MOROCCO EDITION by ASMAE ZBIRI, DOMINIQUE HAESEN and HAMID MAHYOU Using SPIRITS

Transcription

1 MOROCCO EDITION by ASMAE ZBIRI, DOMINIQUE HAESEN and HAMID MAHYOU Using SPIRITS Version :

2 I. INTRODUCTION... 3 II. SPIRITS (Software for the Processing and Interpretation of Remotely sensed Image Time Series)... 3 III. Downlond... 3 IV. Project... 4 V. Application... 4 Ex 1 : Downlond TS : Time series : (MODIS, 8 days, NDVI, North africa, 250 m) Rename : Import / Conversion from Geotiff to ASCII format Adapt HDR Reprojecting Smoothing Extract ROI Rasterize SHP Extract RUM Ex 2 : Downlond TS : emodis, 10 days, NDVI, North africa, 250 m Rastering SHP Extraction ROI by Simple masking scaling/reclassifing Quick look Resampling EX 3: Treatment of METEO DATA Adapt HDR Reproject grille hdr Thinig grille hdr Conversion Meteo data to IMG Validation data EMODIS and ECMWF

3 I. INTRODUCTION For SPIRITS programmers or analysts who need to generalize their programs or improve programming efficiency, Asmae Zbiri and Dominique Haesen thoroughly updates this first edition highly successful of Guide to the SPIRITS Macro Language with an extensive collection of new macro language techniques and examples. The first comprehensive guide to practical Environmental risk modeling. Risk managers who want to stay competitive in today's marketplace need this book to streamline their modeling processes. Despite the high demand for in-house models, this pioneering guidebook is the only complete, focused resource of expert guidance on building and validating accurate, state-of-the-art Environmental risk management models. Written by a proven authorial team with international experience, this hands-on road map takes you from the fundamentals of Environmental risk management to implementing proven strategies in a real-world environment using SPIRITS software. II. SPIRITS (Software for the Processing and Interpretation of Remotely sensed Image Time Series) SPIRITS is a Windows-based software aiming at the analysis of remotely sensed earth observation data. Although it includes a wide range of general purpose functionalities, the focus lies on the processing of time series of images, derived from low resolution sensors such as SPOT- VEGETATION, NOAA-AVHRR, METOP-AVHRR, TERRA-MODIS, ENVISAT-MERIS and MSG-SEVIRI. SPIRITS has been developed by VITO s remote sensing unit on behalf of (and sponsored by) the European Commission s Joint Research Centre (EC-JRC) in Ispra, Italy. The JRC-MARS group (Monitoring Agricultural ResourceS) continuously supplies the EC directorates with agro-statistical information on crop areas and yields for Europe and the major production areas of the world. SPIRITS is a free software environment for analyzing satellite derived image time series in crop and vegetation monitoring. With this toolbox, you can process and examine time series of low and medium resolution sensors. It can be used to perform and to automatize many spatial and temporal processing steps on time series and to extract spatially aggregated statistics. Vegetation indices and their anomalies can be rapidly mapped and statistics can be plotted in seasonal graphs to be shared with analysts and decision makers. III. Downlond 3

4 IV. Project V. Application Ex 1 : 1. Downlond TS : Time series : (MODIS, 8 days, NDVI, North africa, 250 m) Time series : NDVI Datasets available for region: North Africa, MODIS NDVI MOD09 (MODAPS) (C5) 4

5 2. Rename : Specification : =.*. Filename : "africa_north.*.c05.ndvi.mod44.d16.r modaps.v1" - Input pattern: africa_north.*.c05.ndvi.mod44.d16.r modaps.v1* - output pattern: a_%0i Or - in case the filenames have extensions, like "africa_north.*.c05.ndvi.mod44.d16.r modaps.v1.img" and "africa_north.*.c05.ndvi.mod44.d16.r modaps.v1.hdr" - input pattern africa_north.*.c05.ndvi.mod44.d16.r modaps.v1.* - output pattern a_%0i%1 5

6 3. Import / Conversion from Geotiff to ASCII format Original input image (av i) 6

7 4. Adapt HDR We also find the data needed for the map info: Origin = ( , ) Pixel Size = ( , ) 7

8 Corrects the map info: Map info = {arbitrary, 1, 1, , , 250, 250} 5. Reprojecting Using the wkt file we created from the spatial reference set of the original tiff Map info = {arbitrary, 1, 1, 0, e-320, e-315, e-304} Modis tiff From 8

9 Via info button from generic import tool we find the spatial reference set: PROJCS["Sinusoidal_modis_sphere", GEOGCS["GCS_Unknown datum based upon the Authalic Sphere", DATUM["Not_specified_based_on_Authalic_Sphere", SPHEROID["Sphere", ,0]], PRIMEM["Greenwich",0], UNIT["Degree", ]], PROJECTION["Sinusoidal"], PARAMETER["longitude_of_center",20], PARAMETER["false_easting",0], PARAMETER["false_northing",0], UNIT["Meter",1]] Which we can save in a file (e.g. ref.wkt) 9

10 6. Smoothing To generate a dekadal image or eliminate flags we do smoothing Smoothing weekly to dekadal series: Max missing (center) =254 (flags) 1- The operation smoothing weekly to dekadal series was achieved thanks to a single HDR (input identical in the spatial sense: samples, lines, Map info) 2- Note: The operation necessary for the conversion of the series of images from 8 days to 10 days is the compositing (daily, dekad, month, year) 3- The solution is the smoothing weekly to dekadal series we create inputs identical to the spatial sense (hdr). 10

