Presentation on how to chat with PDF using ChatGPT code interpreter
Introducing GSIF (seminar at Lamont campus)
1. Global Soil Information Facilities
(A methodological framework for Open Soil
Information)
Tomislav Hengl
ISRIC World Soil Information, Wageningen University
Seminar at CIESIN, Sept 14 2011
2. Key issues
What do we know about world soils?
What do you know about the GlobalSoilMap.net project?
How to produce complete GlobalSoilMape.net property
maps?
How will soil information t into the Global Land
Information System?
Seminar at CIESIN, Sept 14 2011
3. My backgrounds
Senior researcher at ISRIC World Soil Information;
PhD in pedometric mapping @ ITC (GIS institute in Enschede)
in 2003;
2 years university assistant; 2.5 years JRC Ispra; 2 years
University of Amsterdam;
My expertise: Geostatistics, Digital Soil Mapping, spatial
data analysis, geomorphometry (vice-chair);
Global Soil Information Facilities
Seminar at CIESIN, Sept 14 2011
5. What am I doing in USA?
Seminar at CIESIN, Sept 14 2011
6. AfricaSoils.net
Thank you!
1. Markus Walsh (Keith Shepherd)
2. Sonya Ahamed Pedro Sanchez
Seminar at CIESIN, Sept 14 2011
7. My main inspirations / principles of work
Open Source software for education and research
Crowd sourcing systems for environmental data
collection
Publicly accessible (soil) data products
Seminar at CIESIN, Sept 14 2011
8. Important assumptions
My research philosophy
is based on 4 important assumptions:
Seminar at CIESIN, Sept 14 2011
9. Assumption #1
Humans (companies and governments)
need to be closely monitored
Seminar at CIESIN, Sept 14 2011
10. Did you know?
Global biodiversity has been heavily degraded due to human
activities. The Living planet index has dropped from 1970s
to 60% and will continue to do so (source: Millennium
Assessment project).
By 2048 we will run out of sh (your children will leave on a
planet where there are hardly any visible sh in the oceans).
15-35% of global irrigation withdrawals are estimated to be
unsustainable (source: WBCSD).
Every year, 9.4 million ha of forests are lost (source: FAO
World agriculture: towards 2015/2030).
Seminar at CIESIN, Sept 14 2011
17. Soil threats
Soils are also more important because we are slowly loosing them:
305 million ha of land has been completely degraded (no
longer suitable for agriculture).
10-50% irrigated land aected by salinization (source:
GLASOD).
For a forest to return takes maybe 100 years; it takes
100400 years to produce 1 cm of topsoil are soils
renewable resource at all?
Seminar at CIESIN, Sept 14 2011
18. Soils might become precious in future
Reports by FAO (2002) show that, in future, 80 percent of
increased crop production in developing countries will have to come
from intensication higher yields, increased multiple cropping
and shorter fallow periods.
Any agricultural or environmental management model
requires soil data as an input to estimation of yields, water
and nutrient dynamics.
World demand for cereals has jumped from 39 million tones (in
1970) to 103 million tones (in 2000) (source: FAO World
agriculture: towards 2015/2030).
Seminar at CIESIN, Sept 14 2011
19. Assumption #3
Soil Information (global)
is one of the poorest GIS layers
Seminar at CIESIN, Sept 14 2011
20. What do we know about world soils?
Harmonized World Soil Database: 1 km resolution gridded
soil property maps (16 properties for top and sub-surface soil).
1:5M scale FAO-UNESCO Soil Map of the Word: from
which ISRIC has produced 5 by 5 arc-minutes global soil
property maps (for 020, 2040, 4060, 6080 and
80100 cm) in combination with the ISRIC-WISE soil prole
database.
