diff --git a/data/occurrences/11_01_23-DJI_0977.csv b/data/occurrences/11_01_23-DJI_0977.csv
new file mode 100644
index 0000000000000000000000000000000000000000..159202a268c82e3daa7d152a149263d00670aeee
--- /dev/null
+++ b/data/occurrences/11_01_23-DJI_0977.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50b50f60dc3c52da042a23ab882e74945849d8e1e759cef1e4a43fd96c1bf35f
+size 14345580
diff --git a/data/occurrences/11_01_23-DJI_0978.csv b/data/occurrences/11_01_23-DJI_0978.csv
new file mode 100644
index 0000000000000000000000000000000000000000..a030b70abe9a04936c1b81edfe2604254478487e
--- /dev/null
+++ b/data/occurrences/11_01_23-DJI_0978.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34d62178f47214ff697be175d4932d76457cecb14de168f95b0fc65736757779
+size 4707447
diff --git a/data/occurrences/11_01_23-DJI_0979.csv b/data/occurrences/11_01_23-DJI_0979.csv
new file mode 100644
index 0000000000000000000000000000000000000000..876454d5b8702005507a18066861d4442a123d15
--- /dev/null
+++ b/data/occurrences/11_01_23-DJI_0979.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76f56cfffbd31ec514a81891026ea75e7fccd095d3350f8072c2a40d55cea9e5
+size 5069349
diff --git a/data/occurrences/11_01_23-DJI_0980.csv b/data/occurrences/11_01_23-DJI_0980.csv
new file mode 100644
index 0000000000000000000000000000000000000000..e9db46c502d445fc126f9e2feab44b2fa90dec2f
--- /dev/null
+++ b/data/occurrences/11_01_23-DJI_0980.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:828e26b2bd31afee01c80c6293d92761473244909c1730bdf0cd816f4826d380
+size 17952129
diff --git a/data/occurrences/12_01_23-DJI_0001.csv b/data/occurrences/12_01_23-DJI_0001.csv
new file mode 100644
index 0000000000000000000000000000000000000000..f6e304a0d3e6854a366d6c82d18968c836e37c7c
--- /dev/null
+++ b/data/occurrences/12_01_23-DJI_0001.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:063b1d2607a004dc420eb8dc3b7931573ba0125629e3c2110198cfaf45a757d4
+size 72939976
diff --git a/data/occurrences/12_01_23-DJI_0002.csv b/data/occurrences/12_01_23-DJI_0002.csv
new file mode 100644
index 0000000000000000000000000000000000000000..d16769babebba8091d2abe65f6ba10c08754192d
--- /dev/null
+++ b/data/occurrences/12_01_23-DJI_0002.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61698a88cccdea95c9dad90209beac8960673f7cf26376efddc59ff129eb14de
+size 15943851
diff --git a/data/occurrences/12_01_23-DJI_0003.csv b/data/occurrences/12_01_23-DJI_0003.csv
new file mode 100644
index 0000000000000000000000000000000000000000..eaee3d6ba2380800fad708e3f059e37773fa7b3e
--- /dev/null
+++ b/data/occurrences/12_01_23-DJI_0003.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:473cf4b2158212bfb07a30e2e041f999b961fc965e9a6a74a9563bbbe09649f2
+size 393014
diff --git a/data/occurrences/12_01_23-DJI_0006.csv b/data/occurrences/12_01_23-DJI_0006.csv
new file mode 100644
index 0000000000000000000000000000000000000000..09ea64932b71dcb80e1bb7aa9431e0281def22cd
--- /dev/null
+++ b/data/occurrences/12_01_23-DJI_0006.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24e9f7957961bbb119652b31fde8c00c97781cab02381d4a943f5bd23efdfb9a
+size 13992446
diff --git a/data/occurrences/12_01_23-DJI_0007.csv b/data/occurrences/12_01_23-DJI_0007.csv
new file mode 100644
index 0000000000000000000000000000000000000000..526e6537876a37f1f32add48995b26ad9a4c6345
--- /dev/null
+++ b/data/occurrences/12_01_23-DJI_0007.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2d6c94bd52278800a2885797cb328e824c4ec8b901c09141e1f0458beb8f3ea
+size 23336478
diff --git a/data/occurrences/12_01_23-DJI_0008.csv b/data/occurrences/12_01_23-DJI_0008.csv
new file mode 100644
index 0000000000000000000000000000000000000000..ef3da5c55aabfd91e310c67cc563434e1c1d9ada
--- /dev/null
+++ b/data/occurrences/12_01_23-DJI_0008.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:962c8c1f1feeb6703cb73230b800e5fac262a45047ebb73ce5248921562a8a61
+size 10031966
diff --git a/data/occurrences/12_01_23-DJI_0987.csv b/data/occurrences/12_01_23-DJI_0987.csv
new file mode 100644
index 0000000000000000000000000000000000000000..19d92f2bd1805c2dc359978ca0b2999bcd3c8c94
--- /dev/null
+++ b/data/occurrences/12_01_23-DJI_0987.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cbcf7f65fc5aff7f098e319a134450be27a3ed0555f9bd9709c39c92896b7a05
+size 11339992
diff --git a/data/occurrences/12_01_23-DJI_0988.csv b/data/occurrences/12_01_23-DJI_0988.csv
new file mode 100644
index 0000000000000000000000000000000000000000..1ae7e674c1f54cafa570d975395a0fe6db8a627f
--- /dev/null
+++ b/data/occurrences/12_01_23-DJI_0988.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf9047be691bd6bd5b4b5d44a0da450b6dd7a3830d0bd44d081e044c74cd56d7
+size 15173906
diff --git a/data/occurrences/12_01_23-DJI_0989.csv b/data/occurrences/12_01_23-DJI_0989.csv
new file mode 100644
index 0000000000000000000000000000000000000000..f2f77b4513fa9a2b6089d026e47d710d6ec1ff93
--- /dev/null
+++ b/data/occurrences/12_01_23-DJI_0989.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b18174dbd8f092e650ff1a8e0a030e859cda0d306738ddad86d581a1aa555dfb
+size 3635599
diff --git a/data/occurrences/12_01_23-DJI_0992.csv b/data/occurrences/12_01_23-DJI_0992.csv
new file mode 100644
index 0000000000000000000000000000000000000000..57079ed16adb889b023b44d2de827f2cc3bba862
--- /dev/null
+++ b/data/occurrences/12_01_23-DJI_0992.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d2ee072132b6da70013ff058c4008eafa2dc91f75d83c880eb86a9827c3053c
+size 4940182
diff --git a/data/occurrences/12_01_23-DJI_0994.csv b/data/occurrences/12_01_23-DJI_0994.csv
new file mode 100644
index 0000000000000000000000000000000000000000..80ecbcb407aa6f393efeb4885907dc8c9c84a645
--- /dev/null
+++ b/data/occurrences/12_01_23-DJI_0994.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:feff00240d115a46be0513177508aa462740c86c838d2d69ce73a2390c5bff50
+size 3355881
diff --git a/data/occurrences/12_01_23-DJI_0997.csv b/data/occurrences/12_01_23-DJI_0997.csv
new file mode 100644
index 0000000000000000000000000000000000000000..35d987ea1a6999f576f92765aa3c2aefc98ec2ac
--- /dev/null
+++ b/data/occurrences/12_01_23-DJI_0997.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36cf5cfc61e879023b4b72a114b0ba780f29e8976f355a1bfe4cecb3409403ab
+size 24568166
diff --git a/data/occurrences/12_01_23-DJI_0998.csv b/data/occurrences/12_01_23-DJI_0998.csv
new file mode 100644
index 0000000000000000000000000000000000000000..5518acedc913a188d0729ecf4876661fce0308ac
--- /dev/null
+++ b/data/occurrences/12_01_23-DJI_0998.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:71f72036b13beeaa38f914eaa87271c31f9895c877291b0569cadb0039376cbd
+size 4709447
diff --git a/data/occurrences/13_01_23-DJI_0009.csv b/data/occurrences/13_01_23-DJI_0009.csv
new file mode 100644
index 0000000000000000000000000000000000000000..a71adee18194cde810e3242dd091dba3d68a274a
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0009.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8be9eadb5f7261283e20e39eb4ab9784d5aab6eff38cbaf8c599dccb7d75612
+size 885424
diff --git a/data/occurrences/13_01_23-DJI_0011.csv b/data/occurrences/13_01_23-DJI_0011.csv
new file mode 100644
index 0000000000000000000000000000000000000000..9b17c7fa3525cd399985bd98b2544cb7aa6eec43
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0011.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:11c5a419893474e8367afd0621a3bd3d8f609111dc0268579d73c0f2f753bbe3
+size 8603958
diff --git a/data/occurrences/13_01_23-DJI_0012.csv b/data/occurrences/13_01_23-DJI_0012.csv
new file mode 100644
index 0000000000000000000000000000000000000000..7749df5d7ea91ba676a37859208524854e7b5d04
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0012.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4324d88a6e239f9d697665caabbe709b86e0e58bdf08066795480f23e241eff0
+size 4987483
diff --git a/data/occurrences/13_01_23-DJI_0013.csv b/data/occurrences/13_01_23-DJI_0013.csv
new file mode 100644
index 0000000000000000000000000000000000000000..f4fd5e6cdab35aa52392c936d9d42d50e61dc3be
