONAL WORK OF THE INDICATOR EXPERT GROUP

In parallel to the development of the indicators, the Expert Network has developed a paper with

Working Definitions of Water Scarcity and Droughts in follow up to the assessment of water

scarcity & droughts in the RBM plans where it became apparent that these phenomena are notinterpreted in the same way by the different RB authorities The definitions paper is now agreed bythe Expert Group and attached to this paper

Furthermore the EG has initiated work on Environmental flows as a tool to better understand the

relationship between water scarcity, droughts and achieving the objectives of the Water FrameworkDirective A paper is currently being finalized and given that the relevance of environmental flows ismuch wider than the link to water scarcity & droughts the EG suggests that that paper could besubmitted and discussed more widely in the CIS structure

5. RECOMMENDATIONS TO WATER DIRECTORS:

The EG requests WDs to

 endorse the WEI+ indicator as part of the overall indicator set for water scarcity and drought,with the understanding that thresholds still need to be tested and agreed

 Take note of the Working Definitions on Water Scarcity & Droughts (attached)

 agree to the paper on Environmental Flows (draft attached) to be circulated to other relevantgroups in the CIS, in particular WG A & C and the EG Climate Change and Agriculture

3

Relevance of the indicator to drought

The Fraction of Absorbed Photosynthetically Active Radiation (fAPAR) represents the fraction ofthe solar energy which is absorbed by the vegetation canopy fAPAR is a biophysical variabledirectly correlated with the primary productivity of the vegetation, since the intercepted PAR is theenergy (carried by photons) underlying the biochemical productivity processes of plants fAPAR isone of the Essential Climate Variables recognized by the UN Global Climate Observing System(GCOS) and by the FAO Global Terrestrial Observing System (GTOS) as of great potential tocharacterize the climate of the Earth

Due to its sensitivity to vegetation stress, fAPAR has been proposed as a drought indicator

(Gobron et al 2005 and 2007) Indeed droughts can cause a reduction in the vegetation growth

rate, which is affected by changes either in the solar interception of the plant or in the light useefficiency

Policy relevance

Water Framework Directive WFD (Directive 2000/60/EC of the European Parliament and of theCouncil of 23 October 2000 establishing a framework for Community action in the field of waterpolicy)

- Environmental objectives: Exemption for temporary deterioration in the status (Art 4 (6))

- Programme of measures: Additional measures are not practicable (Art 11 (5))

Communication of the EC to the Council and European Parliament: “Addressing the challenge ofwater scarcity and droughts in the European Union” (published on July 2007)

- Developing drought risk management plans (Paragraph 2.3.1.)

- Developing an observatory and an early warning system on droughts (Paragraph 2.3.2).

- Improve knowledge and data collection: Water scarcity and drought information system

4

a Detailed methodology for the calculation of the indicator

fAPAR is difficult to measure directly but can be inferred from models describing the transfer ofsolar radiation in plant canopies, using Earth Observation (EO) information as input data fAPARestimates are retrieved using EO information by numerically inverting physically-based models.The fAPAR estimates used within EDO are operationally produced by the European SpaceAgency (ESA) They are derived from the multispectral images acquired by the MediumResolution Imaging Spectrometer (MERIS) onboard ENVISAT by means of the MERIS Global

Vegetation Index (MGVI) algorithm, developed at the JRC (Gobron et al 2004)

MGVI is a physically-based index which transforms the calibrated multi-spectral directionalreflectance into a single numerical value while minimizing possible disturbing factors It isconstrained by means of an optimization procedure to provide an estimate of the fAPAR of avegetation canopy The objective of the algorithm is to reach the maximum sensitivity to thepresence and changes in healthy live green vegetation while at the same time minimizing thesensitivity to atmospheric scattering and absorption effects, to soil color and brightness effects,and to temporal and spatial variations in the geometry of illumination and observation

The MGVI level-3 aggregation processor is routinely operated on the ESA Grid Processing onDemand (G-POD) system The algorithms have been developed and are maintained by theEuropean Commission Joint Research Centre (JRC) More information on the algorithms can be

found in Pinty B et al (2002) and Gobron N et al (2004)

 Data acquisition: The 10-day fAPAR estimates are regularly produced by the EuropeanSpace Agency (ESA) as MERIS fAPAR Level-3 Aggregated Products following the approach

by Aussedat et al (2007)

