Land AVHRR Workshop

---

CANBERRA, Monday 25th March 1996

Report by Tim McVicar
May 1, 1996


1. INTRODUCTION

The primary aim of the LAND AVHRR Workshop was to promote ideas and to break down any real or perceived barriers between the data acquisition, modelling and application communities.
AVHRR data were viewed in the workshop as being processed to three product levels, these are base, standard and research products. There are several operational environment management systems in place that rely heavily on AVHRR data. In the workshop these applications were given the opportunity to voice their concerns and hopes for future AVHRR processing at the three levels above.

From the general discussions there arose a series of common questions:

• What generic changes can occur to the data acquisition, base and standard product development which will mean improvements to all applications?
• What mechanisms need to be in place by the data acquisition, base and standard product development community to ensure that the users needs are adequately meet? Are the users needs and desires realistic and achievable?
• How can the application community make better decisions from the currently available data sets?

To provide a framework for readers who did not attend the workshop, the listing of presenters is given later. Some additional copies of the workshop proceedings are available from Tim McVicar on a first come first served basis.

2. PRESENTATIONS

The presentations were broken down into four main sections.

1. Generic issues of data quality and management.
Presentations by Wilson, Turner, Grant and Mitchell addressed this topic.

2. Current operational usages of reflective AVHRR data.
Presentations by Smith, Allan and Dudgeon.

3. Linking AVHRR measurements with non-remote sensing environmental models.
Presentations by Danaher and McVicar.

4. Introduction of the MODIS system, which will probably replace AVHRR data for many applications later this decade.
Presentation by Strahler.

Following and during the formal presentations there were discussions as to what was required and more importantly as to how to progress the requirements that were discussed. These are listed below the timetable.

Timetable

9:15 - 9:30 Welcome & Introduction by David Jupp, Chair.
CSIRO Earth Observation Centre

9:30 - 9:50 Murray Wilson
AVHRR Activities at the CSIRO Earth Observation Centre
CSIRO Earth Observation Centre

9:50 - 10:10 Peter Turner
A Standard Archiving Format for NOAA Full Pass Data
CSIRO Division of Atmospheric Research

10:10 - 10:35 Ian Grant
The Australian Continental Integrated Ground-truth Site Network: Aims and Applications
CSIRO Division of Atmospheric Research

10:35-11:00 Ross Mitchell
Current Calibration Practice for NOAA AVHRR Channels 1 and 2
CSIRO Division of Atmospheric Research

11:30 - 12:00 Richard Smith
Operational Use of NOAA-AVHRR by W.A. State Government
Western Australia Department of Land Administration

12:00 - 12:30 Grant Allan
Spatial and Temporal Fire Data for the N.T. derived from NOAA AVHRR Satellite Images
Conservation Commission of Northern Territory

12:30- 1:00 Graham Dudgeon
Operational Use of NOAA AVHRR NDVI Data for Agriculture in NSW
New South Wales Department of Agriculture

2:00 - 2:30 Tim Danaher and John Carter
Tree Density Mapping and Pasture Model Validation Using 1% of Australia's NOAA AVHRR Data
Queensland Department of Primary Industries

2:30 - 3:00 Tim McVicar
Linking a Water Balance Model with Thermal AVHRR Data
CSIRO Division of Water Resources

3:00 - 3:30 Alan Strahler
AVHRR Applications in Support of MODIS Product Development
Boston University, Department of Geography

3. OUTCOMES AND DISCUSSIONS FROM THE LAND AVHRR WORKSHOP

Data receiving stations should process data so that the base and standard products are both correct and repeatable within and between stations. That is, for the same region observed by two stations during a single overpass both the base and standard product digital counts must be the same.

• 1. Accurate and quantifiable measures of correctness should be defined jointly and then adhered to.
• 2. Geometric accuracy was suggested to be at RMS of 1pixel. This quality specification can be determined by a number of GCP's being selected in each image to ensure that the geometric accuracy is being meet.
• 3. Absolute radiometric correction was suggested to be specifiable to 5%. The main issue for such a specification is how to provide the supporting measures. Options include a combination of ground based or remotely sensed data collected from other satellite systems to enable radiometric quality to be documented for both base and standard products.
• 4. Additional data quality levels to the receiving stations standards should be produced and monitored by appropriate groups.

There needs to be understanding of the complexities involved by all members of the AVHRR community. For example, all members would like a correct, accurate and fast cloud detection and cloud contamination algorithm in place in all receiving stations. Some initial research has been undertaken on this topic. However, due to computational demands few if any algorithms have been operationally implemented. There still exists the opportunity that research will be of benefit due to a combination of algorithm development and faster processing computers. We would all like the AVHRR data to be physically correct and obtainable in a timely fashion to predefined standards. To do this, we must define the problems and offer solutions to them. There is an important opportunity and obligation for CSIRO to engage some basic research. Solutions should be scientifically correct and also operationally implementable. This has obvious advantages to all members of the AVHRR community.

Several issues were raised about how to ensure that present data archives are made consistent and correct. Such archives will form a baseline for future environmental monitoring projects and the need to ensure correctness and consistency is paramount. It was decided that rather then reprocess all images based on the notion that someone may want to use such data in the future, what is required is the knowledge, and specifications to be clearly, completely and correctly documented to allow reprocessing of special data archives. Is all this information known and documented?

It is thought that CSIRO Marine Labs, CSIRO DAR, AIMS, WASTAC and the BofMet need to get their systems all consistent and meeting the same common set of standards. Operational running of receiving stations and base data archiving also needs to be eventually transferred from CSIRO to BofMet. There is currently an agreement being discussed between the two organisations and greater discussions need to be undertaken not only between these two but also include ACRES and the AIMS/Townsville facility.

Data standards need to address the needs of the user community. Most important is the operational data user community, those agencies who are currently using AVHRR data to make management decisions. Any changes to data formats, standards etc need to be documented and feedback from the user community obtained before they take place. This needs to be rigorously planned and implemented on predetermined known dates and not undertaken in an ad hoc way.

There were also concerns about data availability expressed by some members of the user community. Management systems have been developed in which AVHRR data are the central element and continuity, both of geometrically and radiometrically correct data must be ensured to the level of control that the data acquisition, base and standard product community have. Also, issues of CSIRO data gatherers providing data to research applications internal to CSIRO must be addressed. This is an issue for CSIRO to address possibly under the banner of the recently formed EOC.

People must be willing to accept that some changes in the Australian AVHRR reception and subsequent processing pathways will occur. The changes will allow for a better consistent national set of standards to be met. All members of the remote sensing community that receive, process and use AVHRR data must be taken into account before any changes are made.

4. RECOMMENDATIONS

Out of the wide ranging discussions and issues addressed there were three which were agreed as of particular importance for the AVHRR community to achieve as soon as possible. These are:

• Data receiving stations should process data so that the base and standard products are consistent and to a specifiable and repeatable standard within and between stations;
• Geometric accuracy should aim to be RMS of 1pixel with recommended algorithms and processing plus measures of quality to be developed and implemented as the next step;
• Absolute radiometric correction should aim to 5% with measures of quality to be developed and implemented as the next step.

The resonsibility for taking steps to achieve these recommendations rests jointly with CSIRO, BofMet, WASTAC, AIMS as well as the prime data users in the State and Commonwealth Departments. Therefore a fourth recommendation is:

• CSIRO EOC should convene a group consisting of data receivers and data users to establish the workable solutions for these goals.

Tim McVicar
May 1, 1996

Copyright CSIRO 1999 ©