5 MOD13Q1/MOD13A1 (16-day 250/500-m) VI

This product is generated using the daily MODIS Level-2G (L2G) surface reflectance, pointer file, geo-angle file and 1-km state file (Fig. 2). Examples of the MOD13Q1 MODIS products for the Southwest USA are included at the end of this document (Fig. 8).

$\includegraphics[width=4in]{figs/MOD13-250-500m-ProdFlow.png}$

Figure 2: 250/500-m MODIS VI production flow diagram.

5.1 Algorithm Description

The VI algorithm operates on a per-pixel basis and requires multiple observations (days) to generate a composited VI. Due to orbit overlap, multiple observations may exist for one day and a maximum of four observations may be collected. In theory, this can result in a maximum of 64 observations over a 16-day cycle, however, due to the presence of clouds and the actual sensor spatial coverage, this number will range between 64 and 0 with decreasing observations from polar to equatorial latitudes. The MOD13A1 algorithm separates all observations by their orbits providing a means to further filter the input data.

Once all 16 days are collected, the MODIS VI algorithm applies a filter to the data based on quality, cloud, and viewing geometry (Fig. 3). Cloud-contaminated pixels and extreme off-nadir sensor views are considered lower quality. A cloud-free, nadir view pixel with no residual atmospheric contamination represents the best quality pixel. Only the higher quality, cloud-free, filtered data are retained for compositing. Thus, the number of acceptable pixels over a 16-day compositing period is typically less than 10 and often varies between 1 and 5, especially when one considers a mean global cloud cover of 50-60%. The goal of the compositing methodology is to extract a single value per pixel from all the retained filtered data, which is representative of each pixel over the particular 16-day period. The VI compositing technique uses an enhanced criteria for normal-to-ideal observations, but switches to an optional backup method when conditions are less then ideal. These techniques are:

Main: Constrained View angle - Maximum Value Composite (CV-MVC)
Backup: Maximum Value Composite (MVC)

$\includegraphics[width=4.5in]{figs/CompAlgorithmDataFlow-v3.png}$

Figure 3: MODIS VI Compositing algorithm data flow.

The technique employed depends on the number and quality of observations. The MVC is similar to that used in the AVHRR-NDVI product, in which the pixel observation with the highest NDVI value is selected to represent the entire period (16 days). Furthermore, the algorithm will choose the orbit observation with the highest NDVI if presented with multiple observations for the same day (multiple orbits).

The CV-MVC is an enhanced MVC technique, in which the number of observations $n$ ( $n$ being set to 2 at the moment) with the highest NDVI are compared and the observation with the smallest view angle, i.e. closest to nadir view, is chosen to represent the 16-day composite cycle.

All compositing methodologies result in spatial discontinuities, which are inevitable and result from the fact that disparate days can always be chosen for adjacent pixels over the 16-day period. Thus, adjacent selected pixels may originate from different days, with different sun-pixel-sensor viewing geometries and different atmospheric and residual cloud/smoke contamination.

5.2 Scientific Data Sets

The 250m/500-m VI product has the following characteristics (Table 1).

Table 1: Product MOD13A1: 16-day 250/500-m VI.

Science Data Set	Units	Data type	Valid Range	Scale factor
XYZm 16 days NDVI	NDVI	int16	-2000, 10000	0.0001
XYZm 16 days EVI	EVI	int16	-2000, 10000	0.0001
XYZm 16 days VI Quality detailed QA	Bits	uint16	0, 65534	NA
XYZm 16 days red reflectance (Band 1)	Reflectance	int16	0, 10000	0.0001
XYZm 16 days NIR reflectance (Band 2)	Reflectance	int16	0, 10000	0.0001
XYZm 16 days blue reflectance (Band 3)	Reflectance	int16	0, 10000	0.0001
XYZm 16 days MIR reflectance (Band 7)	Reflectance	int16	0, 10000	0.0001
XYZm 16 days view zenith angle	Degree	int16	-9000, 9000	0.01
XYZm 16 days sun zenith angle	Degree	int16	-9000, 9000	0.01
XYZm 16 days relative azimuth angle	Degree	int16	-3600, 3600	0.1
XYZm 16 days composite day of the year	Day of year	int16	1, 366	NA
XYZm 16 days pixel reliability summary QA	Rank	int8	0, 3	NA

XYZ means either 250 or 500 for MOD13Q1 and MOD13A1 products respectively.

