cumulative detections

README.md

@@ -109,9 +109,11 @@ The system was developed for use in the Amazon, but it has also been seen to wor
 
 This most recent assessment was run with an ensemble of six models: [48px_v3.2-3.7ensemble_2024-02-13.h5](https://github.com/earthrise-media/mining-detector/blob/main/models/48px_v3.2-3.7ensemble_2024-02-13.h5). We recorded outputs for all patches with a mean score over 0.5, on a scale from 0 to 1. 
 
-[Output data](https://github.com/earthrise-media/mining-detector/tree/main/data/outputs/48px_v3.2-3.7ensemble) are saved year by year and presented in two formats. The first format records the mean score and the six individual predictions from models 3.2-3.7 for each saved patch. The second, streamlined, format, with filenames tagged _dissolved-0.6_, saves only patches meeting a higher 0.6 mean score threshold and then merges adjacent patches into larger polygons. 
+[Output data](https://github.com/earthrise-media/mining-detector/tree/main/data/outputs/48px_v3.2-3.7ensemble) are saved year by year and presented in three formats. The first format records the mean score and the six individual predictions from models 3.2-3.7 for each saved patch. The second, streamlined, format, with filenames tagged _dissolved-0.6_, saves only patches meeting a higher 0.6 mean score threshold and then merges adjacent patches into larger polygons. 
 
-The dissolved predictions are presented on [Amazon Mining Watch](https://amazonminingwatch.org/) and should suffice for most users. At lower prediction threshold, the ensemble captures more mining at the cost of more false positive detections; at higher threshold, the ensemble is stingier with its predictions and more likely to be correct in the mines it surfaces. The choice of 0.6 reflects our own preference in this tradeoff. Users wanting to tune the prediction threshold can work with the data in the patch format. 
+The dissolved predictions are presented on [Amazon Mining Watch](https://amazonminingwatch.org/) and should suffice for most users. At lower prediction threshold, the ensemble captures more mining at the cost of more false positive detections; at higher threshold, the ensemble is stingier with its predictions and more likely to be correct in the mines it surfaces. The choice of 0.6 reflects our own preference in this tradeoff. Users wanting to tune the prediction threshold can work with the data in the patch format.
+
+Finally, because of year-to-year variance in detections of small mine scars, and because mine scars can fade from detection where vegetation regrows, we include a set of _cumulative_ detections. These datasets aggregate the dissolved yearly detections from 2018 through the later year indicated in the filename, delineating places where mining has ever been detected to that point. By 2023, the cumulative area mapped is almost 50% larger than the area mapped in 2023 alone.
 
 #### Assessement of mining in the Amazon basin in 2020 (v1 Amazon Mining Watch dataset)