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0013.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1da985cabc3df39703a7163a7bad6c0d981e0cead7d047944e1a533f69a8a142
+size 2605445
diff --git a/data/occurrences/13_01_23-DJI_0015.csv b/data/occurrences/13_01_23-DJI_0015.csv
new file mode 100644
index 0000000000000000000000000000000000000000..42c120fc4d7c2d640e973aef8be9cbd70bc4a8be
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0015.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83eccb91f1099871d449bbaf494865dc5346a3cfeca66ef9c342ec6b195960f4
+size 724085
diff --git a/data/occurrences/13_01_23-DJI_0016.csv b/data/occurrences/13_01_23-DJI_0016.csv
new file mode 100644
index 0000000000000000000000000000000000000000..f0d29abaa093ae50ad8f625065c93318a58b9aa5
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0016.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da174aed2695485960517dbb90fce5b66c21a3568df64ef94958229d4a5bee06
+size 1101965
diff --git a/data/occurrences/13_01_23-DJI_0018.csv b/data/occurrences/13_01_23-DJI_0018.csv
new file mode 100644
index 0000000000000000000000000000000000000000..af0d27b8460efc1caa0e74812107c1a05e4f3919
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0018.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee1ec8c76429905ad91cca2c78f1594f3296ded5b668f5be942da35d8c5a67c3
+size 3502983
diff --git a/data/occurrences/13_01_23-DJI_0019.csv b/data/occurrences/13_01_23-DJI_0019.csv
new file mode 100644
index 0000000000000000000000000000000000000000..eefd8a79808a460bf571b53ecd1f67626fb741a3
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0019.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e8412006abf0c006a8aa7363ff910b2792edd4b2921a9e154c91e6af3ae5c5e
+size 21287954
diff --git a/data/occurrences/13_01_23-DJI_0020.csv b/data/occurrences/13_01_23-DJI_0020.csv
new file mode 100644
index 0000000000000000000000000000000000000000..7344ff62d9998bf5e6b1f95855677f98ef1a7c5e
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0020.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e221f97e792e9969eae4b24b061d6bcc2c91a60b95f6e0c928afff8ec6b8bf1c
+size 36827652
diff --git a/data/occurrences/13_01_23-DJI_0021.csv b/data/occurrences/13_01_23-DJI_0021.csv
new file mode 100644
index 0000000000000000000000000000000000000000..5e5c76e1efe5c48f4222a9266f99f0992001c9a4
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0021.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6edb66a9d153f7042744bc11c72183011426a7e23e9133547cd5eab2072ccbb8
+size 2885156
diff --git a/data/occurrences/13_01_23-DJI_0022.csv b/data/occurrences/13_01_23-DJI_0022.csv
new file mode 100644
index 0000000000000000000000000000000000000000..bbc46b858bb63ee565eb5cc057762746cfc7d1a4
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0022.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c089d5fef1035693d88210a7fa2d797b845849a2239faf6ed6e1124f9259e84b
+size 13267907
diff --git a/data/occurrences/13_01_23-DJI_0023.csv b/data/occurrences/13_01_23-DJI_0023.csv
new file mode 100644
index 0000000000000000000000000000000000000000..60b9035692e323dc4e9e204e79c3ed1a67ca72da
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0023.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54ded963edf1eaeb70e8bc753351b438a673e01feba1b77d81e17f9bfa4a91d7
+size 21795825
diff --git a/data/occurrences/13_01_23-DJI_0024.csv b/data/occurrences/13_01_23-DJI_0024.csv
new file mode 100644
index 0000000000000000000000000000000000000000..b7cff602f619dd1b6f771d0ff7c25a1227ec503c
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0024.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1516741845d4a926a920159878f298881ebb75ff4e83359bb776cbe2690d19a7
+size 20764115
diff --git a/data/occurrences/13_01_23-DJI_0029.csv b/data/occurrences/13_01_23-DJI_0029.csv
new file mode 100644
index 0000000000000000000000000000000000000000..19b6d9b160e667c8fcc36dec3f7cc2f8ac0e9a90
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0029.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76db9f9a0b32de8496616a2bfb4874d690c43b00128927706802d94eb016d65d
+size 4526391
diff --git a/data/occurrences/13_01_23-DJI_0031.csv b/data/occurrences/13_01_23-DJI_0031.csv
new file mode 100644
index 0000000000000000000000000000000000000000..0fd863e6dee9c1e1926e6571cefa3dcf6219ee38
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0031.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc706e8dcba3caea6aa5f7c23b4bdd3d4d693841b8978299f6922befc7b70b95
+size 7834907
diff --git a/data/occurrences/13_01_23-DJI_0032.csv b/data/occurrences/13_01_23-DJI_0032.csv
new file mode 100644
index 0000000000000000000000000000000000000000..1b6a1b965a77a9c3a69fb405341b32985fad829a
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0032.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50c4a57bbb42815cf6286454ffddf4c19cffcf8d9addae4973c4f646f5da762f
+size 24098088
diff --git a/data/occurrences/13_01_23-DJI_0033.csv b/data/occurrences/13_01_23-DJI_0033.csv
new file mode 100644
index 0000000000000000000000000000000000000000..f85fb5d7685c874d5aa3d9950611ca0671aeff0b
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0033.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e1a024fbb641d4d33c3d99202f0bf17fa505628d23ff222f5a42b31f619c944
+size 28102013
diff --git a/data/occurrences/13_01_23-DJI_0035.csv b/data/occurrences/13_01_23-DJI_0035.csv
new file mode 100644
index 0000000000000000000000000000000000000000..a95bc0199d850d14234b6c4bd8a3ea4c8d3305fe
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0035.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aaa12b6ea3f628d4b7d64ea00df020dab7096f99285ed23c340c9eb9f19583b2
+size 1195985
diff --git a/data/occurrences/13_01_23-DJI_0036.csv b/data/occurrences/13_01_23-DJI_0036.csv
new file mode 100644
index 0000000000000000000000000000000000000000..e5117d964d77243341ea0cb52bd03fe8107961e1
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0036.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19d06fe004cd653358b2fd45a7729581e2e6d5379da0033175194f1a20eba1ca
+size 7280882
diff --git a/data/occurrences/13_01_23-DJI_0037.csv b/data/occurrences/13_01_23-DJI_0037.csv
new file mode 100644
index 0000000000000000000000000000000000000000..a81009b85135f59e8b6db52d1bfcf8936279927d
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0037.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50e6e6bb98c69f6a8d40a3a8bd5a6d69f73ce2173116776fb64dc0331c197456
+size 28862771
diff --git a/data/occurrences/13_01_23-DJI_0038.csv b/data/occurrences/13_01_23-DJI_0038.csv
new file mode 100644
index 0000000000000000000000000000000000000000..2dd51fb113533833fc96dfdf42b9ba34acb97030
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0038.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22593b4b450125cf5f380dd8978493394c1ebfb0d534a52729ee0e107e6e2cea
+size 6426417
diff --git a/data/occurrences/13_01_23-DJI_0039.csv b/data/occurrences/13_01_23-DJI_0039.csv
new file mode 100644
index 0000000000000000000000000000000000000000..172523a1a1721d961b77f5b45447bcd5fdbadd89
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0039.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2f6b7aaec38b4b17343bcdf40a79f27165f66f6c21b3dd5a7f7a74aaefb1534
+size 17464554
diff --git a/data/occurrences/13_01_23-DJI_0040.csv b/data/occurrences/13_01_23-DJI_0040.csv
new file mode 100644
index 0000000000000000000000000000000000000000..3622df62f6459c96eb80ef4603841ad2d695727a
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0040.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ea2b254059a04308ca48ec3785e314bea344916ebd07f8c67ae42fe4b3fd3137
+size 73829509
diff --git a/data/occurrences/13_01_23-DJI_0041.csv b/data/occurrences/13_01_23-DJI_0041.csv
new file mode 100644
index 0000000000000000000000000000000000000000..1a45942a9a425c47edd46f48a6f480a42dea4d47
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0041.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f91e0b3e92eaff10d209a5ff8015a7b1c83732cbc977d188901e49c3a1383e7
+size 37163078
diff --git a/data/occurrences/13_01_23-DJI_0042.csv b/data/occurrences/13_01_23-DJI_0042.csv