5.3 Product Specific Metadata

A listing of the metadata fields used for QA evaluations of the MOD13 Q1/A1 VI product is included in Table 2.

Table 2: Metadata fields for QA evaluation of MOD13 Q1/A1.

I. Inventory Metadata fields for all VI products (searchable)

QAPERCENTINTERPOLATEDDATA

QAPERCENTMISSINGDATA

QAPERCENTOUTOFBOUNDSDATA

QAPERCENTCLOUDCOVER

QAPERCENTGOODQUALITY

QAPERCENTOTHERQUALITY

QAPERCENTNOTPRODUCEDCLOUD

QAPERCENTNOTPRODUCEDOTHER

II. Product specific metadata (searchable)
Product	Specific Metadata variable name (Best Quality)
MOD13Q1	NDVI250M16DAYQCLASSPERCENTAGE
MOD13Q1	EVI250M16DAYQCLASSPERCENTAGE
MOD13A1	NDVI500M16DAYQCLASSPERCENTAGE
MOD13A1	EVI500M16DAYQCLASSPERCENTAGE

III. Archived Metadata (not searchable)
Product	Metadata variable name (Array of QA usefulness histogram)
MOD13Q1	QAPERCENTPOORQ250M16DAYNDVI
MOD13Q1	QAPERCENTPOORQ250M16DAYEVI
MOD13A1	QAPERCENTPOORQ500M16DAYNDVI
MOD13A1	QAPERCENTPOORQ500M16DAYEVI

5.4 Global and Local Metadata Attributes

As in all MODIS products, the global metadata is written to the output file during the generation process and could be used for searching the archive about the product.

5.5 Quality Assurance

The quality of the MOD13A1 product is indicated and assessed through the quality assessment (QA) metadata objects and QA science data sets (SDS’s). The QA metadata objects summarize tile-level (granule) quality with several single words and numeric numbers, and thus are useful for data ordering and screening processes. The QA SDS’s, on the other hand, document product quality on a pixel-by-pixel basis and thus are useful for data analyses and application uses of the data.

QA Metadata

There are 18 QA metadata objects in the MOD13 Q1/A1 product. These objects (Table 3) are characterized by the following five attributes:

Object name: Uniquely identifies and describes the content of each object.
Object type: Describes the object as either an ECS mandatory, MODLAND mandatory, or VI product specific metadata object, and also as either text or numeric.
Description: Briefly describes the object, its valid value or format, and its sample value(s).
Level: Describes whether the object value is given for each SDS or not.

The ECS QA metadata are mandatory to all of the EOS products (the first 10 objects in Table 3), all of which are given for each SDS of the MOD13 Q1/A1 product. The first 6 objects are called QAFlags, including AutomaticQualityFlag, OperationalQualityFlag, ScienceQualityFlag, and their explanations. The AutomaticQualityFlag object indicates a result of an automatic QA performed during product generation and the following criteria are used to set its value:

Set to ’Passed’ if QAPercentMissingData $\leq$ 5%
Set to ’Suspect’ if QAPercentMissingData $>$ 5% or $<$ 50%
Set to ’Failed’ if QAPercentMissingData $>$ 50%

where the ’QAPercentMissingData’ is also an ECS QA metadata object and is described below. Explanation of the result of the AutomaticQualityFlag is given in the AutomaticQualityFlagExplanation metadata object.

The OperationalQualityFlag indicates the results of manual, non-science QA performed by processing facility personnel (DAAC or PI), i.e., if data are not corrupted in the transfer, archival, and retrieval processes. The flag has the value of ’Not Being Investigated’ if no non-science QA is performed. If the flag has the value other than ’Passed’ or ’Not Being Investigated’, explanation is given in the OperationalQualityFlagExplanation object.

