COMBINED USE OF POLAR ORBITING AND GEO-STATIONARY SATELLITES TO
IMPROVE TIME INTERPOLATION IN DYNAMIC CROP MODELS FOR FOOD
SECURITY ASSESSMENT
V. Venus? *, D. Rugege"
* Dept. of Natural Resources, International Institute for Geo-Information Science and Earth Observation (ITC),
Hengelosestraat 99 P.O. Box 7500 AA Enschede, The Netherlands - venus@ite.nl
° RSA Centre for Environment and Development University of Natal, Pietermaritzburg Private Bag, X01 Scottsville 3209,
South Africa - rugeged@nu.ac.za
PS, WG VII/2 — Sustainable Agriculture & EcoSystem Approach
KEYWORDS: Remote Sensing, Agriculture, Monitoring, Temperature, Temporal, Multisensor
ABSTRACT:
Use of satellite data in crop growth monitoring could provide great value for regional food security assessments. By using the difference
between remotely sensed crop canopy temperature and the corresponding ambient temperature at the time of the satellite overpass the
daily actual rate of transpiration can be inferred. This relationship allows adjustment of the actual rate of assimilation and hence of actual
crop growth. Although promising results were obtained using methods based on this premise, the sensitivity of these methods to temporal
variability outside the time-window of the satellite overpass is a concern. Based on our findings we show that temporal aspects are
indeed not negligible and an improvement in the accuracy of crop productivity assessments can be achieved if data from satellites with
different temporal and spatial resolutions are combined. In this study, data from the Advanced Very High Resolution Radiometers
(AVHRR) instrument aboard the polar orbiting satellite National Oceanic and Atmospheric Administration #14 (NOAA-14) and data from
the Visible Infrared Spin Scan Radiometer (VISSR) instrument onboard the Geostationary Meteorological Satellite #5 (GMS-5) are
integrated in a dynamic crop growth simulation procedure. The existing estimation method we used to evaluate our results against solely
dependents on data from polar orbiting satellites, which observe the earth surface too infrequently to yield sufficient clear-sky
observations (only 24 out of 100 days of the crop cycle were cloud-free). More observations of temperature differences between the crop
canopy and ambient air can be obtained when coupled with geo-stationary satellite measurements that represent the diurnal cycle. The
linear interpolation procedure applied to obtain proxies for missing days improved accordingly. The results indicate that Storage Organ
Mass (SOM) values can be determined from the new method with a higher degree of certainty as compared to the existing method.
When evaluated against SOM values as observed at Quzhou, P.R. of China, experimental maize fields, the estimates are within an
accuracy of about 150 kg ha'1, a relative error of less than 1,8%. This also confirms our hypothesis that observations from geo-stationary
satellites as an additional data source, which are more frequently made than measurements from polar orbiting satellites, can be useful to
explain temporal dynamics of crop stress to better estimate regional crop productivity.
1. INTRODUCTION production are depressed) more energy is left for canopy heating,
and vice versa. In other words, the difference between the
Various approaches for estimating crop production have been remotely sensed crop canopy temperature and the corresponding
proposed and tested since the 1960's to assist food (security) ^ ambient temperature is co-determined by the actual rate of crop
planners. These mainly aimed at improving traditional crop status transpiration. This temperature difference as detected at the
reports and used techniques varying from crop growth simulation moment of the satellite pass is then converted into daily
on a point-to-point basis to empirically derived index values that equivalent values. If the transpiration term is isolated from the
link satellite data with observed crop productivity. For regional ^ energy budget and divided by the theoretical transpiration rate of
applications, quantitative crop growth modeling is a most a constraint-free reference crop, a so-called ‘coefficient of water
promising development since it considers the dynamics of sufficiency' with daily equivalent values (c//20, 0-1) results,
essential physiological and environmental processes and thus aids indicating the degree stomata closure and therewith the degree to
in the universal quantification of productivity of food and fiber which photosynthetic activity is reduced by the compounded
crops. Observations from satellites have been found useful to infer constraints to the actual crop. Recurrent reading at short intervals
the required parameters for crop growth modeling on a real-time accounts for the dynamics of crop growth and produces
and area basis, but procedures still remain challenging successive, near real-time estimates of actual crop performance.
1.1 Rational However, in many cases data collection is hindered by the
; : presence of clouds. This is particularly true for polar orbiting
Only few satellite sensors have a sufficient number of channels to satellites; for only 24 out of 109 days of the crop cycle in 1999 it
derive input parameters meaningful for crop growth simulation. was possible to obtain cloud-free pixels from NOAA-14 imagery
Key to remotely sensed (RS) production estimation is the crop’s to infer the crop performance as will be detailed shortly hereafter.
energy budget. Incident solar radiation incident on the crop ^ Some scientists proposed interpretation techniques to overcome
canopy is used in part for vaporization of water (crop the temporal limitations of using satellite observations (Jin and
transpiration). If less water is used (and assimilation and Dickinson, 1999). Driessen and Rugege (2002) argued however
* Corresponding author.
212
Internati
Lorre
that no i
best it re
estimatec
research
use of mt
12 Obj
The over
technolog
satellite «
crop pro
resolutiot
from sate
detection
2 ME
Analytica
food anc
Netherlar
de Vries,
growth b
intervals
steady ra
organ ma
interval;
calculatio
productio
reference
total dry |
basic dat
algorithm
growth s
follows a
regional €
values.
11 Cro|
As a min
(PS-1), d
which prc
light, the
Crop:
PS-1: P.Y
The level
not the ;
normally
weed plar
In many
constraint
dry region
clear sunn
drainage)
restrict lo:
caused by
therefore
routine th:
water req
constraint
2) calcula
as a fun
mechanist
