2_monitor_assess_batch_c2¶
This script reads a collection of baseline models and generates tasks to export condition assessment results to Earth Engine assets or to Google Drive. Updated "batch" version loops over a list of states and monitoring years creating tasks by monitoring year and state, enabling more flexible production of assessment products over large areas. Results can be added to the Map for testing; however this is not recommended when using this script to generate products for many states and/or years.
Output specifications¶
Monitoring results are generated for an ROI and final assessment layer is exported to a specified Image Collection.
| setting | description |
|---|---|
| VERSION | current version, 10-1 |
| BASELINES_COLLECTION | Collection ID for baselines results |
| OUTPUT_CRS | ESPG code, e.g. 'EPSG:5070' |
| OUTPUT_SCALE | multiplier, default: 1000 |
| KERNEL | kernel for result smoothing, default: 1 |
| OUTPUT_NAME | string to include in name of output Image Collection |
| OUTPUT_COLLECTION | new Collection to export results to (if exported to Asset) |
| EXPORT_TYPE | flag for export, options: Asset or Drive |
| STATE | list of USPS state abbreviations e.g. ['MA'] or ['MA', 'CT', 'RI'] |
Monitoring parameters¶
Condition scores are calculated using the last n baseline models for each acquisition date, and the resulting scores are averaged over all acquisition dates in the monitoring period. Scores in Path overlap zones are combined by weighting based on number of observations (acquisitions) in the monitoring period, resulting in a single map per year per state. Note: spectral index (DEPENDENT) and harmonic frequencies (HARMONICS) are determined from baseline asset metadata.
The following parameters are used to generate monitoring results:
| parameter | type | default | explanation |
|---|---|---|---|
| START_YEAR | string | 1995 |
first year in monitoring year list (YYYY) |
| END_YEAR | string | 2021 |
last year in monitoring year list (YYYY) , end year is inclusive |
| DAY_START | string | '05-01' |
start of monitoring period (MM-DD) |
| DAY_END | string | '10-01' |
end of monitoring period (MM-DD) , note end date is not inclusive |
| CC_THRESH | int | 80 |
maximum CLOUD_COVER_LAND for monitoring images |
| OBS_MIN | int | 30 |
exclude baselines with low number of observations |
| N_MODELS | int | 5 |
number of models to ensemble |
| MAX_STDDEV | int | 5 |
max allowable per Path standard deviation |
Results¶
Condition assessment products include Path-integrated mean (score_mean) and standard deviation (score_stddev) of condition scores, as well as the number of observations during the monitoring period (total_nobs). Updated version also now includes additional bands for mean and standard deviation of observed Tasseled Cap Greenness (tcg_mean and tcg_stddev), which provide context on the nature of post-disturbance dynamics and may be useful for separating emphermeral changes in vegetation condition from more changes in land cover. All outputs multiplied by output multiplier (default: 1000) and converted to in16 (short) to reduce output file size.
Example output:
| Index | Name | Type | CRS | Nominal Scale |
|---|---|---|---|---|
| 0 | score_mean | signed int16 | EPSG:5070 | 30 |
| 1 | score_stddev | signed int16 | EPSG:5070 | 30 |
| 2 | tcg_mean | signed int16 | EPSG:5070 | 30 |
| 3 | tcg_stddev | signed int16 | EPSG:5070 | 30 |
| 4 | total_nobs | signed int16 | EPSG:5070 | 30 |
Properties:
| Property | Value |
|---|---|
| monitor_end | 2010-10-01 |
| monitor_start | 2010-05-10 |
| monitor_year | 2010 |
| n_models | 5 |
| spectral_band | tcg |
| state | CT |