diff --git a/msc_pygeoapi/loader/thunderstorm_outlook.py b/msc_pygeoapi/loader/thunderstorm_outlook.py
index 2459527d..378524f3 100644
--- a/msc_pygeoapi/loader/thunderstorm_outlook.py
+++ b/msc_pygeoapi/loader/thunderstorm_outlook.py
@@ -107,52 +107,68 @@ def generate_geojson_features(self):
             data = json.load(f)['features']
 
         features = []
-        for feature in data:
-
-            # flatten metobject properties
-            metobj = feature['properties']['metobject']
-            for item, values in metobj.items():
-                try:
-                    metobj_flat_item = self.flatten_json(item, values,
-                                                         'metobject')
-                    feature['properties'].update(metobj_flat_item)
-                    filename = f.name.split('/')[-1]
-                    file_id = re.sub(r'_v\d+\.json', '', filename)
-                    feature['properties']['file_id'] = file_id
-                except Exception as err:
-                    msg = f'Error while flattening Thunderstorm JSON {err}'
-                    LOGGER.error(f'{msg}')
-                    pass
-
-            del feature['properties']['metobject']
-
-            feat_exp_datetime = feature['properties']['expiration_datetime']
-            exp_datetime = datetime.strptime(feat_exp_datetime,
-                                             DATETIME_FORMAT)
-
-            if exp_datetime > datetime.now():
-                features.append(feature)
-
-        # check if id is already in ES and if amendment is +=1
-        amendment = features[0]['properties']['amendment']
-        file_id = features[0]['properties']['file_id']
-        is_newer = self.check_if_newer(file_id, amendment)
-
-        if is_newer['update']:
-            for outlook in features:
-                action = {
-                    '_id': outlook['properties']['id'],
-                    '_index': INDEX_NAME,
-                    '_op_type': 'update',
-                    'doc': outlook,
-                    'doc_as_upsert': True,
+        filename = f.name.split('/')[-1]
+        file_id = re.sub(r'_v\d+\.json', '', filename)
+
+        if len(data) > 0:
+            for feature in data:
+
+                # flatten metobject properties
+                metobj = feature['properties']['metobject']
+                for item, values in metobj.items():
+                    try:
+                        metobj_flat_item = self.flatten_json(item,
+                                                             values,
+                                                             'metobject')
+                        feature['properties'].update(metobj_flat_item)
+                        feature['properties']['file_id'] = file_id
+                    except Exception as err:
+                        msg = f'Error while flattening Thunderstorm JSON {err}'
+                        LOGGER.error(f'{msg}')
+                        pass
+
+                del feature['properties']['metobject']
+                f_exp_datetime = feature['properties']['expiration_datetime']
+                exp_datetime = datetime.strptime(f_exp_datetime,
+                                                 DATETIME_FORMAT)
+
+                if exp_datetime > datetime.now():
+                    features.append(feature)
+
+            # check if id is already in ES and if amendment is +=1
+            amendment = features[0]['properties']['amendment']
+            is_newer = self.check_if_newer(file_id, amendment)
+
+            if is_newer['update']:
+                for outlook in features:
+                    action = {
+                        '_id': outlook['properties']['id'],
+                        '_index': INDEX_NAME,
+                        '_op_type': 'update',
+                        'doc': outlook,
+                        'doc_as_upsert': True,
+                    }
+
+                    yield action
+
+                for id_ in is_newer['id_list']:
+                    self.conn.Elasticsearch.delete(index=INDEX_NAME,
+                                                   id=id_)
+        else:
+            LOGGER.warning(f'empty thunderstorm outlook json in {filename}')
+
+            version = re.search(r'v(\d+)\.json$', filename).group(1)
+            if int(version) > 1:
+                # we need to delete the associated outlooks
+                query = {
+                    "query": {
+                        "match": {
+                            "properties.file_id": file_id
+                        }
+                    }
                 }
-
-                yield action
-
-            for id_ in is_newer['id_list']:
-                self.conn.Elasticsearch.delete(index=INDEX_NAME,
-                                               id=id_)
+                self.conn.Elasticsearch.delete_by_query(index=INDEX_NAME,
+                                                        body=query)
 
     def flatten_json(self, key, values, parent_key=''):
         """