The Distributed Active Archive Center (DAAC) soil
property maps
USGS-produced soil property maps
Atlas of the Biosphere soil maps
Seminar at CIESIN, Sept 14 2011
21. HWSD vs GlobCov
GlobCover HWSD
Seminar at CIESIN, Sept 14 2011
25. Assumption #4
Global Resource Planning System
can do much better than a local one
Seminar at CIESIN, Sept 14 2011
26. GLIS
Soil properties (soil information system)
- physical and chemical soil properties, nutrient
capacity, water storage, acidity/salinity…
Model library Live weather channel (meteorological forecasting)
- anticipated temperature (min, max), rainfall, frost
hazard, drought hazard, flood hazard…
Fertilization
Irrigation Plant monitoring channel (MODIS/ENVISAT)
Pest treatment - current biomass production, biomass anomalies
Best crop calendar (pest and diseases), plant health…
Yield estimates
Environmental risks Socio-economic data (site-specific)
GLOBAL - administrative units, new laws and regulations,
LAND INFORMATION market activity, closest offices, agro-dealers…
SYSTEM
Suggest the best
land use practice Query site
attributes
Information Update with
incorrect? ground truth data
Spatial location (site)
Seminar at CIESIN, Sept 14 2011
28. GlobalSoilMap.net
An international initiative to make soil property maps (7+3) at
six depths at 3 arcsecs (100 m).
the leitmotif is to assemble, collate, and rescue as much of
the worlds existing soil data ;
The soil-equivalent of the OneGeology.org, GBIF, GlobCover
and similar projects.
The biggest DSM project ever!
Seminar at CIESIN, Sept 14 2011
29. GlobalSoilMap.net in comparison with other projects
4.0
GLWD
EcoRegions HWSDv1
5.6 km MOD12C1
MOD13C2 CHLO/SST
3.5
FRA
Resolution (m) in log-scale
WorldClim
GPWv3
3.0
DMSP-OLSv4
GlobCov2 OneGeology?
2.5
SRTM GADM GlobalSoilMap?
2.0
1990 1995 2000 2005 2010 2015 2020
Year
Seminar at CIESIN, Sept 14 2011
30. World soils in numbers
Total land area: 14.8 billion ha
73.6%)
Estimated total productive soil area: 10.9 billion ha (
Drylands (deserts, semi-deserts): 3.6 billion ha (24.3%)
Wetlands (swamps, marshes, and bogs): 440 million ha (3%)
Arable and permanent crops: 1.5 billion ha (11%)
Potential areas suitable in varying degrees for the rainfed
production of arable and permanent crops: 2.8 billion ha
Seminar at CIESIN, Sept 14 2011
31. Global Soil Mapping (in numbers)
The total productive soil areas: about 104 million square
km.
Seminar at CIESIN, Sept 14 2011
32. Global Soil Mapping (in numbers)
The total productive soil areas: about 104 million square
km.
k
To map the world at 100 m (1:200 ), would cost about
5 billion EUR (0.5 EUR per ha) using traditional methods.
Seminar at CIESIN, Sept 14 2011
33. Global Soil Mapping (in numbers)
The total productive soil areas: about 104 million square
km.
k
To map the world at 100 m (1:200 ), would cost about
5 billion EUR (0.5 EUR per ha) using traditional methods.
We would require some 65M proles according to the strict
rules of Avery (1987).
Seminar at CIESIN, Sept 14 2011
34. Global Soil Mapping (in numbers)
The total productive soil areas: about 104 million square
km.
k
To map the world at 100 m (1:200 ), would cost about
5 billion EUR (0.5 EUR per ha) using traditional methods.
We would require some 65M proles according to the strict
rules of Avery (1987).
World map at 0.008333333 arcdegrees (ca.1 km) resolution is
an image of size 43,200 Ö21,600 pixels.
Seminar at CIESIN, Sept 14 2011
35. Global Soil Mapping (in numbers)
The total productive soil areas: about 104 million square
km.
k
To map the world at 100 m (1:200 ), would cost about
5 billion EUR (0.5 EUR per ha) using traditional methods.
We would require some 65M proles according to the strict
rules of Avery (1987).
World map at 0.008333333 arcdegrees (ca.1 km) resolution is
an image of size 43,200 Ö21,600 pixels.
27 billion pixels needed to represent the whole world in
100 m (productive soil areas).
Seminar at CIESIN, Sept 14 2011
36. Productive soil areas
Figure: Soil productive area mask derived using the MODIS LAI images.
Projected in the Transverse Mercator system used e.g.in Google Maps.
Seminar at CIESIN, Sept 14 2011
37. Maybe GlobalSoilMap.net will not cost as much?
Technology might be the solution!
