Introduction

Surveying and data acquisition method, in a nutshell and in a broad meaning has always been on the behalf of the benefit of human kind. Even the centimeters of error occurred in the Irrigation Route determination project may cost heavily. Likewise, the significant error based on Earth Observation Data may cause climate prediction to alter, false disaster alarm system etc. Hence the significance of Earth Observation Data cannot be neglected at present and validation of data with ground observation or with any other independent method is a must.

‘Validation’, therefore refers to the comparison between satellite sensor measurements and those made by an independent method. The measured quantity must be properly validated and processed and checked before any significant product can be derived from the measured quantity. The process of validation therefore involves geographical sampling and interpretation, as well as consideration of the physical measurements and the algorithms used.

Problems that are associated with validation are many. The area that is covered by independent or particularly field method is much less than the area covered by remotely sensed data, because of cost and time or other reasons. And it is not guaranteed the scale at which the data is collected at field matches with the sensor’s field of view. Therefore it is important that the sample collected should be highly accurate in order for Earth Observation Data to have significant product.

Discussion

Validation, in addition to calibration is an essential component in nearly all remote-sensing based studies. Therefore, it is important to calibrate and validate the in-situ measurement with that of various remote sensing models. It is also important that these in-situ measurement products have themselves been validated and their accuracy and uncertainty established.

The following section insights with some of the recent researches conducted highlighting the main problem and its associated solution. Some other research topic has been presented too.

In the recent years there has been significant research in the validation activities. Some key problems have been defined and the solution has been suggested.

Problem: Effect of mixed pixels in validation efforts for products generated from the Moderate Resolution Imaging Spectroradiometer (MODIS).

Solution: To minimize the effect of mixed pixels, Milne and Cohen (1999) suggested locating field plots in areas with low spatial variance in Landsat data.

Problem: Validating estimates of leaf area index (LAI) derived from MODIS.

Solution: In a related study conducted in Botswana, Tian et al. (2002) addressed this problem by using field plots ranging from 34 to 2,756 m². Because the scale at which the field data were collected was not consistent with the spatial resolution of MODIS data, a direct comparison was not feasible. To this, Tian et al. (2002) used Landsat data to identify homogeneous areas on the basis of spectral similarity and adjacency using a segmentation algorithm. These regions were then used to acquire the field data to the scale of the MODIS pixels.

Problem: Validation of MODIS hotspot.

Solution: In the past, there were number of satellite performing the work of validation of hotspot: Advanced Very High Resolution Radiometer (AVHRR) and Geostationary Operational Environmental Satellite (GOES) are some to mention, who beside their native function were also performing the work of validation of hotspot. However, MODIS is the first sensor specifically designed and developed to include capability for forest fires detection. At present, the revisiting time of the Terra and Aqua satellites (where MODIS resides) over Thailand is every one or two days. This helps an under-coverage limitation of the near real time monitoring system for Thailand. Additional satellites with similar or better than MODIS capabilities are needed to provide more complete information for more efficient forest fire control and management.

In addition to addressing the existing problems, there has also been significant effort in the developments in the ‘validation’ of satellite sensor products for the land surface, Integrating the Cross-sensor Calibration and Validation System for GEOSS Support, Validation of Remote Sensing Content-Based Information Retrieval (RS-CBIR) Systems upon Scarce Data.

Also before the data is reached to the user from vendor, various guidelines for data providers have been drafted by GEO Data Sharing Working Group (DSWG) to understand the importance of data quality information associated with GEOSS data resources. This is essential for users to understand the quality of data sets and to combine this quality information with other metadata components in order to determine the appropriateness, or fitness, of these data sets for the users’ applications and/or purposes or combination of various datasets.

Conclusion

New innovations are required for more technological advancement. In recent times, the paradigm known as Data Intensive Science is currently changing the way research and innovation is being conducted. This paradigm is based on access and analysis of large amounts of existing or new data that were or are created not only by scientific instruments and computers, but also by processing and collating existing archived data. Remotely sensed data, in particular, are gathered daily in large amounts of information. They not gather information about the planet but are also used to monitor and assess the status of the natural and built environments. Researches are based on service-oriented architectures that are facilitating the linkage of data or resources and processing.

A systematic approach is adopted in resenting the definition and concept of validation, some innovative research conducted; presenting some of the major problem associated in the real world and associated solutions for those problems.