new file mode 100644
index 0000000000000000000000000000000000000000..a8e0e08fd4ce9c140de4e672783f8508fe55614b
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0042.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8100a3f59c6caa6d27fb0e7f2aef33c855b261c15275adfe5725073d8edc8c5
+size 131545325
diff --git a/data/occurrences/13_01_23-DJI_0043.csv b/data/occurrences/13_01_23-DJI_0043.csv
new file mode 100644
index 0000000000000000000000000000000000000000..2a898f4d70b7dab8b9128ed2cfca0c6a0032bb4d
--- /dev/null
+++ b/data/occurrences/13_01_23-DJI_0043.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8306170cb37736ab0714faca6a810e085b6476e718e4b2b03b0f106c5c6745e5
+size 4416255
diff --git a/data/occurrences/16_01_23_flight_1-DJI_0001.csv b/data/occurrences/16_01_23_flight_1-DJI_0001.csv
new file mode 100644
index 0000000000000000000000000000000000000000..a9b1d9ac14feb6f9ca1027d4282c0b648e344607
--- /dev/null
+++ b/data/occurrences/16_01_23_flight_1-DJI_0001.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a68315f8cd40869e70a0a98138b08da2f5cc68df773feb6504780f4f3cefbbb2
+size 25789369
diff --git a/data/occurrences/16_01_23_flight_1-DJI_0002.csv b/data/occurrences/16_01_23_flight_1-DJI_0002.csv
new file mode 100644
index 0000000000000000000000000000000000000000..54f48c7e6dfd29f338998fc067c5d071433d9dfd
--- /dev/null
+++ b/data/occurrences/16_01_23_flight_1-DJI_0002.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c9e0b2aab6d9e4db5c34c844f3e48b4c68639b0727478d14ce573c40e0cf6ad
+size 90351965
diff --git a/data/occurrences/16_01_23_flight_1-DJI_0003.csv b/data/occurrences/16_01_23_flight_1-DJI_0003.csv
new file mode 100644
index 0000000000000000000000000000000000000000..4abaa07e973b31f30de946afa327228e366e13a5
--- /dev/null
+++ b/data/occurrences/16_01_23_flight_1-DJI_0003.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9f81e6faa82153c3a3716a56c4495a5683d1d245446895703bea52808b8ebf7
+size 98499971
diff --git a/data/occurrences/16_01_23_flight_2-DJI_0001.csv b/data/occurrences/16_01_23_flight_2-DJI_0001.csv
new file mode 100644
index 0000000000000000000000000000000000000000..8c4a45867db8d0bbaf64b9eda104466287be8f7a
--- /dev/null
+++ b/data/occurrences/16_01_23_flight_2-DJI_0001.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e836ca3ffdc2da432bfc2c28807e5ac3d4f2fc6204e5a33e77f8fa0b62e2d48
+size 23856551
diff --git a/data/occurrences/16_01_23_flight_2-DJI_0002.csv b/data/occurrences/16_01_23_flight_2-DJI_0002.csv
new file mode 100644
index 0000000000000000000000000000000000000000..013a789877d53d6beb9940fe372a24f5e84068bc
--- /dev/null
+++ b/data/occurrences/16_01_23_flight_2-DJI_0002.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c8846aead3c54aabd8753ad56bc8bb0ffecbd80ff18f78d10d418e39bf9cc15
+size 64329673
diff --git a/data/occurrences/16_01_23_flight_2-DJI_0003.csv b/data/occurrences/16_01_23_flight_2-DJI_0003.csv
new file mode 100644
index 0000000000000000000000000000000000000000..1a4c3df20b57dfac94bec4e19ced2017b5dbc383
--- /dev/null
+++ b/data/occurrences/16_01_23_flight_2-DJI_0003.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b325a42653886f4ab23c64b9c8af1bea09dd88ff04ff963315b00c91e6f6e47
+size 85046093
diff --git a/data/occurrences/16_01_23_flight_2-DJI_0004.csv b/data/occurrences/16_01_23_flight_2-DJI_0004.csv
new file mode 100644
index 0000000000000000000000000000000000000000..cd5bd9ad610a5ebc51c1aa69d2acb95bdee674ae
--- /dev/null
+++ b/data/occurrences/16_01_23_flight_2-DJI_0004.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8f873e5f110ef4f480fc7d6bfbcf141926e003a9ff3a46b2d4709e2c58868e9
+size 53917955
diff --git a/data/occurrences/17_01_2023_session_1-DJI_0006.csv b/data/occurrences/17_01_2023_session_1-DJI_0006.csv
new file mode 100644
index 0000000000000000000000000000000000000000..a5858b4e943a07cbd416f142fe1e1f0d21ad9d79
--- /dev/null
+++ b/data/occurrences/17_01_2023_session_1-DJI_0006.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06625afa027c79c937de33cb67f1c34212e92470b748845a99b788789d293b3c
+size 14238708
diff --git a/data/occurrences/17_01_2023_session_1-DJI_0007.csv b/data/occurrences/17_01_2023_session_1-DJI_0007.csv
new file mode 100644
index 0000000000000000000000000000000000000000..a297d10f9a34df3ce986fbeaaf3d87f5686ff9c1
--- /dev/null
+++ b/data/occurrences/17_01_2023_session_1-DJI_0007.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c586d1bbe51de3d448934258f2205c1dc0b656ff30b26fbc7cd716aca99cf007
+size 16140825
diff --git a/data/occurrences/17_01_2023_session_2-DJI_0008.csv b/data/occurrences/17_01_2023_session_2-DJI_0008.csv
new file mode 100644
index 0000000000000000000000000000000000000000..66050b2877c6ddc8cb62f02e99428aa2eb1f3d6d
--- /dev/null
+++ b/data/occurrences/17_01_2023_session_2-DJI_0008.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07042f1cce31321e4d45da5a70288830dbdf6d28671da17b7cc41f324b3beaa7
+size 2633162
diff --git a/data/occurrences/17_01_2023_session_2-DJI_0010.csv b/data/occurrences/17_01_2023_session_2-DJI_0010.csv
new file mode 100644
index 0000000000000000000000000000000000000000..323663b8b2ccf9255215e5d85eb969d25d0ee5a4
--- /dev/null
+++ b/data/occurrences/17_01_2023_session_2-DJI_0010.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b713bb0d490cb7eeecffaedca935ed4b82ed6293d6912168a0c26914f2dcc18
+size 30948909
diff --git a/data/occurrences/17_01_2023_session_2-DJI_0011.csv b/data/occurrences/17_01_2023_session_2-DJI_0011.csv
new file mode 100644
index 0000000000000000000000000000000000000000..e4a4573a053598923a31bc66e6394c4f606ab90c
--- /dev/null
+++ b/data/occurrences/17_01_2023_session_2-DJI_0011.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d554a6a74001d272aa56953333a5b592c37dcf416ee81d8201c36cb05512aa4f
+size 30877129
diff --git a/data/occurrences/17_01_2023_session_2-DJI_0012.csv b/data/occurrences/17_01_2023_session_2-DJI_0012.csv
new file mode 100644
index 0000000000000000000000000000000000000000..ac462db0ecbe570be3df3cfa8d5c9a96046987f1
--- /dev/null
+++ b/data/occurrences/17_01_2023_session_2-DJI_0012.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c77e0793b2af2598b894b4fb8c4347dfa155b3624f55a0fa353f8ba76681910
+size 30969780
diff --git a/data/occurrences/17_01_2023_session_2-DJI_0013.csv b/data/occurrences/17_01_2023_session_2-DJI_0013.csv
new file mode 100644
index 0000000000000000000000000000000000000000..b0de72fc15382fe8558c151037e63d61c7df8587
--- /dev/null
+++ b/data/occurrences/17_01_2023_session_2-DJI_0013.csv
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e7e2f18c4ecf41b9d78419003846d9908a58bbc8bef14fffd538ee4b7cbee297
+size 23657660
diff --git a/data/session_events.csv b/data/session_events.csv
index a2d21a7de1f9771f08698626e5d80217a007c850..e91e405c11d84f1ca2ba0f1f07457a46d59450c4 100644
--- a/data/session_events.csv
+++ b/data/session_events.csv
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:86cf41400ed25b3d655d4072fd5c3c2e9ca016d349d0d40bc10a80fbfc9d3a3a
-size 19731
+oid sha256:413b6f40f961211e3becd6b34f5feb9850cf91ee4c9644fc52569ea5f36da65d
+size 22306
diff --git a/data/video_events.csv b/data/video_events.csv
index efc349b35a159a045833d8f4cf99d48cec56ff0e..f0b429b6d58b8dd9edb47d0866c4a754f76989ae 100644
--- a/data/video_events.csv
+++ b/data/video_events.csv
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0242eaf58c4fb0cff768e1d1b06c2ba3cba9fef0d2a23018032e3a52e1cf3585
-size 12616
+oid sha256:6a07a6876b5974f9c9e4973fb1ccc112b42c2f049667acc246597c6ecada28cf
+size 88171
diff --git a/metadata/DATA_DICTIONARY.md b/metadata/DATA_DICTIONARY.md
new file mode 100644
index 0000000000000000000000000000000000000000..5ed20566583d2b632037227506a73a5f6d26ff88
--- /dev/null
+++ b/metadata/DATA_DICTIONARY.md
@@ -0,0 +1,382 @@
+# KABR Behavior Telemetry Data Dictionary
+
+This document provides detailed descriptions of all data files and their fields in the KABR Behavior Telemetry dataset.
+
+## Dataset Files
+
+### Core Data Files
+
+#### `data/occurrences/`
+Frame-by-frame occurrence records for each video, combining detection tracks, behavior annotations, and telemetry data.