 Anomaly estimation: fAPAR anomalies are produced at JRC as follows:

where X t is the fAPAR of the 10-day period t of the current year and , is the mean fAPAR and δ is the standard deviation calculated for the same 10-day period t using the available

time series (archive)

b Reference period for calculating the Statistics

For the production of fAPAR anomalies, JRC extended the MERIS fAPAR time series (rangingfrom end 2002 to the current day) with fAPAR estimations obtained from the Sea-viewing WideField-of-view Sensor (SeaWiFS) from mid 1997 to end 2002 with an algorithm completely

compatible with the MGVI (Gobron et al 2002) This allowed for the creation of a reference

period starting mid 1997

5

5 Data source and frequency of data collection

MGVI data are delivered as a subscription service within the Service Support Environment (SSE)

of the European Space Agency This service is called "MGVI Catalogue Search and Download" and can be accessed via this link: http://services.eoportal.org/portal/service/

ShowServiceInfo.do?serviceId=7180CB90&categoryId=89802980

Frequency of data collection: every 10 days

6 Quality Information

a Strength & weaknesses at data level

[+] Every ten days, the MGVI gives a spatially continuous picture of the vegetation status/health

at a high spatial resolution (~1km) for the entire European continent

[-] Drought and water stress are not the only factors that can cause a decrease of MGVIvalues/anomalies Changes in land cover or pests and diseases can also be responsible for suchvariation of the signal Therefore this indicator must be used jointly with other indicators givinginformation on rainfall and soil moisture deficits in order to determine if the variation in thevegetation response (signal) is linked with a drought event or not

[-] Anomalies are dependent of the time series (reference period) available to calculate the term average and the standard deviations This period should be long enough to characterize thearea where the index is calculated As the reference period is still short (1997 to today), theanomalies are to be interpreted with care

long-b Performance of the indicator

MGVI has been used successfully to assess the impact of the 2003 drought on plant productivity

in Europe (Gobron et al 2005) Moreover Rossi et al (2008) highlighted the potential of this

indicator for drought detection and monitoring by comparing it to other drought indicators such asthe Standardized Precipitation Index (SPI) Recently, fAPAR has been capable to reflect theimpact of the 2011 spring drought in western Europe (mainly France, Germany, Benelux, UK) andcentral Europe (http://desert.jrc.ec.europa.eu/action/php/index.php?action=view&id=118)

Products

fAPAR and fAPAR anomalies can be presented in the form of maps and graphs, providinginformation both on the spatial distribution of the vegetation activity and the temporal evolutionover longer time periods Gridded data can easily be aggregated over administrative or naturalentities such as river basins This allows for the qualitative and quantitative comparison of theintensity and duration of the fAPAR anomalies with recorded impacts such as yield reductions,low flows, or the lowering of groundwater levels, for example

The Map Server of the European Drought Observatory (EDO) displays the latest available fAPAR10-day composite image and the fAPAR anomaly image calculated by comparing this image tothe historical series in the same 10-day period (Figure 1) In the future historical images will beretrievable

The fAPAR product is dimensionless It is ranging from 0 to 1 in terms of real values, with 1corresponding to a maximum of vegetation activity On the map these values are represented byyellow to green colours

The fAPAR anomaly product is given in units of standard deviation Anomalies are commonlyranging from -4 to +4, represented by colours ranging from red to green, red showing negativeanomalies

6

Both products are easy to read The interpretation must take into account the fact that thisindicator is showing a variation in the vegetation health and/or cover This variation can be aconsequence of a rainfall / soil moisture deficit but can also be due to other factors.

Figure 1: fAPAR (left) and fAPAR anomaly (right) images for the ten-day period of 1-10 July

2011, as shown on the EDO Map Server

Assessment of capacity for drought monitoring

Gobron et al (2000, 2006a, 2006b) present an evaluation for different canopy radiation transfer

regimes using the current fAPAR products derived from SeaWiFS against ground-basedestimations

The MGVI/fAPAR algorithm seems to perform better than NDVI as a drought indicator, as NDVIshows a much larger temporal variability than fAPAR, and does not always allow for the detection

of droughts (Gobron et al 2007).