Automated mapping
Global soil covariates SRTM DEM GDEM TanDEM-X,
MODIS LST, Meteo images (SMOS), TRMM
Downscaling methods
Soil spectroscopy (rapid soil sampling)
Seminar at CIESIN, Sept 14 2011
38. The 3(4) bottles of vine
At the GSM2011.org meeting at JRC Ispra several people have
oered to award the DSM team that delivers a complete
country/continent size GlobalSoilMap.net product:
1 bottle if it contains complete list of soil properties;
1 bottle if it includes uncertainty estimates;
1 bottle if its accuracy is satisfactory;
(1 bottle if it is being used by agronomist);
Seminar at CIESIN, Sept 14 2011
39. ISRIC's response to the GSM initiatives
Global Soil Information Facilities
a set of open tools and data portals
Seminar at CIESIN, Sept 14 2011
40. GSIF components
1. Cyber infrastructure for input, analysis and visualization
of data.
2. Global databases (legacy data, gridded covariates) that
are main inputs to global soil mapping.
3. Software tools (modules and packages) and manuals for
creation of geoinformation, for instance, according to
the GlobalSoilMap.net specications.
4. Standards and protocols for data entry, map generation
and data sharing.
Seminar at CIESIN, Sept 14 2011
41. Overview
Open Soil Profiles Soil covariates (worldgrids)
Global Continental scale Country/state-level
Soil variables
Soil site info
Soil analytical data
Descriptive properties 5.6 km repository 1 km repository 100 / 250 m repository
R packages
GSIF package
Map import module
Data entry module
Harmonization module
Spline fitting
Spatial analysis module
plotKML
(GSIF Servers) cyber infrastructure
Data import to R
Data visualization
Data export
Soil property maps
Six+four key soil parameters Webmapping API
Global coverage (organic carbon, pH, clay, silt,
sand, coarse fragments) Real-time spatial prediction
at six standard depths (0-5, 5- (Google Maps)
15, 15-30, 30-60, 60-100, 100- GlobalSoilMap.net functionality
200 cm) for web-applications
and with included upper and Geo-serving and geoprocessing
lower 95% probability ranges
functionality
100 m (250 m, 1 km and 5.6 km)
Seminar at CIESIN, Sept 14 2011
42. Proposed implementation
1. Produce a suite of utilities to import, re-format, analyze
and visualize spatial soil data
2. Design them so they t the needs of operational global
soil mapping
3. Focus on using R+OSGeo
4. Get the whole DSM community involved (in design, in
development, in use)
5. Provide training in development and use to countries and
nodes
Seminar at CIESIN, Sept 14 2011
43. List of utilities
1. Global soil mapping (core) package GSIF
2. Soil visualization package plotKML
3. Soil Reference Library SRL
4. Geo-services (PythonWPS, Geoserver, RServe, GDAL utilities)
Seminar at CIESIN, Sept 14 2011
44. Main principles of programming
1. Hide complexity from the users (scale, eective precision,
3D geostat)
2. Deliver data and results so that no software training is required
to open it ( KML)
3. Link to R+OSGeo community ( do not invent functionality
that already exists and is operational)
Seminar at CIESIN, Sept 14 2011
45. The software triangle
Statistical
computing
GDAL
ground
GRASS GIS
overlays,
time-series KML
Browsing of
geo-data
GIS analysis
Seminar at CIESIN, Sept 14 2011
46. Functionality (plotKML)
Visualize soil proles measurements (using the original soil
colors);
Visualize soil prole photographs;
Plot results of prediction (soil property maps) using standard
color schemes;
Visualize uncertainty in the soil property maps;
Seminar at CIESIN, Sept 14 2011
52. Why KML? (1)
Google Earth is #1: 350 millions of downloads!