+
+**Files:** One CSV per video, named `{date}-{video_id}.csv` (e.g., `11_01_23-DJI_0977.csv`)
+
+**Key Fields:**
+- `date`: Recording date in DD_MM_YY format
+- `video_id`: DJI video identifier (e.g., DJI_0977)
+- `frame`: Frame number in the video sequence (0-indexed)
+- `date_time`: Timestamp in format "YYYY-MM-DD HH:MM:SS,milliseconds,microseconds"
+- `id`: Mini-scene identifier (track ID for behavioral sequences)
+- `latitude`: GPS latitude in decimal degrees (WGS84)
+- `longitude`: GPS longitude in decimal degrees (WGS84)
+- `altitude`: Altitude in meters above sea level
+- `iso`: Camera ISO setting
+- `shutter`: Shutter speed value
+- `fnum`: Aperture f-number
+- `ev`: Exposure value
+- `ct`: Color temperature
+- `color_md`: Color mode descriptor
+- `focal_len`: Lens focal length in millimeters
+- `dzoom_ratio`: Digital zoom ratio
+- `xtl`, `ytl`, `xbr`, `ybr`: Bounding box coordinates (top-left and bottom-right)
+- `z_order_x`, `z_order_y`: Display depth ordering for overlapping annotations
+- `label`: Object class label (e.g., "zebra", "giraffe")
+- `source`: Annotation source
+- `keyframe_x`, `keyframe_y`: Whether frame is a tracking keyframe
+- `outside_x`, `outside_y`: Whether object is outside frame bounds
+- `occluded_x`, `occluded_y`: Whether object is occluded
+- `points`: Polygon points for behavioral sequences
+- `behaviour`: Behavioral classification (e.g., "walking", "grazing", "running")
+
+**Record Count:** Varies by video; typically 10,000-66,000 frames per video
+**Coverage:** 47 videos with complete data
+**Missing Values:** Frames without detections/annotations have null values for annotation fields
+
+---
+
+#### `data/video_events.csv`
+Darwin Core Event records for individual video recordings.
+
+**Record Count:** 68 videos (47 with occurrence data)
+
+**Fields:**
+
+**Identifiers:**
+- `eventID`: Unique event identifier (format: "KABR-2023:{session}:{video_id}")
+- `parentEventID`: Links to session_events (format: "KABR-2023:{session}")
+
+**Event Classification:**
+- `eventType`: Type of sampling event ("video recording")
+- `eventDate`: Recording date (ISO 8601: YYYY-MM-DD)
+- `eventTime`: Video start time (HH:MM:SS)
+- `endTime`: Video end time (HH:MM:SS)
+
+**Geographic Information:**
+- `decimalLatitude`: Launch point latitude in decimal degrees (WGS84)
+- `decimalLongitude`: Launch point longitude in decimal degrees (WGS84)
+- `geodeticDatum`: Coordinate reference system ("WGS84")
+- `minimumElevationInMeters`: Minimum altitude during video
+- `maximumElevationInMeters`: Maximum altitude during video
+- `footprintWKT`: Geographic bounding box in Well-Known Text format
+
+**Sampling Protocol:**
+- `samplingProtocol`: "Continuous aerial video recording"
+
+**Associated Resources:**
+- `associatedMedia`: JSON object with paths to detection and behavior annotation files
+  ```json
+  {
+    "detection": "path/to/DJI_XXXX_tracks.xml",
+    "behavior": ["path/to/trackID.xml", ...]
+  }
+  ```
+
+**Remarks:**
+- `eventRemarks`: Description of video file (e.g., "Video file DJI_0977.MP4")
+
+---
+
+#### `data/session_events.csv`
+Darwin Core Event records for field sessions (missions/flights).
+
+**Record Count:** 17 sessions (14 with occurrence data)
+
+**Fields:**
+
+**Identifiers and Classification:**
+- `eventID`: Unique session identifier (format: "KABR-2023:{date}_{session}")
+- `parentEventID`: Dataset identifier ("KABR-2023")
+- `eventType`: Type of event ("drone survey")
+
+**Temporal Coverage:**
+- `eventDate`: Session date (ISO 8601: YYYY-MM-DD)
+- `eventTime`: Session start time (HH:MM:SS)
+- `endTime`: Session end time (HH:MM:SS)
+- `year`, `month`, `day`: Parsed date components
+
+**Geographic Coverage:**
+- `launchLatitude`: Drone launch point latitude (from first video)
+- `launchLongitude`: Drone launch point longitude (from first video)
+- `decimalLatitude`: Latitude range across session "[min, max]"
+- `decimalLongitude`: Longitude range across session "[min, max]"
+- `geodeticDatum`: "WGS84"
+- `coordinateUncertaintyInMeters`: GPS precision
+- `minimumElevationInMeters`: Minimum altitude across all videos
+- `maximumElevationInMeters`: Maximum altitude across all videos
+- `footprintWKT`: Session bounding box in WKT format
+- `locationID`: "MPALA-KENYA"
+- `locality`: "Mpala Research Centre"
+- `country`: "Kenya"
+- `countryCode`: "KE"
+
+**Habitat and Environment:**
+- `habitat`: Habitat type description (e.g., "Open grassy habitat with some scattered bushes")
+
+**Sampling Details:**
+- `samplingProtocol`: Detailed protocol description
+- `sampleSizeValue`: Empty (intended for area surveyed)
+- `sampleSizeUnit`: "minutes"
+- `samplingEffort`: Number and description of video segments
+
+**Taxonomic Information:**
+- `organismQuantity`: Number of individual animals observed
+- `organismQuantityType`: "herd size"
+
+**Observations:**
+- `eventRemarks`: Field notes and observations
+- `_species_common`: Common name(s) of observed species
+- `_species_commonNames`: Additional common names
+- `_n_individuals`: Number of individuals
+- `_n_videos`: Number of videos in session
+
+**Associated Data:**
+- `associatedMedia`: JSON object listing videos, focal follows, and scan samples
+  ```json
+  {
+    "video": ["DJI_0977", "DJI_0978", ...],
+    "focal": [],
+    "scan": []
+  }
+  ```
+- `associatedReferences`: Airdata telemetry file names
+- `_telemetry_file_raw`: Raw telemetry file status notes
+
+**Ecological Extensions (Humboldt Eco):**
+- `eco:inventoryTypes`: "restrictedSearch"
+- `eco:protocolNames`: "KABR Drone Video Survey Protocol"
+- `eco:protocolDescriptions`: Protocol description
+- `eco:protocolReferences`: "https://doi.org/10.48550/arXiv.2510.02030"
+- `eco:isAbundanceReported`: True
+- `eco:isAbundanceCapReported`: False
+- `eco:abundanceUnit`: "individuals"
+- `eco:isVegetationCoverReported`: True/False
+- `eco:vegetationCoverUnit`: "Bitterlich score (0-10 scale)" (when applicable)
+- `eco:isTaxonCompletenessReported`: True
+- `eco:taxonCompletenessProtocols`: "All visible target taxa annotated in video frames with bounding boxes"
+- `eco:isAbsenceReported`: False
+- `eco:hasNonTargetTaxa`: Empty
+- `eco:nonTargetTaxa`: Empty
+- `eco:targetTaxonomicScope`: Scientific names of target species
+- `eco:excludedTaxonomicScope`: Empty
+- `eco:samplingPerformedBy`: Empty
+- `eco:siteCount`: 1
+- `eco:siteNestingDescription`: "Single focal group follow per session"
+- `eco:verbatimSiteDescriptions`: Empty
+
+**Quality Flags:**
+- `_needs_verification`: Data quality flag
+
+---
+
+### Scripts
+
+#### `scripts/merge_behavior_telemetry.py`
+Merges SRT metadata, detection tracks, and behavior annotations into occurrence files.
+
+**Usage:**
+```bash
+python scripts/merge_behavior_telemetry.py \
+  --data_path /path/to/video/directories \
+  --outpath /path/to/output/ \
+  [--skip-airdata]
+```
+
+**Arguments:**
+- `--data_path`: Directory containing video folders (format: DATE-VIDEO_ID)
+- `--session_data_path`: Path to SRT files (default: preset path)
+- `--flight_logs_path`: Path to decrypted flight logs (default: preset path)
+- `--skip-airdata`: Skip merging with flight log data
+- `--write`: Whether to write output (default: True)
+- `--outpath`: Output directory for CSV files
+
+**Input Requirements:**
+- Video directories with structure:
+  - `metadata/{video_id}_tracks.xml`: Detection bounding boxes
+  - `actions/*.xml`: Behavior annotation files
+- SRT files with GPS and camera metadata
+- (Optional) Flight log CSV files with telemetry
+
+**Output:**
+- One CSV per video in occurrence format
+
+---
+
+#### `scripts/update_video_events.py`
+Updates video_events.csv with associatedMedia paths to detection and behavior files.
+
+**Usage:**
+```bash
+python scripts/update_video_events.py \
+  --video_events data/video_events.csv \
+  --data_path /path/to/video/directories \
+  [--output output_path.csv]
+```
+
+**What it does:**
+- Scans video directories for detection and behavior annotation files
+- Creates relative paths from kabr-behavior-telemetry/data to annotation files
+- Updates associatedMedia field with JSON structure
+
+---
+
+#### `scripts/add_event_times.py`
+Extracts start and end times from occurrence files and adds to video_events.csv.
