AMATERASS JICA Training

AMATERASSSatellite Solar RadiationData Analysis

This training material reconstructs the JICA Training conducted from 2023 to 2025 under the JST–JICA SATREPS framework.

FIELD: METEOSAT-IODC fld.3km / 0.03° / 15 min
AMATERASS METEOSAT-IODC surface downward shortwave flux (1 May 2026, 09:12 UTC)
AMATERASS METEOSAT-IODC surface downward shortwave flux (1 May 2026, 09:12 UTC)

About this training material

This material reconstructs the JICA Training conducted from 2023 to 2025 under the JST–JICA SATREPS framework.

1

Review the tasks

First, review what will be processed and verified in the DAY 1–3 curriculum.

2

Analyze real data

Use real AMATERASS data to perform spatial subsetting, temporal averaging, and point time-series extraction.

3

Verify the results

Visualize the results with Tmap and gnuplot and verify that the numerical processing was performed correctly.

The three days cover spatial, temporal, and point-based analysis

DAY 1Extract a study region
DAY 2Calculate daily and monthly means
DAY 3Extract point time series
PREPARATION / SYSTEM CHECK

Analysis environment and data

TRAINING GOALPrepare an environment in which Java, GNU Wget 1.x, bzip2, GNU date, gnuplot, and Tmap are available.
4000 × 4000METEOSAT-IODC grid
0.03°Latitude/longitude grid spacing
15 minProduct time interval
Float32 / BEIEEE 754 / Big-endian
0

Check the execution environment

Check each command used in the training. Rather than hiding these checks in a script, run the actual commands directly.

Create the working directory
mkdir -p ~/amaterass_jica
cd ~/amaterass_jica
Check Java, bzip2, date, and gnuplot
java -version
javac -version
bzip2 --version
date --version
gnuplot --version
Check GNU Wget 1.x for FTP access
wget1 --version
wget --version
GNU Wget: GNU Wget 1.x is required because METEOSAT-IODC data are retrieved via FTP. Each download script uses wget1 when available; otherwise it uses wget only if it is GNU Wget 1.x rather than Wget2.
1

Prepare Tmap

Tmap is used to visualize AMATERASS data as maps. Download and compile Tmap, then run it directly in each DAY while checking the required arguments.

Download and compile Tmap
cd ~/amaterass_jica
wget -c https://downloads.amaterass.science/tmap_v3.0cc_20260815d.tar.gz
tar xzf tmap_v3.0cc_20260815d.tar.gz
cd tmap
javac *.java
✓
CHECK
Run ls *.class. The Tmap setup is complete if tmap.class and the other class files are listed.
2

Data used in this training

The representative scene is 1 May 2026 at 09:12 UTC. DAY 2 uses data from May 2026 to calculate the monthly mean.

Time: The first 12 digits of each filename indicate UTC. METEOSAT-IODC does not provide a per-pixel actual-observation-time file like the HIMAWARI Japan-area product, so this training uses the UTC reference time in the filename.
DAY 1 / SPATIAL SUBSET

Spatial subsetting from wide-area data

MISSION GOALRead the METEOSAT-IODC solar-radiation data for 1 May 2026 at 09:12 UTC and use latitude and longitude to extract a study region over Central Asia. Use Tmap to visualize the full domain and verify the subset.
STEP 1View the full domain
STEP 2Use LAT/LNG
STEP 3Extract the region
STEP 4Verify with Tmap
1

Solar-radiation distribution over the full IODC domain

First, view the full 4000 × 4000 grid and identify where the study region lies within the full domain. A black rectangle is overlaid here to make the subset extent clear.

Prepare the DAY 1 input data
wget -O download_day1.sh https://amaterass.science/jica-training/download_day1.sh
chmod +x download_day1.sh
./download_day1.sh
~/amaterass_jica/data/
├── reference/
│   ├── standard_4000x4000.iodc.42.5.lat.fld.3km.bin
│   └── standard_4000x4000.iodc.42.5.lng.fld.3km.bin
└── 20260501/
    └── 202605010912.dwn.sw.flx.sfc.fld.3km.bin
METEOSAT-IODC surface downward shortwave flux and Central Asia subset extent (1 May 2026, 09:12 UTC)
Surface downward shortwave flux over the full METEOSAT-IODC domain (1 May 2026, 09:12 UTC). The region outlined in black is extracted in DAY 1.
✓
CHECK
Confirm the solar-radiation distribution over the Indian Ocean domain and the location of Central Asia.
2

Define the study region by latitude and longitude

Set the upper-left corner to 47.02°N, 51.0°E and the lower-right corner to 34.02°N, 73.0°E, then determine the corresponding rows and columns from the LAT/LNG grids.