The ScienceQualityFlag indicates the results of manual, science-QA performed by personnel at the VI Science Computing Facility (SCF). As for the OperationalQualityFlag, the flag has the value of ’Not Being Investigated’ if science QA is not performed. Explanation is given in the ScienceQualityFlagExplanation object if the flag has the value other than ’Passed’ or ’Not Being Investigated’.

The last 4 ECS QA metadata objects are called ’QAStats’. The QAStats indicate the percentages of pixels in the tile of which values are either interpolated (QAPercentInterpolatedData), missing (QAPercentMissingData), out of a valid range (QAPercentOutOfBoundData), or contaminated by cloud cover (QAPercentCloudCover).

There are 4 MODLAND mandatory QA metadata objects, all of which are designed to complement the ECS QA metadata objects. These indicate the percentages of pixels in the tile that are either good quality (QAPercentGoodQuality), unreliable quality (QAPercentOtherQuality), covered by cloud (QAPercentNotProducedCloud), or not produced due to bad quality other than cloud cover (QAPercentNotProducedOther). Different from the ECS QA metadata, only one set of values are given per tile.

The last 4 QA metadata objects in Table 3 are designed specifically for the MODIS VI product(s) (Product Specific Attributes, PSAs). Both NDVI500M16DAYQCLASSPERCENTAGE and EVI500M16DAYQCLASSPERCENTAGE objects indicate the percentages of pixels with good quality in the tile and, thus, should be equal to the QAPercentGoodQuality value unless there is a significant difference between the NDVI and EVI performance for the same tile.

The QAPERCENTPOORQ500M16DAYNDVI and QAPERCENTPOORQ500M16DAYNDVI indicate, respectively, the percent frequency distributions of the NDVI and EVI quality. Their values are computed as sums of the NDVI and EVI usefulness indices (described in the QA Science Data Set section) and, thus, include 16 integer numbers. The 16 numbers are ordered in the descending qualities from left to right and a sum of 16 numbers is always equal to 100. The first numbers in the QAPERCENTPOORQ500M16DAYNDVI and QAPERCENTPOORQ500M16DAYNDVI objects are equal to the values given in the NDVI500M16DAYQCLASSPERCENTAGE and EVI500M16DAYQCLASSPERCENTAGE objects, respectively.

Table 3: List of the QA Metadata Objects for the MOD13 Q1/A1 products (XYZ refers to either 250 or 500 m).