+
+**Usage:**
+```bash
+python scripts/add_event_times.py \
+  --video_events data/video_events.csv \
+  --occurrences data/occurrences/ \
+  [--output output_path.csv]
+```
+
+**What it does:**
+- Reads date_time from first and last rows of each occurrence file
+- Extracts time component (HH:MM:SS)
+- Updates eventTime and endTime fields
+
+---
+
+#### `scripts/add_gps_data.py`
+Extracts GPS statistics from occurrence files and adds to event files.
+
+**Usage:**
+```bash
+python scripts/add_gps_data.py \
+  --video_events data/video_events.csv \
+  --session_events data/session_events.csv \
+  --occurrences data/occurrences/ \
+  [--output_video output_video.csv] \
+  [--output_session output_session.csv]
+```
+
+**What it does:**
+
+For **video_events.csv**:
+- Extracts launch point (first GPS coordinate)
+- Calculates min/max lat/lon bounds
+- Determines altitude range
+- Creates WKT footprint
+
+For **session_events.csv**:
+- Uses first video's launch point as session launch
+- Aggregates bounds across all videos in session
+- Formats lat/lon as "[min, max]" ranges
+- Creates session-level WKT footprint
+
+---
+
+## Data Relationships
+
+### Hierarchical Structure
+```
+KABR-2023 (Dataset)
+├── KABR-2023:11_01_23_session_1 (Session)
+│   ├── KABR-2023:11_01_23_session_1:DJI_0488 (Video Event)
+│   │   └── occurrences/11_01_23-DJI_0488.csv (Frame records)
+│   └── ...
+├── KABR-2023:11_01_23_session_2 (Session)
+│   ├── KABR-2023:11_01_23_session_2:DJI_0977 (Video Event)
+│   │   └── occurrences/11_01_23-DJI_0977.csv (Frame records)
+│   └── ...
+...
+```
+
+### Linkages
+- **session_events** ← linked by parentEventID ← **video_events**
+- **video_events** ← associatedMedia paths ← **annotation files**
+- **video_events** → referenced by occurrence files → **occurrences/**
+
+---
+
+## Data Processing Pipeline
+
+1. **Field Collection**: Drone flights with video recording
+2. **Video Processing**: Frame extraction and annotation
+3. **Telemetry Extraction**: SRT files parsed for GPS and camera metadata
+4. **Annotation**: Detection boxes and behavior labels added
+5. **Merging** (`merge_behavior_telemetry.py`): Combine all data sources
+6. **Event Creation**: Generate Darwin Core event records
+7. **Metadata Enhancement**:
+   - Add associatedMedia paths (`update_video_events.py`)
+   - Add temporal bounds (`add_event_times.py`)
+   - Add GPS statistics (`add_gps_data.py`)
+
+---
+
+## Darwin Core Compliance
+
+This dataset follows Darwin Core standards (TDWG) with extensions for ecological inventory data (Humboldt Eco).
+
+**Core Classes:**
+- **Event**: Recording sessions and individual videos
+- **Occurrence**: Frame-level animal detections (in occurrence files)
+
+**Key Standards:**
+- Temporal data in ISO 8601 format
+- Coordinates in WGS84 decimal degrees
+- Controlled vocabularies for taxonomic names
+- Unique identifiers for all records
+
+---
+
+## File Formats
+
+### CSV Files
+- **Encoding**: UTF-8
+- **Delimiter**: Comma (`,`)
+- **Quoting**: Double quotes for fields containing delimiters
+- **Missing Values**: Empty strings or `NaN`
+
+### JSON Fields
+- **Structure**: Valid JSON objects in CSV fields
+- **Encoding**: Double-quotes escaped as `""`
+- **Arrays**: Square bracket notation `["item1", "item2"]`
+
+### WKT (Well-Known Text)
+- **Format**: `POLYGON((lon1 lat1, lon2 lat2, ...))`
+- **Coordinate Order**: Longitude first, then latitude
+- **Datum**: WGS84 (EPSG:4326)
+
+---
+
+## Quality Assurance
+
+### Known Limitations
+- **Missing Occurrence Data**: 21 videos lack occurrence files due to:
+  - Missing SRT files
+  - Corrupted source data
+  - Processing errors
+  - Empty annotations
+
+- **GPS Accuracy**: ±5-10 meters typical
+- **Timestamp Precision**: Millisecond resolution but may have ±1 second drift
+
+### Data Quality Flags
+- `_needs_verification`: Manual review recommended
+- `_telemetry_file_raw`: Notes on telemetry file status
+
+---
+
+## Version History
+
+- **v1.1** (2026-01-02): Added GPS data, event times, and occurrence files
+- **v1.0** (2024-12-31): Initial release with session and video events
+
+---
+
+## Contact
+
+For questions about this data:
+- **Email**: kline.377@osu.edu
+- **Issues**: [GitHub repository](https://github.com/Imageomics/kabr-behavior-telemetry)
diff --git a/scripts/add_event_times.py b/scripts/add_event_times.py
new file mode 100644
index 0000000000000000000000000000000000000000..4d7e6a6ad33cafe0cdd84cbc045ebe976d38cf54
--- /dev/null
+++ b/scripts/add_event_times.py
@@ -0,0 +1,119 @@
+import pandas as pd
+import os
+from datetime import datetime
+
+def add_event_times(
+    video_events_path,
+    occurrences_path,
+    output_path=None
+):
+    """
+    Update video_events.csv with eventTime and endTime from occurrence files.
+
+    Args:
+        video_events_path: Path to video_events.csv
+        occurrences_path: Path to occurrences directory
+        output_path: Path to write updated CSV (if None, overwrites input)
+    """
+    # Read video_events.csv
+    df = pd.read_csv(video_events_path)
+
+    # Parse the eventID to extract video_id
+    for idx, row in df.iterrows():
+        event_id = row['eventID']
+        parts = event_id.split(':')
+
+        if len(parts) < 3:
+            print(f"Warning: Could not parse eventID: {event_id}")
+            continue
+
+        date_session = parts[1]
+        video_id = parts[2]
+
+        # Extract the date portion (without session)
+        date_parts = date_session.split('_session_')
+        if len(date_parts) > 1:
+            date_part = date_parts[0]
+        else:
+            date_part = date_session
+
+        # Construct the occurrence filename
+        occurrence_file = f"{date_part}-{video_id}.csv"
+        occurrence_path = os.path.join(occurrences_path, occurrence_file)
+
+        if not os.path.exists(occurrence_path):
+            print(f"⚠ {video_id}: No occurrence file found")
+            continue
+
+        try:
+            # Read the occurrence file
+            occ_df = pd.read_csv(occurrence_path)
+
+            if 'date_time' not in occ_df.columns or occ_df.empty:
+                print(f"⚠ {video_id}: No date_time data")
+                continue
+
+            # Get first and last non-null date_time values
+            date_times = occ_df['date_time'].dropna()
+
+            if date_times.empty:
+                print(f"⚠ {video_id}: All date_time values are null")
+                continue
+
+            # Extract the first and last timestamps
+            # Format: "2023-01-11 16:04:03,114,286"
+            first_dt_str = str(date_times.iloc[0])
+            last_dt_str = str(date_times.iloc[-1])
+
+            # Parse to extract just the time portion (HH:MM:SS)
+            first_time = first_dt_str.split(',')[0].split(' ')[1] if ' ' in first_dt_str else None
+            last_time = last_dt_str.split(',')[0].split(' ')[1] if ' ' in last_dt_str else None
+
+            if first_time and last_time:
+                # Update the dataframe
+                df.at[idx, 'eventTime'] = first_time
+                df.at[idx, 'endTime'] = last_time
+                print(f"✓ {video_id}: {first_time} - {last_time}")
+            else:
+                print(f"⚠ {video_id}: Could not parse time")
+
+        except Exception as e:
+            print(f"✗ {video_id}: Error - {str(e)}")
+
+    # Write the updated CSV
+    if output_path is None:
+        output_path = video_events_path
+
+    df.to_csv(output_path, index=False)
+    print(f"\nUpdated video_events.csv written to: {output_path}")
+
+if __name__ == "__main__":
+    import argparse
+
+    parser = argparse.ArgumentParser(description="Add event times to video_events.csv from occurrence files")
+    parser.add_argument(
+        "--video_events",
+        type=str,
+        required=True,
+        help="Path to video_events.csv"
+    )
+    parser.add_argument(
+        "--occurrences",
+        type=str,
+        required=True,
+        help="Path to occurrences directory"
+    )
+    parser.add_argument(
+        "--output",
+        type=str,
+        default=None,
+        help="Output path (default: overwrites input)"
+    )
+
+    args = parser.parse_args()
+
+    add_event_times(
+        args.video_events,
+        args.occurrences,
+        args.output
+    )
diff --git a/scripts/add_gps_data.py b/scripts/add_gps_data.py
new file mode 100644
index 0000000000000000000000000000000000000000..35b090a34365c21e13d6a6e8e82d43e71db741e6
--- /dev/null
+++ b/scripts/add_gps_data.py
@@ -0,0 +1,252 @@
+import pandas as pd
+import numpy as np
+import os
+import json
+
+def extract_gps_from_occurrence(occurrence_path):
+    """
+    Extract GPS statistics from an occurrence file.