Full IODC domain4000 × 4000
→
LAT / LNGLocate target coordinates
→
Study region47.02–34.02°N
51.0–73.0°E
3

Extract the study region

Apply the identified rows and columns to the solar-radiation grid and write only the selected region as a Big-endian Float32 binary file.

Download the analysis program
wget -O analysis_iodc_day1.tar.gz https://amaterass.science/jica-training/analysis_iodc_day1.tar.gz
tar xzf analysis_iodc_day1.tar.gz
javac analysis.iodc.day1/*.java
Run
java -cp analysis.iodc.day1 IODCCrop \
  data/20260501/202605010912.dwn.sw.flx.sfc.fld.3km.bin \
  data/reference/standard_4000x4000.iodc.42.5.lat.fld.3km.bin \
  data/reference/standard_4000x4000.iodc.42.5.lng.fld.3km.bin \
  51.0 73.0 34.02 47.02 \
  202605010912.central_asia.bin

This creates 202605010912.central_asia.bin and 202605010912.central_asia.bin.grid.txt, which contains the grid information required by Tmap.

4

Verify the subset with Tmap

Visualize the subset binary with Tmap and confirm that the specified region was extracted correctly.

The file 202605010912.central_asia.bin.grid.txt, created during the subset operation, stores the grid dimensions and latitude/longitude extent required by Tmap. Inspect its contents first.

Inspect the grid information
cat 202605010912.central_asia.bin.grid.txt
Read the grid information and run Tmap directly
GRID=202605010912.central_asia.bin.grid.txt

WIDTH=$(sed -n 's/^width=//p' "$GRID")
HEIGHT=$(sed -n 's/^height=//p' "$GRID")
NORTH=$(sed -n 's/^north=//p' "$GRID")
LAT_SPAN=$(sed -n 's/^lat_span=//p' "$GRID")
LAT_DIV=$(sed -n 's/^lat_div=//p' "$GRID")
WEST=$(sed -n 's/^west=//p' "$GRID")
LON_SPAN=$(sed -n 's/^lon_span=//p' "$GRID")
LON_DIV=$(sed -n 's/^lon_div=//p' "$GRID")

cd tmap
java tmap ../202605010912.central_asia.bin 1400 0 8 "W/m2" auto png   "$NORTH" "$LAT_SPAN" "$LAT_DIV"   "$WEST" "$LON_SPAN" "$LON_DIV"   "$WIDTH" "$HEIGHT"
cd ..
✓
DAY 1 CLEAR
DAY 1 is complete when Tmap displays the specified latitude/longitude extent correctly and only the spatial region has been extracted, without averaging the original values.
DAY 2 / TEMPORAL AVERAGING

Calculate hourly, daily, and monthly means

MISSION GOALCalculate an hourly mean from 15-minute observations, then calculate a daily mean from 24 hourly means. Finally, calculate a monthly mean from one month of daily means, following the same sequence used in the original JICA Training.
STEP 115-min observations → hourly mean
STEP 224 hours → daily mean
STEP 3Daily means → monthly mean
1

Calculate an hourly mean from 15-minute observations

METEOSAT-IODC observations are provided at 15-minute intervals. For example, the 08 UTC hour contains observations at 08:12, 08:27, 08:42, and 08:57. Average the observations available within that hour at each grid point to obtain the hourly mean for 08 UTC.

Conceptual diagram of hourly and daily averaging used in the JICA Training
Diagram used in the JICA Training, showing the sequence from 15-minute observations to an hourly mean and from 24 hourly means to a daily mean.
Missing observations: If observations are missing, do not treat them as zero. Divide by the number of observations actually available within that hour.
Download and run the DAY 2 data-retrieval script
wget -O download_day2.sh https://amaterass.science/jica-training/download_day2.sh
chmod +x download_day2.sh
./download_day2.sh
Download the analysis program
wget -O analysis_iodc_day2.tar.gz https://amaterass.science/jica-training/analysis_iodc_day2.tar.gz
tar xzf analysis_iodc_day2.tar.gz
javac analysis.iodc.day2/*.java
Example: hourly mean for 09 UTC
java -cp analysis.iodc.day2 IODCAverage hourly data/20260501 2026050109 \
  2026050109.hourlymean.dwn.sw.flx.sfc.fld.3km.bin
2

Calculate the daily mean from 24 hourly means

Applying the same procedure over one day yields hourly means from 00 to 23 UTC. The daily mean is the grid-point-wise average of those 24 hourly means.

Daily mean = (H00 + H01 + ... + H23) / 24
java -cp analysis.iodc.day2 IODCAverage daily data/20260501 20260501 \
  20260501.dailymean.dwn.sw.flx.sfc.fld.3km.bin
3

Aggregate daily means into a monthly mean

Once the daily means for the month are available, average them at each grid point to obtain the monthly mean. Here, daily means from 1–31 May 2026 are used.