Object Name	Object Type	Description	Level
AutomaticQuality Flag	ECS Mandatory QAFlags, Text	Result of an automatic quality assessment performed during product generation. Valid value: ’Passed’, ’Suspect’, or ’Failed’	Per-SDS, Per-Tile
AutomaticQuality FlagExplanation	ECS Mandatory QAFlags, Text	Explanation of the result of the automatic quality assessment. Valid value: Up to 255 characters. Sample value: ’Run was successful But no land data found/processed’	Per-SDS, Per-Tile
OperationalQuality Flag	ECS Mandatory QAFlags, Text	Result of an manual, non-science quality assessment performed by production facility personnel after production. Valid value: ’Passed’, ’Suspect’, ’Failed’, ’Inferred Passed’, ’Inferred Failed’, ’Being Investigated’, or ’Not Being Investigated’	Per-SDS, Per-Tile
OperationalQuality FlagExplanation	ECS Mandatory QAFlags, Text	Explanation of the result of the manual, non-science quality assessment. Valid value: Up to 255 characters	Per-SDS, Per-Tile
ScienceQuality Flag	ECS Mandatory QAFlags, Text	Result of an manual, science quality assessment performed by production facility personnel after production. Valid value: ’Passed’, ’Suspect’, ’Failed’, ’Inferred Passed’, ’Inferred Failed’, ’Being Investigated’, or ’Not Being Investigated’	Per-SDS, Per-Tile
ScienceQuality FlagExplanation	ECS Mandatory QAFlags, Text	Explanation of the result of the manual, science quality assessment. Valid value: Up to 255 characters	Per-SDS, Per-Tile
QAPercent InterpolatedData	ECS Mandatory QAStats, Numeric	Percentage of interpolated data in the tile. Valid value: 0 100. Sample value: 12	Per-SDS, Per-Tile
QAPercent MissingData	ECS Mandatory QAStats, Numeric	Percentage of missing data in the tile. Valid value: 0 100. Sample value: 8	Per-SDS, Per-Tile
QAPercent OutOfBoundData	ECS Mandatory QAStats, Numeric	Percentage of data in the tile of which values are out of a valid range. Valid value: 0 100. Sample value: 2	Per-SDS, Per-Tile
QAPercent CloudCover	ECS Mandatory QAStats, Numeric	Percentage of cloud covered data in the tile. Valid value: 0 100. Sample value: 15	Per-SDS, Per-Tile
QAPercent GoodQuality	MODLAND Mandatory, Numeric	Percentage of data produced with good quality in the tile. Valid value: 0 100. Sample value: 4	Per-Tile
QAPercent OtherQuality	MODLAND Mandatory, Numeric	Percentage of data produced with unreliable quality in the tile. Valid value: 0 100. Sample value: 56	Per-Tile
QAPercent NotProducedCloud	MODLAND Mandatory, Numeric	Percentage of data produced but contaminated with clouds in the tile. Valid value: 0 100. Sample value: 32	Per-Tile
QAPercent NotProducedOther	MODLAND Mandatory, Numeric	Percentage of data not produced due to bad quality in the tile. Valid value: 0 100. Sample value: 8	Per-Tile
NDVIXYZM16DAY QCLASS PERCENTAGE	VI Product Specific, Numeric	Percentage of NDVI data produced with good quality in the tile. Valid value: 0 100. Sample value: 4	Per-Tile
EVIXYZM16DAY QCLASS PERCENTAGE	VI Product Specific, Numeric	Percentage of EVI data produced with good quality in the tile. Valid value: 0 100. Sample value: 4	Per-Tile
QAPERCENT POORQ XYZM16DAYNDVI	VI Product Specific, Numeric	Summary statistics (percent frequency distribution) of the NDVI usefulness index over the tile. Valid format: (N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N), where N = 0 100. Sample value: (4,0,0,0,44,6,18,15,5,0,0,0,0,0,0,8)	Per-Tile
QAPERCENT POORQ XYZM16DAYEVI	VI Product Specific, Numeric	Summary statistics (percent frequency distribution) of the NDVI usefulness index over the tile. Valid format: (N, N, N, N, N, N, N, N, N, N, N, N, N, N, N, N), where N = 0-100. Sample value: (4,0,0,0,44,6,18,15,5,0,0,0,0,0,0,8)	Per-Tile

QA Science Data Sets

A summary Quality layer has been included in the MOD13Q1: “pixel reliability”. This layer contains ranked values describing overall pixel quality (Table 4).

Table 4: MOD13Q1/A1 Pixel Reliability.

Rank Key	Summary QA	Description
-1	Fill/No Data	Not Processed
0	Good Data	Use with confidence
1	Marginal data	Useful, but look at other QA information
2	Snow/Ice	Target covered with snow/ice
3	Cloudy	Target not visible, covered with cloud

Because evaluation of the past 6 years of MODIS C3 and C4 data collections revealed insignificant differences between the Quality assignments for NDVI versus EVI, C5 MOD13 products include a single Quality layer pertinent to both indices, rather than one layer for each (Table 5). This reduces data volume as well as user confusion with multiple Quality layers.

QA bits are designed to document conditions under which each pixel was acquired and processed.

Table 5: Descriptions of the VI Quality Assessment Science Data Sets (QA SDS).