References

Aussedat O., Taberner M., Gobron N., and Pinty B (2007) MERIS Level 3 Land SurfaceAggregated Products Description Institute for Environment and Sustainability, EUR Report

No 22643 EN, 16pp

Gobron, N., Pinty, B., Verstraete, M.M., and Widlowski, J.-L (2000) Advanced Vegetation Indices

Optimized for Up-Coming Sensors: Design, Performance and Applications IEEE Transactions

on Geoscience and Remote Sensing, 38: 2489-2505.

Gobron N., Pinty B., Mélin F., Taberner M., and Verstraete M M (2002) Sea Wide Field-of-ViewSensor (SeaWiFS) - Level 2 Land Surface Products - Algorithm Theoretical Basis Document.Institute for Environment and Sustainability, EUR Report No 20144 EN, 23 p

Gobron N., Aussedat O., Pinty B., Taberner M., and Verstraete M M (2004) Medium ResolutionImaging Spectrometer (MERIS) - An optimized FAPAR Algorithm - Theoretical BasisDocument Revision 3.0 Institute for Environment and Sustainability, EUR Report No 21386

EN, 20 p

http://fapar.jrc.ec.europa.eu/WWW/Data/Pages/FAPAR_Home/FAPAR_Home_Publications/atbd_meris_v4_gen.pdf

Gobron N., Pinty B., Mélin F., Taberner M., Verstraete M.M., Belward A., Lavergne T., andWidlowski J.-L (2005) The state vegetation in Europe following the 2003 drought

International Journal Remote Sensing Letters, 26 (9): 2013-2020.

Gobron, N., Aussedat, O., Pinty, B., Robustelli, M., Taberner, M., Lavergne, T (2006a) TechnicalAssistance for the Validation the ENVISAT MGVI Geophysical Product Final Report EURReport 22246 EN, European Commission - DG Joint Research Centre, Institute for

7

Environment and Sustainability, 103pp.

Gobron, N., Pinty, B., Aussedat, O., Chen, J M., Cohen, W B., Fensholt, R., Gond, V., Lavergne,T., Mélin, F., Privette, J L., Sandholt, I., Taberner, M., Turner, D P., Verstraete, M M.,Widlowski, J.-L (2006b) Evaluation of Fraction of Absorbed Photosynthetically ActiveRadiation Products for Different Canopy Radiation Transfer Regimes: Methodology andResults Using Joint Research Center Products Derived from SeaWiFS Against Ground-Based

Estimations Journal of Geophysical Research – Atmospheres, 111(13), D13110.

Gobron, N., Pinty, B., Mélin, F., Taberner, M., Verstraete, M.M., Robustelli, M., Widlowski, J.-L

(2007) Evaluation of the MERIS/ENVISAT fAPAR Product Advances in Space Research 39:

105-115

Pinty B., Gobron N., Mélin F., and Verstraete M.M (2002) Time Composite Algorithm TheoreticalBasis Document Institute for Environment and Sustainability, EUR Report No 20150 EN, 8pp

http://fapar.jrc.ec.europa.eu/WWW/Data/Pages/FAPAR_Home/FAPAR_Home_Publications/pinty_etal_ies_2002.pdf

Rossi, S., Weissteiner, C., Laguardia, G., Kurnik, B., Robustelli, M., Niemeyer, S., and Gobron, N

2008 “Potential of MERIS fAPAR for drought detection”, in Lacoste, H., and Ouwehand, L.(eds.), “Proceedings of the 2nd MERIS/(A)ATSR User Workshop”, 22–26 September 2008,Frascati Italy (ESA SP-666), ESA Communication Production Office, November 2008

http://envisat.esa.int/pub/ESA_DOC/meris_workshop_2008/papers%20/p63_rossi.pdf

8

ANNEX 2

SPI: Standardized Precipitation Index

Indicator definition

The Standardized Precipitation Index (SPI-n) is a statistical indicator comparing the total

precipitation received during a period of n months with the long-term rainfall distribution for the same period of time SPI is calculated on a monthly basis for a moving window of n months, where n indicates the rainfall accumulation period, which is typically of 1, 3, 6, 9, 12, 24 and 48

months The corresponding SPIs are denoted as SPI-1, SPI-3, SPI-6, etc

In order to allow for the statistical comparison of wetter and drier climates, SPI is based on atransformation of the accumulated precipitation into a standard normal variable with zero mean

and variance equal to one SPI results are given in units of standard deviation from the

long-term mean of the standardized distribution

In 2010 WMO selected the SPI as a key meteorological drought indicator to be producedoperationally by meteorological services