Monthly mean = (D1 + D2 + ... + D31) / 31
java -cp analysis.iodc.day2 IODCAverage monthly data/daily 202605 \
  202605.monthlymean.dwn.sw.flx.sfc.fld.3km.bin
✓
DAY 2 CLEAR
DAY 2 is complete when you have verified the processing sequence from 15-minute observations to hourly means, from hourly means to a daily mean, and from daily means to a monthly mean.
DAY 3 / POINT TIME SERIES

Extract solar-radiation time series at specified locations

MISSION GOALFind the grid point nearest to each specified latitude/longitude coordinate and extract the solar-radiation values for 1 May 2026 as a time series. Finally, visualize the series with gnuplot.
STEP 1Specify locations
STEP 2Nearest grid point
STEP 3Extract time series
STEP 4gnuplot
1

Specify the analysis locations

Use the following two locations.

Point A

45.02°N, 56.48°E

Point B

41.02°N, 58.48°E

2

Find the nearest grid point

Scan the LAT/LNG grids and identify the grid point with the minimum distance to the specified coordinate. The same program also writes the solar-radiation time series at that grid point.

Download and run the DAY 3 data-retrieval script
wget -O download_day3.sh https://amaterass.science/jica-training/download_day3.sh
chmod +x download_day3.sh
./download_day3.sh
Download the analysis program
wget -O analysis_iodc_day3.tar.gz https://amaterass.science/jica-training/analysis_iodc_day3.tar.gz
tar xzf analysis_iodc_day3.tar.gz
javac analysis.iodc.day3/*.java
3

Extract a one-day time series

Read the value at the same grid point for each observation time and save the result as a two-column UTC / solar flux data file.

Point A
java -cp analysis.iodc.day3 IODCPointSeries data/20260501 \
  data/reference/standard_4000x4000.iodc.42.5.lat.fld.3km.bin \
  data/reference/standard_4000x4000.iodc.42.5.lng.fld.3km.bin \
  20260501 45.02 56.48 point_A_20260501.dat
Point B
java -cp analysis.iodc.day3 IODCPointSeries data/20260501 \
  data/reference/standard_4000x4000.iodc.42.5.lat.fld.3km.bin \
  data/reference/standard_4000x4000.iodc.42.5.lng.fld.3km.bin \
  20260501 41.02 58.48 point_B_20260501.dat
Output
├── point_A_20260501.dat
└── point_B_20260501.dat
4

Visualize the diurnal variation with gnuplot

Plot the two extracted time series on the same graph and examine the diurnal variation in solar radiation and the differences between the two locations.

In gnuplot, specify the output format, axes, time format, and input files. Enter the commands directly to plot both locations on the same graph.

Start gnuplot
gnuplot
Run the following commands in gnuplot
set terminal pngcairo size 1200,700 enhanced font ",14"
set output "point_timeseries_20260501.png"
set title "Surface downward shortwave flux: 1 May 2026"
set xlabel "UTC"
set ylabel "W/m^2"
set xdata time
set timefmt "%Y-%m-%dT%H:%M:%S"
set format x "%H:%M"
set grid
set key top left

plot "point_A_20260501.dat" using 1:2 with linespoints title "45.02N, 56.48E",      "point_B_20260501.dat" using 1:2 with linespoints title "41.02N, 58.48E"

exit

This creates point_timeseries_20260501.png.

✓
DAY 3 CLEAR
DAY 3 is complete when you can extract a one-day time series at the specified locations from the 4000 × 4000 solar-radiation grid and verify it with gnuplot.

References and further information

This training uses METEOSAT-IODC solar-radiation data to demonstrate three fundamental operations: spatial subsetting, temporal averaging, and point time-series extraction. Refer to the AMATERASS Product Guide as the authoritative source for product specifications.

AMATERASS Product Guide →　Product Download →

Download all sample programs: 
wget -O jica_training_programs.zip https://amaterass.science/jica-training/jica_training_programs.zip

This archive contains the Java source code and data-retrieval script used in DAY 1–3. Commands for analysis and visualization are shown directly in each DAY; completed output data are not included.

Notes and disclaimer

This material is intended for educational and research use.

About this material

This material reconstructs the JICA Training conducted from 2023 to 2025 under the JST–JICA SATREPS framework. Operation in every computing environment and continued availability in the future are not guaranteed.

Analysis results

Images and numerical results obtained through this training are for learning and analysis. Do not use them as the sole basis for decisions involving life or property.

Data and network availability

AMATERASS data and distribution servers may be temporarily unavailable because of maintenance, network problems, specification changes, or other reasons.

Software

Tmap is distributed under the MIT License. Separate licenses may apply to third-party data.

AMATERASS Data Use and Download →　Tmap and License →