+
+    Returns:
+        dict with keys: launch_lat, launch_lon, min_lat, max_lat, min_lon, max_lon, min_alt, max_alt
+    """
+    try:
+        # Read occurrence file with low_memory=False to avoid dtype warnings
+        occ_df = pd.read_csv(occurrence_path, low_memory=False)
+
+        if occ_df.empty:
+            return None
+
+        # Get GPS columns
+        lat_col = occ_df['latitude'].dropna()
+        lon_col = occ_df['longitude'].dropna()
+        alt_col = occ_df['altitude'].dropna()
+
+        if lat_col.empty or lon_col.empty:
+            return None
+
+        # Launch point is the first GPS coordinate
+        launch_lat = float(lat_col.iloc[0])
+        launch_lon = float(lon_col.iloc[0])
+
+        # Calculate min/max
+        stats = {
+            'launch_lat': launch_lat,
+            'launch_lon': launch_lon,
+            'min_lat': float(lat_col.min()),
+            'max_lat': float(lat_col.max()),
+            'min_lon': float(lon_col.min()),
+            'max_lon': float(lon_col.max()),
+        }
+
+        # Add altitude if available
+        if not alt_col.empty:
+            stats['min_alt'] = float(alt_col.min())
+            stats['max_alt'] = float(alt_col.max())
+        else:
+            stats['min_alt'] = None
+            stats['max_alt'] = None
+
+        return stats
+
+    except Exception as e:
+        print(f"Error processing {occurrence_path}: {str(e)}")
+        return None
+
+
+def add_gps_to_video_events(video_events_path, occurrences_path, output_path=None):
+    """
+    Add GPS columns to video_events.csv from occurrence files.
+
+    Adds columns:
+    - decimalLatitude (launch point)
+    - decimalLongitude (launch point)
+    - minimumElevationInMeters
+    - maximumElevationInMeters
+    - footprintWKT (bounding box in WKT format)
+    """
+    # Read video_events.csv
+    df = pd.read_csv(video_events_path)
+
+    # Add new columns if they don't exist
+    new_columns = ['decimalLatitude', 'decimalLongitude',
+                   'minimumElevationInMeters', 'maximumElevationInMeters',
+                   'footprintWKT']
+
+    for col in new_columns:
+        if col not in df.columns:
+            df[col] = np.nan
+
+    # Process each video
+    for idx, row in df.iterrows():
+        event_id = row['eventID']
+        parts = event_id.split(':')
+
+        if len(parts) < 3:
+            continue
+
+        date_session = parts[1]
+        video_id = parts[2]
+
+        # Extract the date portion
+        date_parts = date_session.split('_session_')
+        date_part = date_parts[0] if len(date_parts) > 1 else date_session
+
+        # Construct occurrence filename
+        # Try with underscore first (for flight_1, flight_2 format)
+        occurrence_file = f"{date_part}_{video_id}.csv"
+        occurrence_path = os.path.join(occurrences_path, occurrence_file)
+
+        # If that doesn't exist, try with dash (for older format)
+        if not os.path.exists(occurrence_path):
+            occurrence_file = f"{date_part}-{video_id}.csv"
+            occurrence_path = os.path.join(occurrences_path, occurrence_file)
+
+        if not os.path.exists(occurrence_path):
+            print(f"⚠ {video_id}: No occurrence file")
+            continue
+
+        # Extract GPS data
+        gps_stats = extract_gps_from_occurrence(occurrence_path)
+
+        if gps_stats is None:
+            print(f"⚠ {video_id}: No GPS data")
+            continue
+
+        # Update video_events
+        df.at[idx, 'decimalLatitude'] = gps_stats['launch_lat']
+        df.at[idx, 'decimalLongitude'] = gps_stats['launch_lon']
+
+        if gps_stats['min_alt'] is not None:
+            df.at[idx, 'minimumElevationInMeters'] = gps_stats['min_alt']
+            df.at[idx, 'maximumElevationInMeters'] = gps_stats['max_alt']
+
+        # Create WKT footprint (bounding box)
+        wkt = f"POLYGON(({gps_stats['min_lon']} {gps_stats['min_lat']}, " \
+              f"{gps_stats['max_lon']} {gps_stats['min_lat']}, " \
+              f"{gps_stats['max_lon']} {gps_stats['max_lat']}, " \
+              f"{gps_stats['min_lon']} {gps_stats['max_lat']}, " \
+              f"{gps_stats['min_lon']} {gps_stats['min_lat']}))"
+        df.at[idx, 'footprintWKT'] = wkt
+
+        print(f"✓ {video_id}: Launch ({gps_stats['launch_lat']:.6f}, {gps_stats['launch_lon']:.6f}), "
+              f"Bounds: lat[{gps_stats['min_lat']:.6f}, {gps_stats['max_lat']:.6f}], "
+              f"lon[{gps_stats['min_lon']:.6f}, {gps_stats['max_lon']:.6f}]")
+
+    # Write updated CSV
+    if output_path is None:
+        output_path = video_events_path
+
+    df.to_csv(output_path, index=False)
+    print(f"\nUpdated video_events.csv written to: {output_path}")
+    return df
+
+
+def add_gps_to_session_events(session_events_path, video_events_df, output_path=None):
+    """
+    Add GPS columns to session_events.csv by aggregating from video_events.
+
+    For each session:
+    - launchLatitude/launchLongitude: Launch point of first video in session
+    - decimalLatitude: [min, max] latitude range as string
+    - decimalLongitude: [min, max] longitude range as string
+    - footprintWKT: Bounding box encompassing all videos in session
+    - minimumElevationInMeters/maximumElevationInMeters: Min/max across all videos
+    """
+    # Read session_events.csv
+    session_df = pd.read_csv(session_events_path)
+
+    # Add new columns if they don't exist
+    new_columns = ['launchLatitude', 'launchLongitude',
+                   'minimumElevationInMeters', 'maximumElevationInMeters',
+                   'footprintWKT']
+
+    for col in new_columns:
+        if col not in session_df.columns:
+            session_df[col] = np.nan
+
+    # Process each session
+    for idx, row in session_df.iterrows():
+        session_id = row['eventID']
+
+        # Get all videos for this session
+        session_videos = video_events_df[video_events_df['parentEventID'] == session_id]
+
+        if session_videos.empty:
+            print(f"⚠ {session_id}: No videos found")
+            continue
+
+        # Filter videos with GPS data
+        videos_with_gps = session_videos.dropna(subset=['decimalLatitude', 'decimalLongitude'])
+
+        if videos_with_gps.empty:
+            print(f"⚠ {session_id}: No GPS data in videos")
+            continue
+
+        # Launch point from first video
+        first_video = videos_with_gps.iloc[0]
+        session_df.at[idx, 'launchLatitude'] = first_video['decimalLatitude']
+        session_df.at[idx, 'launchLongitude'] = first_video['decimalLongitude']
+
+        # Calculate session-level min/max across all videos
+        min_lat = videos_with_gps['decimalLatitude'].min()
+        max_lat = videos_with_gps['decimalLatitude'].max()
+        min_lon = videos_with_gps['decimalLongitude'].min()
+        max_lon = videos_with_gps['decimalLongitude'].max()
+
+        # Set decimalLatitude and decimalLongitude to [min, max] ranges
+        session_df.at[idx, 'decimalLatitude'] = f"[{min_lat:.6f}, {max_lat:.6f}]"
+        session_df.at[idx, 'decimalLongitude'] = f"[{min_lon:.6f}, {max_lon:.6f}]"
+
+        # Aggregate elevation
+        if 'minimumElevationInMeters' in videos_with_gps.columns:
+            elev_videos = videos_with_gps.dropna(subset=['minimumElevationInMeters'])
+            if not elev_videos.empty:
+                session_df.at[idx, 'minimumElevationInMeters'] = elev_videos['minimumElevationInMeters'].min()
+                session_df.at[idx, 'maximumElevationInMeters'] = elev_videos['maximumElevationInMeters'].max()
+
+        # Create session footprint
+        wkt = f"POLYGON(({min_lon} {min_lat}, {max_lon} {min_lat}, " \
+              f"{max_lon} {max_lat}, {min_lon} {max_lat}, {min_lon} {min_lat}))"
+        session_df.at[idx, 'footprintWKT'] = wkt
+
+        print(f"✓ {session_id.split(':')[1]}: Launch ({first_video['decimalLatitude']:.6f}, {first_video['decimalLongitude']:.6f}), "
+              f"Session bounds: lat[{min_lat:.6f}, {max_lat:.6f}], lon[{min_lon:.6f}, {max_lon:.6f}]")
+
+    # Write updated CSV
+    if output_path is None:
+        output_path = session_events_path
+
+    session_df.to_csv(output_path, index=False)
+    print(f"\nUpdated session_events.csv written to: {output_path}")
+
+
+def main():
+    import argparse
+
+    parser = argparse.ArgumentParser(description="Add GPS data to video_events and session_events")
+    parser.add_argument("--video_events", type=str, required=True, help="Path to video_events.csv")
+    parser.add_argument("--session_events", type=str, required=True, help="Path to session_events.csv")
+    parser.add_argument("--occurrences", type=str, required=True, help="Path to occurrences directory")
+    parser.add_argument("--output_video", type=str, default=None, help="Output path for video_events (default: overwrite)")
+    parser.add_argument("--output_session", type=str, default=None, help="Output path for session_events (default: overwrite)")
+
+    args = parser.parse_args()
+
+    print("=" * 80)
+    print("STEP 1: Adding GPS data to video_events.csv")
+    print("=" * 80)
+    video_df = add_gps_to_video_events(args.video_events, args.occurrences, args.output_video)
+
+    print("\n" + "=" * 80)
+    print("STEP 2: Adding GPS data to session_events.csv")
+    print("=" * 80)
+    add_gps_to_session_events(args.session_events, video_df, args.output_session)
+
+    print("\n" + "=" * 80)
+    print("DONE!")