Relevance of the Indicator to Drought

A reduction in precipitation with respect to the normal precipitation amount is the primary driver

of drought, resulting in a successive shortage of water for different natural and human needs.Since SPI values are given in units of standard deviation from the standardised mean, negativevalues correspond to drier periods than normal and positive values correspond to wetter periods

than normal The magnitude of the departure from the mean is a probabilistic measure of the severity of a wet or dry event

Since the SPI can be calculated over different rainfall accumulation periods, different SPIs allowfor estimating different potential impacts of a meteorological drought:

- SPIs for short accumulation periods (e.g., SPI-1 to SPI-3) are indicators for immediate impacts such as reduced soil moisture, snowpack, and flow in smaller creeks;

- SPIs for medium accumulation periods (e.g., SPI-3 to SPI-12) are indicators for reduced stream flow and reservoir storage; and

- SPIs for long accumulation periods (SPI-12 to SPI-48) are indicators for reduced reservoir and groundwater recharge, for example

The exact relationship between accumulation period and impact depends on the naturalenvironment (e.g., geology, soils) and the human interference (e.g., existence of irrigationschemes) In order to get a full picture of the potential impacts of a drought, SPIs of differentaccumulation periods should be calculated and compared A comparison with other droughtindicators is needed to evaluate actual impacts on the vegetation cover and different economicsectors

Policy Relevance

Water Framework Directive WFD (Directive 2000/60/EC of the European Parliament and of theCouncil of 23 October 2000 establishing a framework for Community action in the field of waterpolicy)

- Environmental objectives: Exemption for temporary deterioration in the status (Art 4 (6))

- Programme of measures: Additional measures are not practicable (Art 11 (5))

9

Communication of the EC to the Council and European Parliament: “Addressing the challenge ofwater scarcity and droughts in the European Union” (published on July 2007)

- Improving drought risk management

 Developing drought risk management plans (Paragraph 2.3.1.)

 Developing an observatory and an early warning system on droughts (Paragraph 2.3.2.)

 Use of the EU Solidarity Fund (Paragraph 2.3.3.)

- Improve knowledge and data collection: Water scarcity and drought information system

SPI Values Category Cumulative Probability

Probability of Event [%]

SPI ≥ 2.00 Extreme wet 0.977 – 1.000 2.3%

1.50 < SPI ≤ 2.00 Severely wet 0.933 – 0.977 4.4%

1.00 < SPI ≤ 1.50 Moderately wet 0.841 – 0.933 9.2%

-1.00 < SPI ≤ 1.00 Near normal 0.159 – 0.841 68.2%

-1.50 < SPI ≤ -1.00 Moderately dry 0.067 – 0.159 9.2%

-2.00 < SPI ≤ -1.50 Severely dry 0.023 – 0.067 4.4%

SPI < -2.00 Extremely dry 0.000 – 0.023 2.3%

The SPI can be computed for any timescale of interest to the user, reflecting potential impactsfrom the agricultural, hydrological and water supply management perspectives (see above)

3 Temporal Scale

Typically of 1-, 3-, 6-, 12-, 24-, 48-months, or even longer time-periods depending on thepotential impact and regional characteristics For statistical reasons a minimum rainfallaccumulation period of one month is recommended SPI is then calculated for every month with

a moving window of the selected time-length

4 Methodology

10

a Detailed Methodology for the calculation of the indicator

Computation of the SPI involves fitting a probability density function to a given frequencydistribution of precipitation totals for a station or grid point and for an accumulation period.Typically the gamma probability density function is used, but others, such as the Pearson-IIIprobability density function may also be used (see below) The statistics for the frequencydistribution are calculated on the basis of a reference period of at least 30 years (see point 4.b).The parameters of the probability density function are then used to find the cumulativeprobability of an observed precipitation event for the required month and temporal scale Thiscumulative probability is then transformed to the standardised normal distribution with mean zeroand variance one, which results in the value of the SPI The procedure of transforming theobserved rainfall via the cumulative distribution functions (CDFs) of the Gamma distribution andthe standardised normal variable to the SPI is illustrated in Figure 1:

Figure 1: Transformation of the observed rainfall via the Gamma cumulative distribution

function (CDF) and the CDF of the standardized normal variable to the SPI.