Bits	Parameter Name	Value	Description
0-1	VI Quality (MODLAND QA Bits)	00	VI produced with good quality
		01	VI produced, but check other QA
		10	Pixel produced, but most probably cloudy
		11	Pixel not produced due to other reasons than clouds
2-5	VI Usefulness	0000	Highest quality
		0001	Lower quality
		0010	Decreasing quality
		0100	Decreasing quality
		1000	Decreasing quality
		1001	Decreasing quality
		1010	Decreasing quality
		1100	Lowest quality
		1101	Quality so low that it is not useful
		1110	L1B data faulty
		1111	Not useful for any other reason/not processed
6-7	Aerosol Quantity	00	Climatology
		01	Low
		10	Intermediate
		11	High
8	Adjacent cloud detected	0	No
		1	Yes
9	Atmosphere BRDF Correction	0	No
		1	Yes
10	Mixed Clouds	0	No
		1	Yes
11-13	Land/Water Mask	000	Shallow ocean
		001	Land (Nothing else but land)
		010	Ocean coastlines and lake shorelines
		011	Shallow inland water
		100	Ephemeral water
		101	Deep inland water
		110	Moderate or continental ocean
		111	Deep ocean
14	Possible snow/ice	0	No
		1	Yes
15	Possible shadow	0	No
		1	Yes

The first two bits are used for the MODLAND mandatory per-pixel QA bits that summarize the VI quality of the corresponding pixel locations. Percentages of sums of its four possible values (bit combinations) over a tile will give the MODLAND mandatory QA metadata object values (Table 6).

Table 6: Relationship between the MODLAND Mandatory per-pixel QA Bits and QA Metadata Objects.

VI Quality Bit Combination	Corresponding QA Metadata Object
00: VI produced, good quality	QAPercentGoodQuality
01: VI produced, but check other QA	QAPercentOtherQuality
10: Pixel produced, but most probably cloudy	QAPercentNotProducedCloud
11: Pixel not produced due to other reasons than clouds	QAPercentNotProducedOther

The 2nd QA bit-field is called the VI usefulness index. The usefulness index is a higher resolution quality indicator than the MODLAND mandatory QA bits (16 levels) and its value for a pixel is determined from several conditions, including 1) aerosol quantity, 2) atmospheric correction conditions, 3) cloud cover, 4) shadow, and 5) sun-target-viewing geometry (Table 7). As shown, there is a specific score that is assigned to each condition and a sum of all the scores gives a usefulness index value for the pixel. An index value of 0000 is corresponding to the highest quality, while the lowest quality is equal to a value of 1100 (i.e., 13 levels). The three largest values are reserved for three specific conditions which are shown in Table 5. There are relationships between the VI usefulness index and the MODLAND mandatory QA bits. Pixels with the index value of 0000 and 1111 always have the MODLAND QA bit values of 00 and 11, respectively.

Table 7: VI Usefulness Index Scaling Method for the MOD13 Q1/A1 products.

Parameter Name	Condition	Score
Aerosol Quantity	If aerosol climatology was used for atmospheric correction (00)	2
	If aerosol quantity was high (11)	3
Atmosphere Adjacency Correction	If no adjacency correction was performed (0)	1
Atmosphere BRDF Correction	If no atmosphere-surface BRDF coupled correction was performed (0)	2
Mixed Clouds	If there possibly existed mixed clouds (1)	3
Shadow	If there possibly existed shadow (1)	2
View zenith angle ( $q_ v$ )	If $q_ v >$ 40 $^\circ$	1
Sun zenith angle ( $q_ s$ )	If $q_ s >$ 60 $^\circ$	1

The next three QA bit-fields document atmospheric correction scenarios of each pixel. The bits 6-7 are used to indicate aerosol quantity, and the bits 8 and 9 indicate whether an adjacency correction and atmosphere-surface BRDF coupled correction, respectively, are applied or not.

Bit 10 indicates a possible existence of mixed clouds. As the original spatial resolutions of the red and NIR bands are 250 m, these two bands were spatially aggregated to a 500 m resolution before the computations of VIs. The mixed cloud QA bit is flagged if any of the 250 m resolution pixels that were used for the aggregations were contaminated with cloud.

Bits 11-13 are used for the land/water mask. The input land/water mask to the MOD13 Q1/A1 VI product has 7 land/water classes. The VIs are not computed for pixels over the ocean/inland water class.

Bits 14 and 15 indicate possible existences of snow/ice and shadow, respectively.