Seminar at CIESIN, Sept 14 2011
53. Why KML? (2)
People that made Google Earth understand
(space-time) statistics
Seminar at CIESIN, Sept 14 2011
54. What is Global Soil Mapper?
Global Soil Mapper
is an automated system (R+OSGeo) for
generation of soil property maps
that meet the GlobalSoilMap.net specs
Seminar at CIESIN, Sept 14 2011
55. Global Soil Mapper: the main principles
1. Put emphasis on inputs (point data, soil polygon maps,
covariates) and tools (GSIF)
2. Fit model parameters per soil property for the whole
world
3. Map the world block-by-block (automated mapping)
4. Update the maps as soon as the new point / covariates
arrive (while you sleep)
Seminar at CIESIN, Sept 14 2011
56. GSIF function predict
predict.gsm ( target.var = ORCDRC, observations = soilprofiles.org,
+ covariates = worldgrids.org, model = GMN-RK,
+ newdata = boundingbox )
model = GMN-RK is the default global model ( tted using the
global data);
Seminar at CIESIN, Sept 14 2011
57. GMN-RK
Global Multiscale Nested RK =
a 3D spatial prediction method
based on a four-level nested Regression-Kriging
Seminar at CIESIN, Sept 14 2011
58. Nested RK
z(sB ) = m0 (sB−k ) + e1 (sB−k |sB−[k−1] ) + . . . + ek (sB−1 |sB ) + ε(sB )
where m0 (sB−k ) is the value of the target variable estimated at the
coarsest global scale (B − k ), B−1 , . . . ,B−k are the higher order
components, ek (sB−k |sB−[k−1] ) is the residual variation from scale
sB−k to a ner resolution scale sB−[k−1] , and ε is the spatially
auto-correlated residual soil variation dealt with ordinary kriging.
Seminar at CIESIN, Sept 14 2011
59. Multiscale signal
S4 + S3 + S2 + S1 + e
S4 + S3 + S2 + S1
S4 + S3 + S2
S4 + S3
S4
Figure: Based on McBratney (1998): Some considerations on methods
for spatially aggregating and disaggregating soil information.
Seminar at CIESIN, Sept 14 2011
60. 65k soil proles
Figure: USDA NCSS Characterization Database, CSIRO National Soil
Archive, ISRIC WISE, SPADE, Iran National soil prole database,
Canadian Soil Information System, and African soil proles.
Seminar at CIESIN, Sept 14 2011
62. Gridded maps for Malawi
Parent General Erosion Land
Climate Biomes
material land use deposition management
Rainfall map of the world
5.6 km
MODIS-based long term Land
Surface Temperature (day/night)
Elevation
Geologic Provinces of Africa
1 km
Soil polygon map (FAO classes)
ENVISAT Land Cover map (GlobCov)
MODIS (MCD12Q1) land cover dynamics
250 m
MODIS (MCD13Q1) Enhanced Vegetation
Index (EVI) and medium infrared band (MIR)
TWI, TRI, Slope,
Surface roughness,
100 m Insolation
Landsat ETM
thermal band
Seminar at CIESIN, Sept 14 2011
63. The downscaling approach
Figure: Predictions of soil organic carbon for top depth at various scales.
By running a multiscale global model we can ll in the large gaps in the
data (interpolate instead of extrapolate).
Seminar at CIESIN, Sept 14 2011
67. Conclusions
Value of soil information is likely to grow.
Seminar at CIESIN, Sept 14 2011
68. Conclusions
Value of soil information is likely to grow.
GSIF is a methodological framework for continuous
production of Open Soil Information.
Seminar at CIESIN, Sept 14 2011
69. Conclusions
Value of soil information is likely to grow.
GSIF is a methodological framework for continuous
production of Open Soil Information.
Advantage of using a GMN-RK is that we can employ a
diversity of predictors ( CLORPT factors work at dierent
scales).
Seminar at CIESIN, Sept 14 2011
70. Conclusions
Value of soil information is likely to grow.
GSIF is a methodological framework for continuous
production of Open Soil Information.
Advantage of using a GMN-RK is that we can employ a
diversity of predictors ( CLORPT factors work at dierent
scales).
Global is now (local statistical models will become extinct?).
Seminar at CIESIN, Sept 14 2011
71. Soils of Mars
Astrophysists are selling something very abstract for a
high price. Soils are the basic of human survival, yet we
manage to acquire much less research funds.
Neil McKenzie (CSIRO)
We know more about soils of Mars than about soils of
Africa.
Pedro Sanchez (Earth Institute)
Seminar at CIESIN, Sept 14 2011
72. Next steps
Re-implement the method using a `clean' data
Next step:
set (USA data) and write up step-by-step guidelines.
Publish the GSIF package and WPS for GSM (anyone can
become a digital soil mapper).
Complete and publish plotKML and GSIF R packages.
Map the whole of Africa at 100 m (end of 2012).
Seminar at CIESIN, Sept 14 2011