11 7. Extract ROI Original image (with correct map info) => extracted region of interest. 8. Rasterize SHP Rasterize shape. Select attribute corresponding with regions. Selected data type so its values will fit adapt HDR so the minimum and maximum value is known, specify no data value as well. 11

12 Rasterize SHP: attributes - selected ID_1 9. Extract RUM 12

13 Prepare rum database for the data which will be extracted from input images, using rastized shp as regions. in this case without distinuishing land-use classes - just exracting overall mean values for the regions. add sensor to database add variable to database add regions set to database add land use classes set to database 13

14 import regions in regions set from shape file, for ID select same attribute as used for rasterize Create SPU using the rasterized image, no specific land use. Extract RUM from input IMG with this SPU, using specified sensor, variable, regions set and classes set. Now the data (at this point just 1 value) is available in the database. 14

15 Ex 2 : 1. Downlond TS : emodis, 10 days, NDVI, North africa, 250 m Time series : NDVI emodis, North Africa 15

16 2. Rastering SHP 3. Extraction ROI by Simple masking NDVI_data File: HDR ENVI description = {emodis TERRA Normalized Difference Vegetation Index (NDVI). Source: FEWS NET ( samples = 7500 lines = 7045 bands = 1 16

17 Rainfall_data file: int_africa_rain_ hdr ENVI description = {description = {ECMWF rainfall estimate from ECMWF INTERIM model (scaled from original resolution to 0.25 degrees by MeteoConsult)}} samples = 150 lines = 141 bands = 1 header offset = 0 file type = ENVI standard data type = 2 17

18 4. scaling/reclassifing Reclassification Global Land Cover 2000 from 24 to 5 classes 18

19 5. Quick look 19

20 6. Resampling When we have a series of images with different scale as the mask we have to do resampling, and after mask reclassification to have a same scale with TS HDR. Time series resampling Mask Resampling 20

21 EX 3: Treatment of METEO DATA Meteo data from africa, ECMWF TS 1. Adapt HDR Vlo: Lowest significant IMG-DN (value~dt) = 0 Vhi: Highest significant IMG-DN (value~dt) = 250 Vmin: Lowest observaed + signif. DN (Vlo<=Vmin<=Vhi)= 0 Vmax: Highest observaed + signif. DN (Vlo<=Vmax<=Vhi)= 250 Vint : Scaling Intercept : physical Y = Vint+Vslo*DN= 0 Vslo : Scaling Slope : physical Y = Vint+Vslo*DN= 1 ENVI Description = {Morocco ROI extracted from Africa} samples = 150 lines =141 bands = 1 header offset = 0 file type = ENVI standard data type = 2 interleave = bsq byte order = 0 map info = {Geographic Lat/Lon, 1, 1, , , , , WGS-84} values = {Raifall, mm, 0, 250, 0, 28, 0, 1} flags = {0=missing} Date = Days = 10 Sensor type = ECMWF Comment = {Created for Cross-validation of rainfall data Morocco publication} program = {HDRadapt.exe (V1208/1403)} 21

22 Meteo data from Morocco 2. Reproject grille hdr 3. Thinig grille hdr 4. Conversion Meteo data to IMG The operation requires the creation of the metadata file, because the precipitation data of the stations of Morocco in text format do not have a file called metadata (HDR), to do so have projected the digital data with the scale of the shepfile of Steppe ecosystem and a grid of 11 km resolution was developed, this grid is reprejected and then pixelated and used in the conversion. The final operation is carried out with a model for the conversion of meteorological data according to the winter and summer season. TYPE: DATABASE FORMAT TXT DATE: Decadal VARIABLES: Grid, Alt, Lat, long, Decade, Tmax, Tmin, Rain, ET0, RADiation Latitude Longitude Date Precipitation Fields : champs 1 Latitude (decimal degrees) 2 Longitude (decimal degrees) 3 Date (YYYYmmdd) 4 Precipitation (mm) 22

23 etc Rainfall_DATA Temperature_DATA 23

24 Evapotranspiration_DATA Solar radiation_data 24

25 5. Validation data EMODIS and ECMWF Over a period from 2001 to 2012 Validation of these databases is accomplished using the Arcgis 9.2 software. In particular, the Cokrigeage method taking the elevations as secondary variables under consideration provides better results. The success of the method depends mainly on the level of correlation between the climatic parameter and the elevations (Daly et al, 1994). In the standard application of interpolation methods, built-in validation tools (statistics and error surfaces) are available. The control of true estimates should enable us to assess both the validity of each estimate area against each other and the general trends in estimation errors per station (Crane, 2003). The purpose of the validation is to use all the data to estimate the homogeneous and filtered models. The European model provides excellent data it can be assumed. The observed values are those recorded by the rainfall stations and the estimated values are those of the nationally extrapolated database and the ECMWF. To meet the expectations of land degradation studies with respect to climate hazards, remote sensing and modeling have been solutions to solve the problem of data deficiency in arid zones and arid seeding, thus reinforcing the strategy of Adaptation to climate change. The findings of this study open the prospect of defining a new type of data that reveals a good correlation between time series. 25