Two statistical distributions have been shown to perform equally well in estimating the SPI: theGamma and the Pearson-III distributions The Gamma distribution has been adopted by mostcentres around the world as a model from which to compute SPI It is described by only twoparameters, but offers considerable flexibility in describing the shape of the distribution, from anexponential to a Gaussian form It has the advantage that it is bounded on the left at zero andtherefore excludes the possibility of negative precipitation Additionally, it is positively skewedwith an extended tail to the right, which is especially important for dry areas with low mean and ahigh variability in precipitation

However, for some stations and SPI timescales, distributions such as the beta, Pearson-III ornormal distributions may fit the observations better Ultimately the requirement is to obtain thebest estimate of the probability of the observed precipitation to be transformed to the standard

normal distribution for SPI It therefore follows that the best estimate of SPI will come from the distribution that best fits the station data

As a general rule, it is recommended to use the Gamma distribution as the standard model If, however, a satisfactory fit cannot be achieved, alternative distributions should

compute the SPI using the R statistics package is provided in Annex A.

b Reference period for calculating the Statistics

The World Meteorological Organization (WMO) recommends that precipitation totals for the atleast 30 years are used as reference time-line for calculating rainfall statistics Severalinvestigators recommend calculating the statistics for the SPI from even longer time periods(e.g., 50 or more years) in order to ensure an accurate representation of extreme events Thecurrent WMO standard reference period is 1961-1990 However several Member States usemore recent periods (e.g., 1971-2000, 1981-2010) in order to accommodate changes in theprecipitation regime due to climate change and to compare actual rainfall figures to a morerecent situation In order to ensure comparability of the results across Europe and across scales

it is highly recommended to use a common reference period for the calculation of the SPI.Considering the results of an inventory of the reference periods used in various Member States,the specific needs for accurately representing extreme events, and possible changes in therainfall regimes due to climate change, the Water Scarcity and Drought Expert Group strongly

recommends using the period January 1971 to December 2010 as Reference Period for the

calculation of the SPI

In the case that a lack of data would significantly restrict the number of rainfall stations to be

used, a shorter reference period may be used (e.g., 1981-2010) However, in all cases, the Reference Period used should be clearly indicated with all data presented

5 Data source and frequency of data collection

Station precipitation data can be taken from high resolution national rain gauge networks or fromlower resolution WMO SYNOP stations Data should be available with at least a monthly timestep Historical time series should be homogeneous and as complete as possible WMOrecommends a maximum of 5 years of missing data (maximum 3 years in a row) over a 30 yearsperiod

As a further check on the data, it is recommended that a threshold of 90% completeness beapplied for all SPI timescales and months For example, for the SPI-1 for May, there must be atleast 0.9*31=28 days with observations; and for the SPI-3 for May, there must be at least0.9*(31+30+31)=83 days with data Furthermore, in order to have enough data to fit adistribution for the reference period, 90% of the years must have valid data for the month andSPI timescale For example, for the SPI-3 for May and a 30-year reference period, 0.9*30=27years must have 3-month accumulations for May with at least 83 days of observations

For data visualization, it is necessary to interpolate the SPI to a grid The resolution of the gridshould be chosen based on the density of the stations No advantage is gained frominterpolating sparse station data to a high resolution grid, and such an approach may result inmisleading information Since the SPI is a measure of the deviation from normal at a particulartime and location, the localised effects of factors such as topographic barriers, sea breezes, landsurface type etc on rainfall are already included in the measure Therefore the simplest

interpolation methods are preferred Recommended methods are the basic inverse distance method or the Cressman/Barnes method, which uses a successive correction technique to

take advantage of higher station densities where they exist (see Xia et al 1999, Barnes 1973,

Cressman 1959) For details of the implementation of the Cressman/Barnes method see, forexample, http://www.atmos.millersville.edu/~lead/Obs_Data_Assimilation.html)

Gridded data interpolated from rain gauges at 1o resolution are available from GPCC1 GPCC

“First Guess” data are available monthly on the 5th of each month, and final analyses areavailable through the GPCC monthly monitoring product 2 months after the end of the month Gridded data at 2.5° spatial resolution, interpolated from rain gauges and satellite remotesensing observations are available from GPCP2 monthly, typically 2 months after the end of eachmonth

6 Quality Information

12