+    print("=" * 80)
+
+
+if __name__ == "__main__":
+    main()
diff --git a/scripts/merge_behavior_telemetry.py b/scripts/merge_behavior_telemetry.py
new file mode 100644
index 0000000000000000000000000000000000000000..71c577111348db8c5743daf45368c2ec7306deaa
--- /dev/null
+++ b/scripts/merge_behavior_telemetry.py
@@ -0,0 +1,353 @@
+import re
+import os
+import json
+import pysrt
+import argparse
+import pandas as pd
+from tqdm import tqdm
+from glob import glob
+from datetime import datetime
+import xml.etree.ElementTree as ET
+
+" Based on script authored by Otto Brookes for KABR-2023 project "
+
+
+def pandify_xml_tracks(path2tracks):
+    elems = []
+    et = ET.parse(path2tracks)
+    root = et.getroot()
+    for row in root:
+        for e in row.iter("box"):
+            for k, v in row.attrib.items():
+                e.attrib[k] = v
+            elems.append(e.attrib)
+    track_df = pd.DataFrame(elems)
+    track_df["frame"] = track_df.frame.astype(int)
+    return track_df
+
+
+def extract_frame_no(text):
+    pattern = r": (\d+),"
+    matches = re.findall(pattern, text)
+    numbers = [int(match) for match in matches]
+    assert len(numbers) == 1, "Frame index must be unique"
+    return next(iter(numbers))
+
+
+def extract_meta_data(text):
+    # Extract all [text]
+    pattern = r"\[(.*?)\]"
+    matches = re.findall(pattern, text)
+    data_dict = {}
+    for item in matches:
+        key_value = item.split(":", 1)
+        key = key_value[0].strip()
+        value = key_value[1].strip()
+        data_dict[key] = value
+    return data_dict
+
+
+def pandify_srt_data(path2srt):
+    subs = pysrt.open(path2srt)
+    all_meta_data = []
+    for s in subs:
+        split_text = s.text.split("\n")
+        meta_data = extract_meta_data(split_text[2])
+        meta_data["frame"] = extract_frame_no(split_text[0])
+        meta_data["date_time"] = split_text[1]
+        all_meta_data.append(meta_data)
+    srt_df = pd.DataFrame(all_meta_data)
+    srt_df["frame"] = srt_df["frame"] - 1
+    return srt_df
+
+
+def get_per_frame_annotations(path2xml):
+    et = ET.parse(path2xml)
+    root = et.getroot()
+    per_frame_annotations = []
+    for row in root.findall("track"):
+        for e, j in zip(row.iter("points"), row.iter("attribute")):
+            behaviour = j.text
+            e.attrib["behaviour"] = behaviour
+            per_frame_annotations.append(e.attrib)
+    return per_frame_annotations
+
+
+def add_per_frame_behaviours(merged_df, path2annotations):
+    mini_scenes_df = None
+    ms_annotations = glob(f"{path2annotations}/**/*.xml", recursive=True)
+    for ms in ms_annotations:
+        ms_index = ms.split("/")[-1].split(".")[0]
+        ms_df = merged_df[merged_df.id == str(ms_index)].sort_values(by="frame")
+        first_frame = ms_df.frame.iloc[0]  # ugly - rework later
+        per_frame_anns = pd.DataFrame(get_per_frame_annotations(ms))
+        per_frame_anns["frame"] = per_frame_anns.frame.astype(int) + first_frame
+        ms_df = ms_df.merge(per_frame_anns, on="frame")
+        if mini_scenes_df is None:
+            mini_scenes_df = ms_df
+        else:
+            mini_scenes_df = pd.concat([mini_scenes_df, ms_df])
+    return mini_scenes_df
+
+
+def find_srt_file(session_data_root, date_part, filename):
+    """
+    Recursively search for SRT file matching the date and filename.
+
+    Args:
+        session_data_root: Root path to session_data directory
+        date_part: Date portion of the directory name (e.g., '11_01_23' or '17_01_2023_session_1')
+        filename: DJI filename (e.g., 'DJI_0488')
+
+    Returns:
+        Path to SRT file if found, None otherwise
+    """
+    # Try to find the date directory in session_data
+    date_dir = os.path.join(session_data_root, date_part)
+    if not os.path.exists(date_dir):
+        # Try without session suffix for cases like '16_01_23_session_1' -> '16_01_23'
+        base_date = date_part.split('_session_')[0]
+        date_dir = os.path.join(session_data_root, base_date)
+
+    if not os.path.exists(date_dir):
+        print(f"Warning: Could not find date directory for {date_part}")
+        return None
+
+    # Recursively search for the SRT file
+    srt_filename = f"{filename}.SRT"
+    for root, dirs, files in os.walk(date_dir):
+        if srt_filename in files:
+            return os.path.join(root, srt_filename)
+
+    return None
+
+
+def find_flight_log(flight_logs_path, srt_df):
+    """
+    Find the matching flight log CSV based on datetime from SRT data.
+
+    Args:
+        flight_logs_path: Path to decrypted_flight_logs directory
+        srt_df: DataFrame with SRT data containing date_time column
+
+    Returns:
+        Path to matching flight log CSV, or None if not found
+    """
+    if srt_df.empty or 'date_time' not in srt_df.columns:
+        return None
+
+    # Get first datetime from SRT (format: "2023-01-11 16:04:03,681,492")
+    first_datetime_str = srt_df['date_time'].iloc[0]
+    # Parse just the date and time part (ignore milliseconds)
+    srt_datetime = datetime.strptime(first_datetime_str.split(',')[0], "%Y-%m-%d %H:%M:%S")
+
+    # Search for matching flight log
+    flight_logs = glob(f"{flight_logs_path}/*.csv")
+
+    for log_path in flight_logs:
+        # Read full file to get complete time range (many files are small)
+        try:
+            log_df = pd.read_csv(log_path)
+            if 'datetime(utc)' not in log_df.columns or log_df.empty:
+                continue
+
+            # Convert to datetime and add 3 hours (flight logs are 3 hours behind)
+            log_df['datetime_corrected'] = pd.to_datetime(log_df['datetime(utc)']) + pd.Timedelta(hours=3)
+
+            # Check if SRT datetime falls within flight log timerange
+            log_start = log_df['datetime_corrected'].min()
+            log_end = log_df['datetime_corrected'].max()
+
+            # Skip if dates are invalid
+            if pd.isna(log_start) or pd.isna(log_end):
+                continue
+
+            if log_start <= srt_datetime <= log_end:
+                return log_path
+        except Exception as e:
+            continue
+
+    return None
+
+
+def merge_flight_log_data(merged_df, flight_log_path):
+    """
+    Merge flight log data with the main dataframe based on datetime.
+
+    Args:
+        merged_df: Main dataframe with date_time column
+        flight_log_path: Path to flight log CSV
+
+    Returns:
+        Merged dataframe with flight log data
+    """
+    if flight_log_path is None or not os.path.exists(flight_log_path):
+        return merged_df
+
+    try:
+        # Read flight log
+        flight_df = pd.read_csv(flight_log_path)
+
+        # Prepare datetime columns for merging
+        # SRT format: "2023-01-11 16:04:03,681,492" -> convert to "2023-01-11 16:04:03"
+        merged_df['datetime_merge'] = merged_df['date_time'].apply(
+            lambda x: x.split(',')[0] if pd.notna(x) else None
+        )
+
+        # Flight log format: "2023-01-11 07:45:46"
+        # IMPORTANT: Flight log datetimes are 3 hours behind actual time - add 3 hours
+        flight_df['datetime_merge'] = pd.to_datetime(flight_df['datetime(utc)']) + pd.Timedelta(hours=3)
+
+        # Merge on datetime
+        merged_df['datetime_merge'] = pd.to_datetime(merged_df['datetime_merge'])
+
+        # Use merge_asof for nearest time matching
+        merged_df = merged_df.sort_values('datetime_merge')
+        flight_df = flight_df.sort_values('datetime_merge')
+
+        # Merge with flight log data
+        result_df = pd.merge_asof(
+            merged_df,
+            flight_df,
+            on='datetime_merge',
+            direction='nearest',
+            tolerance=pd.Timedelta('2s'),  # Increased tolerance to 2 seconds
+            suffixes=('', '_flight')
+        )
+
+        # Drop temporary merge column and handle duplicate columns
+        result_df = result_df.drop('datetime_merge', axis=1)
+
+        # Remove duplicate latitude/longitude/altitude columns from flight log if they exist
+        # Keep the SRT versions (more accurate for video frames)
+        for col in ['latitude', 'longitude', 'altitude']:
+            if f'{col}_flight' in result_df.columns:
+                result_df = result_df.drop(f'{col}_flight', axis=1)
+
+        print(f"  Merged with flight log: {os.path.basename(flight_log_path)}")
+        return result_df
+
+    except Exception as e:
+        print(f"  Warning: Could not merge flight log: {str(e)}")
+        if 'datetime_merge' in merged_df.columns:
+            merged_df = merged_df.drop('datetime_merge', axis=1)
+        return merged_df
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--data_path",
+        type=str,
+        help="Please use the full path to the data dir in this repo!",
+    )
+    parser.add_argument(
+        "--session_data_path",
+        type=str,
+        default="/fs/ess/PAS2136/Kenya-2023/Zebras/session_data",
+        help="Path to session_data directory containing SRT files",
+    )
+    parser.add_argument(
+        "--flight_logs_path",
+        type=str,
+        default="/fs/ess/PAS2136/Kenya-2023/Zebras/Flight_Logs/decrypted_flight_logs",
+        help="Path to decrypted_flight_logs directory",
+    )
+    parser.add_argument(
+        "--skip-airdata",
+        action="store_true",
+        help="Skip merging with airdata/flight log files",
+    )
+    parser.add_argument("--write", type=bool, default=True)
+    parser.add_argument("--outpath", type=str, help="Path to write csvs to")
+    args = parser.parse_args()
+
+    path2data = args.data_path
+    session_data_root = args.session_data_path
+    flight_logs_path = args.flight_logs_path
+    data_dirs = [x for x in os.listdir(path2data) if not x.startswith(".")]
