The Aggregation Algorithms
AVG
AggregatorAverage: This aggregator implements a weighted averaging with per overflight weighting as described in the referring NASA
publication listed in the overview section. This aggregation method is well suited for satellite data acquired
in the visible and NIR by integrating over the pixel area. It offers three parameters:
- Weight coefficient: the weighting coefficient (float). A value of 0.0 disables weighting, default value is 1.0
- Output counts: if selected, the measurement counts per bin-cell are stored in the output
- Output sums: if selected, the measurement value sums per bin-cell are stored in the output instead of mean values.
This feature allows a continuous aggregation of data over a longer period when not all input data is available at once.
Output bands are:
- xxx_mean: the weighted mean value
- xxx_sigma: the weighted standard deviation
- xxx_counts: the measurement counts (optional)
If Output sums is selected:
- xxx_sum: the weighted sum of the input measurements
- xxx_sum_sq: the weighted sum of the squares of the input measurements
- xxx_weights: the aggregated weights
- xxx_counts: the measurement counts (optional)
AVG_OUTLIER
AggregatorAverageOutlierAware: implements an "outlier aware" averaging. It realises this by first calculating a mean and standard-deviation
for the complete set of input measurements. Then the algorithm removes all measurements that are more than n standard deviations away from
the mean value, where n is the so called "deviation factor". The remaining data is averaged using an unweighted version
of the standard "AVG" aggregation described above. Parameters:
- Deviation Factor: the deviation factor (float). Defines the distance to the mean value beyond which data is considered as outlier,
distance = deviationFactor * standardDeviation. Default value is 1.0
Output bands are:
- xxx_mean: the weighted mean value
- xxx_sigma: the weighted standard deviation
- xxx_counts: the measurement counts
MEAN_OBS
AggregatorMeanObs: calculates the standard, unweighted mean and standard deviation using all input measurements per cell.
This aggregation method is suited for input data consisting of discrete measurements, e.g. number of detected fires.
The aggregator has no public parameters.
Output bands are:
- xxx_mean: the mean value
- xxx_sigma: the standard deviation
- xxx_counts: the measurement counts
MIN_MAX
AggregatorMinMax: an aggregator tracking the minimum and maximum value of the measurements over the binning period.
This algorithm can be used to detect data ranges.
The aggregator has no public parameters.
Output bands are:
- xxx_min: the smallest value in the grid cell
- xxx_max: the largest value in the grid cell
ON_MAX_SET
AggregatorOnMaxSet: this aggregator selects input data from a list of input variables under the condition that
another band has the maximum value in the aggregation input set. For example it allows aggregating the output of all
radiance data under the condition that the sensor-zenith angle is maximal. A tracking for a minimal value condition is possible by
defining a virtual variable that contains the negative of the variable to be tracked.
Parameters are:
- Maximum Band Name: selects the band that is tracked for the maximal value
- Source band Names: select a list of input band names. The values of these bands are written to the output if the Maximum Band has the peak value
Output bands are:
- xxx_max: the largest value of the Maximum Bandin the input data
- xxx_mjd: the Modified Julian Day value of the maximm value measurement
- source_bands: all source band values defined in the parameter Source band Names
PERCENTILE
AggregatorPercentile: an aggregator that computes the p-th percentile, i.e. the value of a variable below which a
certain percent (p) of observations fall.
Parameters are:
- Percentile: the precentile value (integer). Value must be between 0 and 100, default is 90.
Output bands are:
- xxx_p: the percentile value of the input data
SUM
AggregatorSum: this aggregator sums up all input measurements per grid cell. It has no public parameters.
Output bands are:
- xxx_sum: the sum of the input data