+    path2write = args.outpath
+
+    good = 0
+    fail = 0
+    failed_files = []
+
+    for d in tqdm(data_dirs):
+        try:
+            # Parse directory name to get date and filename
+            # Format: DATE-FILENAME (e.g., '11_01_23-DJI_0488' or '17_01_2023_session_1-DJI_0005')
+            parts = d.split("-")
+            date_part = parts[0]
+            filename = parts[-1]
+
+            # Formulate paths
+            path2tracks = f"{path2data}/{d}/metadata/{filename}_tracks.xml"
+            path2annotations = f"{path2data}/{d}/actions/"
+
+            # Find SRT file recursively
+            path2srt = find_srt_file(session_data_root, date_part, filename)
+
+            if path2srt is None:
+                raise FileNotFoundError(f"Could not find SRT file for {date_part}-{filename}")
+
+            print(f"Processing {d}: Found SRT at {path2srt}")
+
+            # initialise dfs:
+            srt_df = pandify_srt_data(path2srt)
+            track_df = pandify_xml_tracks(path2tracks)
+            merged_df = srt_df.merge(track_df, on="frame", how="left")
+
+            # Add date and video_id columns to ALL rows
+            merged_df.insert(0, "date", date_part)
+            merged_df.insert(1, "video_id", filename)
+
+            # Move frame to position 2
+            frame_col = merged_df.pop("frame")
+            merged_df.insert(2, "frame", frame_col)
+
+            # Move id (mini-scene id) to position 3
+            if "id" in merged_df.columns:
+                id_col = merged_df.pop("id")
+                merged_df.insert(3, "id", id_col)
+
+            # Ensure date_time is preserved (move to position 4)
+            if "date_time" in merged_df.columns:
+                datetime_col = merged_df.pop("date_time")
+                merged_df.insert(4, "date_time", datetime_col)
+
+            # Find and merge flight log data if path provided and not skipped
+            if flight_logs_path and not args.skip_airdata:
+                flight_log_path = find_flight_log(flight_logs_path, srt_df)
+                if flight_log_path:
+                    merged_df = merge_flight_log_data(merged_df, flight_log_path)
+
+            # Add per frame behaviours to existing df
+            mini_scene_df = add_per_frame_behaviours(merged_df, path2annotations)
+
+            # Merge with frame df to preserve all frames (including those without annotations)
+            frame_df = merged_df[['date', 'video_id', 'frame', 'date_time']]
+            mini_scene_df = frame_df.merge(
+                mini_scene_df, on="frame", how="left"
+            )
+
+            # Remove duplicate date/video_id columns if they exist
+            for col in ['date_x', 'date_y', 'video_id_x', 'video_id_y', 'date_time_x', 'date_time_y']:
+                if col in mini_scene_df.columns:
+                    # Keep the non-null version
+                    base_col = col.rsplit('_', 1)[0]
+                    if f'{base_col}_x' in mini_scene_df.columns and f'{base_col}_y' in mini_scene_df.columns:
+                        mini_scene_df[base_col] = mini_scene_df[f'{base_col}_x'].fillna(mini_scene_df[f'{base_col}_y'])
+                        mini_scene_df = mini_scene_df.drop([f'{base_col}_x', f'{base_col}_y'], axis=1)
+
+            if args.write:
+                mini_scene_df.sort_values(by="frame").to_csv(path2write+f"{d}.csv", index=False)
+            good += 1
+        except Exception as e:
+            failed_files.append(d)
+            print(f"Failed on {d}: {str(e)}")
+            fail += 1
+    print("Pass: ", good, "Fail: ", fail)
+    print("Failed files:", failed_files)
+
+if __name__ == "__main__":
+    main()
\ No newline at end of file
diff --git a/scripts/update_video_events.py b/scripts/update_video_events.py
new file mode 100644
index 0000000000000000000000000000000000000000..3fa7faec3ad4cc85dc04d30f05dd48ce3626da3a
--- /dev/null
+++ b/scripts/update_video_events.py
@@ -0,0 +1,119 @@
+import pandas as pd
+import json
+import os
+from pathlib import Path
+
+def update_video_events(
+    video_events_path,
+    data_path,
+    output_path=None
+):
+    """
+    Update video_events.csv with associatedMedia paths for detections and behavior annotations.
+
+    Args:
+        video_events_path: Path to video_events.csv
+        data_path: Path to the data directory containing video directories
+        output_path: Path to write updated CSV (if None, overwrites input)
+    """
+    # Read video_events.csv
+    df = pd.read_csv(video_events_path)
+
+    # Parse the eventID to extract date and video_id
+    # Format: KABR-2023:DATE_SESSION:VIDEO_ID
+    for idx, row in df.iterrows():
+        event_id = row['eventID']
+        parts = event_id.split(':')
+
+        if len(parts) < 3:
+            print(f"Warning: Could not parse eventID: {event_id}")
+            continue
+
+        date_session = parts[1]
+        video_id = parts[2]
+
+        # Extract the date portion (without session)
+        # e.g., "11_01_23_session_1" -> "11_01_23"
+        date_parts = date_session.split('_session_')
+        if len(date_parts) > 1:
+            date_part = date_parts[0]
+        else:
+            date_part = date_session
+
+        # Construct the directory name
+        dir_name = f"{date_part}-{video_id}"
+
+        # Build paths to detections and behavior files
+        detections_path = os.path.join(data_path, dir_name, "metadata", f"{video_id}_tracks.xml")
+
+        # For behavior annotations, we need to find all XML files in the actions directory
+        actions_dir = os.path.join(data_path, dir_name, "actions")
+        behavior_files = []
+
+        if os.path.exists(actions_dir):
+            behavior_files = [f for f in os.listdir(actions_dir) if f.endswith('.xml')]
+            behavior_files.sort()  # Sort for consistency
+
+        # Check if files exist
+        detections_exists = os.path.exists(detections_path)
+
+        # Create relative paths from the kabr-behavior-telemetry/data directory
+        detections_rel = f"../../../mini-scenes_zebras/kabr-datapalooza-2023/data/{dir_name}/metadata/{video_id}_tracks.xml" if detections_exists else ""
+
+        # Create behavior annotations list with relative paths
+        behavior_rel_list = []
+        if behavior_files:
+            for bf in behavior_files:
+                behavior_rel_list.append(f"../../../mini-scenes_zebras/kabr-datapalooza-2023/data/{dir_name}/actions/{bf}")
+
+        # Update the associatedMedia field with JSON structure
+        associated_media = {
+            "detection": detections_rel,
+            "behavior": behavior_rel_list
+        }
+
+        # Update the dataframe
+        df.at[idx, 'associatedMedia'] = json.dumps(associated_media)
+
+        # Print status
+        status = "✓" if detections_exists else "✗"
+        behavior_count = len(behavior_files) if behavior_files else 0
+        print(f"{status} {video_id}: detections={detections_exists}, behaviors={behavior_count}")
+
+    # Write the updated CSV
+    if output_path is None:
+        output_path = video_events_path
+
+    df.to_csv(output_path, index=False)
+    print(f"\nUpdated video_events.csv written to: {output_path}")
+
+if __name__ == "__main__":
+    import argparse
+
+    parser = argparse.ArgumentParser(description="Update video_events.csv with associatedMedia paths")
+    parser.add_argument(
+        "--video_events",
+        type=str,
+        required=True,
+        help="Path to video_events.csv"
+    )
+    parser.add_argument(
+        "--data_path",
+        type=str,
+        required=True,
+        help="Path to data directory containing video directories"
+    )
+    parser.add_argument(
+        "--output",
+        type=str,
+        default=None,
+        help="Output path (default: overwrites input)"
+    )
+
+    args = parser.parse_args()
+
+    update_video_events(
+        args.video_events,
+        args.data_path,
+        args.output
+    )